Repository: memononen/SDF
Branch: master
Commit: 1ef5b4e4feb6
Files: 8
Total size: 190.8 KB
Directory structure:
gitextract_bltkgvwp/
├── .gitignore
├── LICENSE.txt
├── README.md
├── example/
│ ├── example.c
│ ├── stb_image.c
│ └── stb_image_write.h
├── premake4.lua
└── src/
└── sdf.h
================================================
FILE CONTENTS
================================================
================================================
FILE: .gitignore
================================================
## Compiled source #
*.com
*.class
*.dll
*.exe
*.o
*.so
test
## Logs and databases #
*.log
*.sql
*.sqlite
## OS generated files #
.DS_Store
.DS_Store?
._*
.Spotlight-V100
.Trashes
ehthumbs.db
Thumbs.db
## Build dir
build/*
## xcode specific
*xcuserdata*
================================================
FILE: LICENSE.txt
================================================
The MIT License (MIT)
Copyright (C) 2014 Mikko Mononen (memon@inside.org)
Copyright (C) 2009-2012 Stefan Gustavson (stefan.gustavson@gmail.com)
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE SOFTWARE.
================================================
FILE: README.md
================================================
SDF
===
Sweep-and-update Euclidean distance transform of an antialised image for contour texturing.
The code is based on [edtaa3func.c](http://contourtextures.wikidot.com/) by Stefan Gustavson and improves the original in terms of memory usage and execution time.
The algorithms first traverse the image and uses gradient direction and the edge function from edtaa3 to find an approximated point on the contour of the input image. After this pass the distance at the edge pixels are known, and the code proceeds to update the rest of the distance field using sweep-and-update until the distance feild convergences (or max passes run).
The additional memory required by the code is 3 floats (px,py,distance) per pixel, compared to 5 doubles (or floats) of edtaa3. The sweep-and-update is
done using squared distances and contour points calculated only in the first pass. This is greatly reduces the amount of computation (especially suqare roots) per distance update.
The code procudes comparable, but probably not as accurate distance fields as the original code.
The code is intended to be used to calculate distance fields for [contour texturing](http://contourtextures.wikidot.com/).
## Usage
```C
int sdfBuildDistanceField(unsigned char* out, int outstride, float radius,
const unsigned char* img, int width, int height, int stride);
```
The output distance field is encoded as bytes, where 0 = radius (outside) and 255 = -radius (inside). Input and output can be the same buffer.
* _out_ - Output of the distance transform, one byte per pixel.
* _outstride_ - Bytes per row on output image.
* _radius_ - The radius of the distance field narrow band in pixels.
* _img_ - Input image, one byte per pixel.
* _width_ - Width if the image.
* _height_ - Height if the image.
* _stride_ - Bytes per row on input image.
White (255) pixels are treated as object pixels, zero pixels are treated as background. An attempt is made to treat antialiased edges correctly. The input image must have pixels in the range [0,255], and the antialiased image should be a box-filter sampling of the ideal, crisp edge. If the antialias region is more than 1 pixel wide, the result from this transform will be inaccurate. Pixels at image border are not calculated and are set to 0.
(Explanation borrowed from the original eedtaa3func.c)
```C
void sdfBuildDistanceFieldNoAlloc(unsigned char* out, int outstride, float radius,
const unsigned char* img, int width, int height, int stride,
unsigned char* temp);
```
Same as distXform, but does not allocate any memory. The `temp` array should be enough to fit `width * height * sizeof(float) * 3` bytes.
```C
void sdfCoverageToDistanceField(unsigned char* out, int outstride,
const unsigned char* img, int width, int height, int stride);
```
The output distance field is encoded as bytes, where 0 = sqrt(2) (outside) and 255 = -sqrt(2) (inside). Input and output must be different buffers.
* _out_ - Output of the distance transform, one byte per pixel.
* _outstride_ - Bytes per row on output image.
* _radius_ - The radius of the distance field narrow band in pixels.
* _img_ - Input image, one byte per pixel.
* _width_ - Width if the image.
* _height_ - Height if the image.
* _stride_ - Bytes per row on input image.
This function converts the antialiased image where each pixel represents coverage (box-filter sampling of the ideal, crisp edge) to a distance field with narrow band radius of sqrt(2). This is the fastest way (often up to 10x faster than `sdfBuildDistanceField`) to turn antialised image to contour texture. This function is good if you don't need the distance field for effects (i.e. fat outline or dropshadow).
The code is single header file only. Use following code once in your project to compile the implementation.
```C
#define SDF_IMPLEMENTATION
#include "sdf.h"
```
## License
MIT License
================================================
FILE: example/example.c
================================================
#include <stdio.h>
#include <string.h>
#include <float.h>
#include <GLFW/glfw3.h>
#define SDF_IMPLEMENTATION
#include "sdf.h"
#include "stb_image.c"
#define STB_IMAGE_WRITE_IMPLEMENTATION
#include "stb_image_write.h"
#include <sys/time.h>
#include <stdint.h>
int64_t getPerfTime()
{
struct timeval now;
gettimeofday(&now, 0);
return (int64_t)now.tv_sec*1000000L + (int64_t)now.tv_usec;
}
int deltaTimeUsec(int64_t start, int64_t end)
{
return (int)(end - start);
}
GLuint tex = 0;
GLuint tex2 = 0;
float imageAspect = 1.0f;
int alphaTest = 1;
float scale = 1.0f;
float radius = 2.0f;
float x = 0.0f;
float y = 0.0f;
const char* imageFilename = "../example/test.png";
//const char* imageFilename = "../example/flipper128.png";
struct Image {
unsigned char* data;
int width, height, bpp;
};
struct Image* imgLoad(const char* path, int bpp)
{
struct Image* img = malloc(sizeof(struct Image));
if (img == NULL) goto error;
img->data = stbi_load(path, &img->width, &img->height, &img->bpp, bpp);
if (img->data == NULL) goto error;
return img;
error:
if (img != NULL) {
free(img->data);
free(img);
}
return NULL;
}
struct Image* imgCreate(int w, int h, int bpp)
{
struct Image* img = malloc(sizeof(struct Image));
if (img == NULL) goto error;
img->width = w;
img->height = h;
img->bpp = bpp;
img->data = malloc(w * h * bpp);
if (img->data == NULL) goto error;
return img;
error:
if (img != NULL) {
free(img->data);
free(img);
}
return NULL;
}
void imgSave(struct Image* img, const char* path)
{
stbi_write_png(path, img->width, img->height, img->bpp, img->data, img->width*img->bpp);
}
void imgFree(struct Image* img)
{
if (img == NULL) return;
free(img->data);
free(img);
}
void imgInverse(struct Image* img)
{
int i;
for (i = 0; i < img->width*img->height; i++)
img->data[i] = 255-img->data[i];
}
GLuint imgTexture(struct Image* img)
{
GLuint tex = 0;
glGenTextures(1, &tex);
glBindTexture(GL_TEXTURE_2D, tex);
glPixelStorei(GL_UNPACK_ALIGNMENT,1);
glPixelStorei(GL_UNPACK_ROW_LENGTH, img->width);
glPixelStorei(GL_UNPACK_SKIP_PIXELS, 0);
glPixelStorei(GL_UNPACK_SKIP_ROWS, 0);
glTexImage2D(GL_TEXTURE_2D, 0, GL_ALPHA, img->width, img->height, 0, GL_ALPHA, GL_UNSIGNED_BYTE, img->data);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
glPixelStorei(GL_UNPACK_ALIGNMENT, 4);
glPixelStorei(GL_UNPACK_ROW_LENGTH, 0);
glPixelStorei(GL_UNPACK_SKIP_PIXELS, 0);
glPixelStorei(GL_UNPACK_SKIP_ROWS, 0);
return tex;
}
int loadImage(const char* imageFile, float radius, float* imageAspect, GLuint* tex, GLuint* texSDF)
{
int64_t t0, t1;
struct Image* img = NULL;
struct Image* img2 = NULL;
// Load example image
img = imgLoad(imageFile, 1);
if (img == NULL) {
printf("Could not load image.\n");
return 0;
}
imgInverse(img);
*imageAspect = img->height / (float) img->width;
// Build distance field and save it
img2 = imgCreate(img->width, img->height, 1);
if (img2 == NULL) {
return 0;
}
t0 = getPerfTime();
sdfBuildDistanceField(img2->data, img2->width, radius, img->data, img->width, img->height, img->width);
// sdfCoverageToDistanceField(img2->data, img2->width, img->data, img->width, img->height, img->width);
t1 = getPerfTime();
imgSave(img2, "dist.png");
printf("sdfBuild(%dx%d) %.1fms\n", img->width, img->height, deltaTimeUsec(t0, t1) / 1000.0f);
if (tex != 0) glDeleteTextures(1, tex);
*tex = imgTexture(img);
if (texSDF != 0) glDeleteTextures(1, texSDF);
*texSDF = imgTexture(img2);
imgFree(img);
imgFree(img2);
return 1;
}
static void key(GLFWwindow* window, int key, int scancode, int action, int mods)
{
if (key == GLFW_KEY_ESCAPE && action == GLFW_PRESS)
glfwSetWindowShouldClose(window, GL_TRUE);
if (key == GLFW_KEY_HOME && action == GLFW_PRESS) {
radius += 0.5f;
loadImage(imageFilename, radius, &imageAspect, &tex, &tex2);
}
if (key == GLFW_KEY_END && action == GLFW_PRESS) {
radius -= 0.5f;
if (radius < 0.0) radius = 0.0;
loadImage(imageFilename, radius, &imageAspect, &tex, &tex2);
}
if (key == GLFW_KEY_PAGE_UP && action == GLFW_PRESS) {
scale *= 1.1f;
}
if (key == GLFW_KEY_PAGE_DOWN && action == GLFW_PRESS) {
scale /= 1.1f;
}
if (key == GLFW_KEY_LEFT && action == GLFW_PRESS) {
x += 50.0f * scale;
}
if (key == GLFW_KEY_RIGHT && action == GLFW_PRESS) {
x -= 50.0f * scale;
}
if (key == GLFW_KEY_UP && action == GLFW_PRESS) {
y += 50.0f * scale;
}
if (key == GLFW_KEY_DOWN && action == GLFW_PRESS) {
y -= 50.0f * scale;
}
if (key == GLFW_KEY_A && action == GLFW_PRESS) {
alphaTest = !alphaTest;
}
}
int main()
{
GLFWwindow* window;
const GLFWvidmode* mode;
double time;
if (!glfwInit())
return -1;
mode = glfwGetVideoMode(glfwGetPrimaryMonitor());
window = glfwCreateWindow(mode->width - 40, mode->height - 80, "Distance Transform", NULL, NULL);
if (!window) {
glfwTerminate();
return -1;
}
glfwSetKeyCallback(window, key);
glfwMakeContextCurrent(window);
if (!loadImage(imageFilename, radius, &imageAspect, &tex, &tex2))
return -1;
glEnable(GL_LINE_SMOOTH);
time = glfwGetTime();
while (!glfwWindowShouldClose(window))
{
float o, w, h;
int width, height;
double t = glfwGetTime();
double dt = t - time;
if (dt > 0.5f) dt = 0.5f;
time = t;
glfwGetFramebufferSize(window, &width, &height);
// Update and render
glViewport(0, 0, width, height);
glClearColor(0.3f, 0.3f, 0.32f, 1.0f);
glClear(GL_COLOR_BUFFER_BIT|GL_DEPTH_BUFFER_BIT);
glEnable(GL_BLEND);
glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA);
glDisable(GL_TEXTURE_2D);
// Draw UI
glMatrixMode(GL_PROJECTION);
glLoadIdentity();
glOrtho(0,width,height,0,-1,1);
glMatrixMode(GL_MODELVIEW);
glLoadIdentity();
glDisable(GL_DEPTH_TEST);
glDisable(GL_CULL_FACE);
w = (width-40) * 0.5f * scale;
h = w * imageAspect;
o = w;
// Draw orig texture using bilinear filtering.
glEnable(GL_TEXTURE_2D);
glBindTexture(GL_TEXTURE_2D, tex);
glColor4ub(255,255,255,255);
glBegin(GL_QUADS);
glTexCoord2f(0,0);
glVertex2f(x+o,y);
glTexCoord2f(1,0);
glVertex2f(x+w+o,y);
glTexCoord2f(1,1);
glVertex2f(x+w+o,y+h);
glTexCoord2f(0,1);
glVertex2f(x+o,y+h);
glEnd();
// Draw distance texture using alha testing.
glBindTexture(GL_TEXTURE_2D, tex2);
glColor4ub(255,255,255,255);
if (alphaTest) {
glDisable(GL_BLEND);
glEnable(GL_ALPHA_TEST);
glAlphaFunc(GL_GREATER, 0.5f);
}
glBegin(GL_QUADS);
glTexCoord2f(0,0);
glVertex2f(x,y);
glTexCoord2f(1,0);
glVertex2f(x+w,y);
glTexCoord2f(1,1);
glVertex2f(x+w,y+h);
glTexCoord2f(0,1);
glVertex2f(x,y+h);
glEnd();
glDisable(GL_TEXTURE_2D);
if (alphaTest) {
glEnable(GL_BLEND);
glDisable(GL_ALPHA_TEST);
}
glEnable(GL_DEPTH_TEST);
glfwSwapBuffers(window);
glfwPollEvents();
}
glfwTerminate();
return 0;
}
================================================
FILE: example/stb_image.c
================================================
/* stbi-1.33 - public domain JPEG/PNG reader - http://nothings.org/stb_image.c
when you control the images you're loading
no warranty implied; use at your own risk
QUICK NOTES:
Primarily of interest to game developers and other people who can
avoid problematic images and only need the trivial interface
JPEG baseline (no JPEG progressive)
PNG 8-bit-per-channel only
TGA (not sure what subset, if a subset)
BMP non-1bpp, non-RLE
PSD (composited view only, no extra channels)
GIF (*comp always reports as 4-channel)
HDR (radiance rgbE format)
PIC (Softimage PIC)
- decode from memory or through FILE (define STBI_NO_STDIO to remove code)
- decode from arbitrary I/O callbacks
- overridable dequantizing-IDCT, YCbCr-to-RGB conversion (define STBI_SIMD)
Latest revisions:
1.33 (2011-07-14) minor fixes suggested by Dave Moore
1.32 (2011-07-13) info support for all filetypes (SpartanJ)
1.31 (2011-06-19) a few more leak fixes, bug in PNG handling (SpartanJ)
1.30 (2011-06-11) added ability to load files via io callbacks (Ben Wenger)
1.29 (2010-08-16) various warning fixes from Aurelien Pocheville
1.28 (2010-08-01) fix bug in GIF palette transparency (SpartanJ)
1.27 (2010-08-01) cast-to-uint8 to fix warnings (Laurent Gomila)
allow trailing 0s at end of image data (Laurent Gomila)
1.26 (2010-07-24) fix bug in file buffering for PNG reported by SpartanJ
See end of file for full revision history.
TODO:
stbi_info support for BMP,PSD,HDR,PIC
============================ Contributors =========================
Image formats Optimizations & bugfixes
Sean Barrett (jpeg, png, bmp) Fabian "ryg" Giesen
Nicolas Schulz (hdr, psd)
Jonathan Dummer (tga) Bug fixes & warning fixes
Jean-Marc Lienher (gif) Marc LeBlanc
Tom Seddon (pic) Christpher Lloyd
Thatcher Ulrich (psd) Dave Moore
Won Chun
the Horde3D community
Extensions, features Janez Zemva
Jetro Lauha (stbi_info) Jonathan Blow
James "moose2000" Brown (iPhone PNG) Laurent Gomila
Ben "Disch" Wenger (io callbacks) Aruelien Pocheville
Martin "SpartanJ" Golini Ryamond Barbiero
David Woo
If your name should be here but isn't, let Sean know.
*/
#ifndef STBI_INCLUDE_STB_IMAGE_H
#define STBI_INCLUDE_STB_IMAGE_H
// To get a header file for this, either cut and paste the header,
// or create stb_image.h, #define STBI_HEADER_FILE_ONLY, and
// then include stb_image.c from it.
//// begin header file ////////////////////////////////////////////////////
//
// Limitations:
// - no jpeg progressive support
// - non-HDR formats support 8-bit samples only (jpeg, png)
// - no delayed line count (jpeg) -- IJG doesn't support either
// - no 1-bit BMP
// - GIF always returns *comp=4
//
// Basic usage (see HDR discussion below):
// int x,y,n;
// unsigned char *data = stbi_load(filename, &x, &y, &n, 0);
// // ... process data if not NULL ...
// // ... x = width, y = height, n = # 8-bit components per pixel ...
// // ... replace '0' with '1'..'4' to force that many components per pixel
// // ... but 'n' will always be the number that it would have been if you said 0
// stbi_image_free(data)
//
// Standard parameters:
// int *x -- outputs image width in pixels
// int *y -- outputs image height in pixels
// int *comp -- outputs # of image components in image file
// int req_comp -- if non-zero, # of image components requested in result
//
// The return value from an image loader is an 'unsigned char *' which points
// to the pixel data. The pixel data consists of *y scanlines of *x pixels,
// with each pixel consisting of N interleaved 8-bit components; the first
// pixel pointed to is top-left-most in the image. There is no padding between
// image scanlines or between pixels, regardless of format. The number of
// components N is 'req_comp' if req_comp is non-zero, or *comp otherwise.
// If req_comp is non-zero, *comp has the number of components that _would_
// have been output otherwise. E.g. if you set req_comp to 4, you will always
// get RGBA output, but you can check *comp to easily see if it's opaque.
//
// An output image with N components has the following components interleaved
// in this order in each pixel:
//
// N=#comp components
// 1 grey
// 2 grey, alpha
// 3 red, green, blue
// 4 red, green, blue, alpha
//
// If image loading fails for any reason, the return value will be NULL,
// and *x, *y, *comp will be unchanged. The function stbi_failure_reason()
// can be queried for an extremely brief, end-user unfriendly explanation
// of why the load failed. Define STBI_NO_FAILURE_STRINGS to avoid
// compiling these strings at all, and STBI_FAILURE_USERMSG to get slightly
// more user-friendly ones.
//
// Paletted PNG, BMP, GIF, and PIC images are automatically depalettized.
//
// ===========================================================================
//
// iPhone PNG support:
//
// By default we convert iphone-formatted PNGs back to RGB; nominally they
// would silently load as BGR, except the existing code should have just
// failed on such iPhone PNGs. But you can disable this conversion by
// by calling stbi_convert_iphone_png_to_rgb(0), in which case
// you will always just get the native iphone "format" through.
//
// Call stbi_set_unpremultiply_on_load(1) as well to force a divide per
// pixel to remove any premultiplied alpha *only* if the image file explicitly
// says there's premultiplied data (currently only happens in iPhone images,
// and only if iPhone convert-to-rgb processing is on).
//
// ===========================================================================
//
// HDR image support (disable by defining STBI_NO_HDR)
//
// stb_image now supports loading HDR images in general, and currently
// the Radiance .HDR file format, although the support is provided
// generically. You can still load any file through the existing interface;
// if you attempt to load an HDR file, it will be automatically remapped to
// LDR, assuming gamma 2.2 and an arbitrary scale factor defaulting to 1;
// both of these constants can be reconfigured through this interface:
//
// stbi_hdr_to_ldr_gamma(2.2f);
// stbi_hdr_to_ldr_scale(1.0f);
//
// (note, do not use _inverse_ constants; stbi_image will invert them
// appropriately).
//
// Additionally, there is a new, parallel interface for loading files as
// (linear) floats to preserve the full dynamic range:
//
// float *data = stbi_loadf(filename, &x, &y, &n, 0);
//
// If you load LDR images through this interface, those images will
// be promoted to floating point values, run through the inverse of
// constants corresponding to the above:
//
// stbi_ldr_to_hdr_scale(1.0f);
// stbi_ldr_to_hdr_gamma(2.2f);
//
// Finally, given a filename (or an open file or memory block--see header
// file for details) containing image data, you can query for the "most
// appropriate" interface to use (that is, whether the image is HDR or
// not), using:
//
// stbi_is_hdr(char *filename);
//
// ===========================================================================
//
// I/O callbacks
//
// I/O callbacks allow you to read from arbitrary sources, like packaged
// files or some other source. Data read from callbacks are processed
// through a small internal buffer (currently 128 bytes) to try to reduce
// overhead.
//
// The three functions you must define are "read" (reads some bytes of data),
// "skip" (skips some bytes of data), "eof" (reports if the stream is at the end).
#ifndef STBI_NO_STDIO
#if defined(_MSC_VER) && _MSC_VER >= 0x1400
#define _CRT_SECURE_NO_WARNINGS // suppress bogus warnings about fopen()
#endif
#include <stdio.h>
#endif
#define STBI_VERSION 1
enum
{
STBI_default = 0, // only used for req_comp
STBI_grey = 1,
STBI_grey_alpha = 2,
STBI_rgb = 3,
STBI_rgb_alpha = 4
};
typedef unsigned char stbi_uc;
#ifdef __cplusplus
extern "C" {
#endif
//////////////////////////////////////////////////////////////////////////////
//
// PRIMARY API - works on images of any type
//
//
// load image by filename, open file, or memory buffer
//
extern stbi_uc *stbi_load_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp, int req_comp);
#ifndef STBI_NO_STDIO
extern stbi_uc *stbi_load (char const *filename, int *x, int *y, int *comp, int req_comp);
extern stbi_uc *stbi_load_from_file (FILE *f, int *x, int *y, int *comp, int req_comp);
// for stbi_load_from_file, file pointer is left pointing immediately after image
#endif
typedef struct
{
int (*read) (void *user,char *data,int size); // fill 'data' with 'size' bytes. return number of bytes actually read
void (*skip) (void *user,unsigned n); // skip the next 'n' bytes
int (*eof) (void *user); // returns nonzero if we are at end of file/data
} stbi_io_callbacks;
extern stbi_uc *stbi_load_from_callbacks (stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *comp, int req_comp);
#ifndef STBI_NO_HDR
extern float *stbi_loadf_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp, int req_comp);
#ifndef STBI_NO_STDIO
extern float *stbi_loadf (char const *filename, int *x, int *y, int *comp, int req_comp);
extern float *stbi_loadf_from_file (FILE *f, int *x, int *y, int *comp, int req_comp);
#endif
extern float *stbi_loadf_from_callbacks (stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *comp, int req_comp);
extern void stbi_hdr_to_ldr_gamma(float gamma);
extern void stbi_hdr_to_ldr_scale(float scale);
extern void stbi_ldr_to_hdr_gamma(float gamma);
extern void stbi_ldr_to_hdr_scale(float scale);
#endif // STBI_NO_HDR
// stbi_is_hdr is always defined
extern int stbi_is_hdr_from_callbacks(stbi_io_callbacks const *clbk, void *user);
extern int stbi_is_hdr_from_memory(stbi_uc const *buffer, int len);
#ifndef STBI_NO_STDIO
extern int stbi_is_hdr (char const *filename);
extern int stbi_is_hdr_from_file(FILE *f);
#endif // STBI_NO_STDIO
// get a VERY brief reason for failure
// NOT THREADSAFE
extern const char *stbi_failure_reason (void);
// free the loaded image -- this is just free()
extern void stbi_image_free (void *retval_from_stbi_load);
// get image dimensions & components without fully decoding
extern int stbi_info_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp);
extern int stbi_info_from_callbacks(stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *comp);
#ifndef STBI_NO_STDIO
extern int stbi_info (char const *filename, int *x, int *y, int *comp);
extern int stbi_info_from_file (FILE *f, int *x, int *y, int *comp);
#endif
// for image formats that explicitly notate that they have premultiplied alpha,
// we just return the colors as stored in the file. set this flag to force
// unpremultiplication. results are undefined if the unpremultiply overflow.
extern void stbi_set_unpremultiply_on_load(int flag_true_if_should_unpremultiply);
// indicate whether we should process iphone images back to canonical format,
// or just pass them through "as-is"
extern void stbi_convert_iphone_png_to_rgb(int flag_true_if_should_convert);
// ZLIB client - used by PNG, available for other purposes
extern char *stbi_zlib_decode_malloc_guesssize(const char *buffer, int len, int initial_size, int *outlen);
extern char *stbi_zlib_decode_malloc(const char *buffer, int len, int *outlen);
extern int stbi_zlib_decode_buffer(char *obuffer, int olen, const char *ibuffer, int ilen);
extern char *stbi_zlib_decode_noheader_malloc(const char *buffer, int len, int *outlen);
extern int stbi_zlib_decode_noheader_buffer(char *obuffer, int olen, const char *ibuffer, int ilen);
// define faster low-level operations (typically SIMD support)
#ifdef STBI_SIMD
typedef void (*stbi_idct_8x8)(stbi_uc *out, int out_stride, short data[64], unsigned short *dequantize);
// compute an integer IDCT on "input"
// input[x] = data[x] * dequantize[x]
// write results to 'out': 64 samples, each run of 8 spaced by 'out_stride'
// CLAMP results to 0..255
typedef void (*stbi_YCbCr_to_RGB_run)(stbi_uc *output, stbi_uc const *y, stbi_uc const *cb, stbi_uc const *cr, int count, int step);
// compute a conversion from YCbCr to RGB
// 'count' pixels
// write pixels to 'output'; each pixel is 'step' bytes (either 3 or 4; if 4, write '255' as 4th), order R,G,B
// y: Y input channel
// cb: Cb input channel; scale/biased to be 0..255
// cr: Cr input channel; scale/biased to be 0..255
extern void stbi_install_idct(stbi_idct_8x8 func);
extern void stbi_install_YCbCr_to_RGB(stbi_YCbCr_to_RGB_run func);
#endif // STBI_SIMD
#ifdef __cplusplus
}
#endif
//
//
//// end header file /////////////////////////////////////////////////////
#endif // STBI_INCLUDE_STB_IMAGE_H
#ifndef STBI_HEADER_FILE_ONLY
#ifndef STBI_NO_HDR
#include <math.h> // ldexp
#include <string.h> // strcmp, strtok
#endif
#ifndef STBI_NO_STDIO
#include <stdio.h>
#endif
#include <stdlib.h>
#include <memory.h>
#include <assert.h>
#include <stdarg.h>
#ifndef _MSC_VER
#ifdef __cplusplus
#define stbi_inline inline
#else
#define stbi_inline
#endif
#else
#define stbi_inline __forceinline
#endif
// implementation:
typedef unsigned char uint8;
typedef unsigned short uint16;
typedef signed short int16;
typedef unsigned int uint32;
typedef signed int int32;
typedef unsigned int uint;
// should produce compiler error if size is wrong
typedef unsigned char validate_uint32[sizeof(uint32)==4 ? 1 : -1];
#if defined(STBI_NO_STDIO) && !defined(STBI_NO_WRITE)
#define STBI_NO_WRITE
#endif
#define STBI_NOTUSED(v) (void)sizeof(v)
#ifdef _MSC_VER
#define STBI_HAS_LROTL
#endif
#ifdef STBI_HAS_LROTL
#define stbi_lrot(x,y) _lrotl(x,y)
#else
#define stbi_lrot(x,y) (((x) << (y)) | ((x) >> (32 - (y))))
#endif
///////////////////////////////////////////////
//
// stbi struct and start_xxx functions
// stbi structure is our basic context used by all images, so it
// contains all the IO context, plus some basic image information
typedef struct
{
uint32 img_x, img_y;
int img_n, img_out_n;
stbi_io_callbacks io;
void *io_user_data;
int read_from_callbacks;
int buflen;
uint8 buffer_start[128];
uint8 *img_buffer, *img_buffer_end;
uint8 *img_buffer_original;
} stbi;
static void refill_buffer(stbi *s);
// initialize a memory-decode context
static void start_mem(stbi *s, uint8 const *buffer, int len)
{
s->io.read = NULL;
s->read_from_callbacks = 0;
s->img_buffer = s->img_buffer_original = (uint8 *) buffer;
s->img_buffer_end = (uint8 *) buffer+len;
}
// initialize a callback-based context
static void start_callbacks(stbi *s, stbi_io_callbacks *c, void *user)
{
s->io = *c;
s->io_user_data = user;
s->buflen = sizeof(s->buffer_start);
s->read_from_callbacks = 1;
s->img_buffer_original = s->buffer_start;
refill_buffer(s);
}
#ifndef STBI_NO_STDIO
static int stdio_read(void *user, char *data, int size)
{
return (int) fread(data,1,size,(FILE*) user);
}
static void stdio_skip(void *user, unsigned n)
{
fseek((FILE*) user, n, SEEK_CUR);
}
static int stdio_eof(void *user)
{
return feof((FILE*) user);
}
static stbi_io_callbacks stbi_stdio_callbacks =
{
stdio_read,
stdio_skip,
stdio_eof,
};
static void start_file(stbi *s, FILE *f)
{
start_callbacks(s, &stbi_stdio_callbacks, (void *) f);
}
//static void stop_file(stbi *s) { }
#endif // !STBI_NO_STDIO
static void stbi_rewind(stbi *s)
{
// conceptually rewind SHOULD rewind to the beginning of the stream,
// but we just rewind to the beginning of the initial buffer, because
// we only use it after doing 'test', which only ever looks at at most 92 bytes
s->img_buffer = s->img_buffer_original;
}
static int stbi_jpeg_test(stbi *s);
static stbi_uc *stbi_jpeg_load(stbi *s, int *x, int *y, int *comp, int req_comp);
static int stbi_jpeg_info(stbi *s, int *x, int *y, int *comp);
static int stbi_png_test(stbi *s);
static stbi_uc *stbi_png_load(stbi *s, int *x, int *y, int *comp, int req_comp);
static int stbi_png_info(stbi *s, int *x, int *y, int *comp);
static int stbi_bmp_test(stbi *s);
static stbi_uc *stbi_bmp_load(stbi *s, int *x, int *y, int *comp, int req_comp);
static int stbi_tga_test(stbi *s);
static stbi_uc *stbi_tga_load(stbi *s, int *x, int *y, int *comp, int req_comp);
static int stbi_tga_info(stbi *s, int *x, int *y, int *comp);
static int stbi_psd_test(stbi *s);
static stbi_uc *stbi_psd_load(stbi *s, int *x, int *y, int *comp, int req_comp);
static int stbi_hdr_test(stbi *s);
static float *stbi_hdr_load(stbi *s, int *x, int *y, int *comp, int req_comp);
static int stbi_pic_test(stbi *s);
static stbi_uc *stbi_pic_load(stbi *s, int *x, int *y, int *comp, int req_comp);
static int stbi_gif_test(stbi *s);
static stbi_uc *stbi_gif_load(stbi *s, int *x, int *y, int *comp, int req_comp);
static int stbi_gif_info(stbi *s, int *x, int *y, int *comp);
// this is not threadsafe
static const char *failure_reason;
const char *stbi_failure_reason(void)
{
return failure_reason;
}
static int e(const char *str)
{
failure_reason = str;
return 0;
}
// e - error
// epf - error returning pointer to float
// epuc - error returning pointer to unsigned char
#ifdef STBI_NO_FAILURE_STRINGS
#define e(x,y) 0
#elif defined(STBI_FAILURE_USERMSG)
#define e(x,y) e(y)
#else
#define e(x,y) e(x)
#endif
#define epf(x,y) ((float *) (e(x,y)?NULL:NULL))
#define epuc(x,y) ((unsigned char *) (e(x,y)?NULL:NULL))
void stbi_image_free(void *retval_from_stbi_load)
{
free(retval_from_stbi_load);
}
#ifndef STBI_NO_HDR
static float *ldr_to_hdr(stbi_uc *data, int x, int y, int comp);
static stbi_uc *hdr_to_ldr(float *data, int x, int y, int comp);
#endif
static unsigned char *stbi_load_main(stbi *s, int *x, int *y, int *comp, int req_comp)
{
if (stbi_jpeg_test(s)) return stbi_jpeg_load(s,x,y,comp,req_comp);
if (stbi_png_test(s)) return stbi_png_load(s,x,y,comp,req_comp);
if (stbi_bmp_test(s)) return stbi_bmp_load(s,x,y,comp,req_comp);
if (stbi_gif_test(s)) return stbi_gif_load(s,x,y,comp,req_comp);
if (stbi_psd_test(s)) return stbi_psd_load(s,x,y,comp,req_comp);
if (stbi_pic_test(s)) return stbi_pic_load(s,x,y,comp,req_comp);
#ifndef STBI_NO_HDR
if (stbi_hdr_test(s)) {
float *hdr = stbi_hdr_load(s, x,y,comp,req_comp);
return hdr_to_ldr(hdr, *x, *y, req_comp ? req_comp : *comp);
}
#endif
// test tga last because it's a crappy test!
if (stbi_tga_test(s))
return stbi_tga_load(s,x,y,comp,req_comp);
return epuc("unknown image type", "Image not of any known type, or corrupt");
}
#ifndef STBI_NO_STDIO
unsigned char *stbi_load(char const *filename, int *x, int *y, int *comp, int req_comp)
{
FILE *f = fopen(filename, "rb");
unsigned char *result;
if (!f) return epuc("can't fopen", "Unable to open file");
result = stbi_load_from_file(f,x,y,comp,req_comp);
fclose(f);
return result;
}
unsigned char *stbi_load_from_file(FILE *f, int *x, int *y, int *comp, int req_comp)
{
stbi s;
start_file(&s,f);
return stbi_load_main(&s,x,y,comp,req_comp);
}
#endif //!STBI_NO_STDIO
unsigned char *stbi_load_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp, int req_comp)
{
stbi s;
start_mem(&s,buffer,len);
return stbi_load_main(&s,x,y,comp,req_comp);
}
unsigned char *stbi_load_from_callbacks(stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *comp, int req_comp)
{
stbi s;
start_callbacks(&s, (stbi_io_callbacks *) clbk, user);
return stbi_load_main(&s,x,y,comp,req_comp);
}
#ifndef STBI_NO_HDR
float *stbi_loadf_main(stbi *s, int *x, int *y, int *comp, int req_comp)
{
unsigned char *data;
#ifndef STBI_NO_HDR
if (stbi_hdr_test(s))
return stbi_hdr_load(s,x,y,comp,req_comp);
#endif
data = stbi_load_main(s, x, y, comp, req_comp);
if (data)
return ldr_to_hdr(data, *x, *y, req_comp ? req_comp : *comp);
return epf("unknown image type", "Image not of any known type, or corrupt");
}
float *stbi_loadf_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp, int req_comp)
{
stbi s;
start_mem(&s,buffer,len);
return stbi_loadf_main(&s,x,y,comp,req_comp);
}
float *stbi_loadf_from_callbacks(stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *comp, int req_comp)
{
stbi s;
start_callbacks(&s, (stbi_io_callbacks *) clbk, user);
return stbi_loadf_main(&s,x,y,comp,req_comp);
}
#ifndef STBI_NO_STDIO
float *stbi_loadf(char const *filename, int *x, int *y, int *comp, int req_comp)
{
FILE *f = fopen(filename, "rb");
float *result;
if (!f) return epf("can't fopen", "Unable to open file");
result = stbi_loadf_from_file(f,x,y,comp,req_comp);
fclose(f);
return result;
}
float *stbi_loadf_from_file(FILE *f, int *x, int *y, int *comp, int req_comp)
{
stbi s;
start_file(&s,f);
return stbi_loadf_main(&s,x,y,comp,req_comp);
}
#endif // !STBI_NO_STDIO
#endif // !STBI_NO_HDR
// these is-hdr-or-not is defined independent of whether STBI_NO_HDR is
// defined, for API simplicity; if STBI_NO_HDR is defined, it always
// reports false!
int stbi_is_hdr_from_memory(stbi_uc const *buffer, int len)
{
#ifndef STBI_NO_HDR
stbi s;
start_mem(&s,buffer,len);
return stbi_hdr_test(&s);
#else
STBI_NOTUSED(buffer);
STBI_NOTUSED(len);
return 0;
#endif
}
#ifndef STBI_NO_STDIO
extern int stbi_is_hdr (char const *filename)
{
FILE *f = fopen(filename, "rb");
int result=0;
if (f) {
result = stbi_is_hdr_from_file(f);
fclose(f);
}
return result;
}
extern int stbi_is_hdr_from_file(FILE *f)
{
#ifndef STBI_NO_HDR
stbi s;
start_file(&s,f);
return stbi_hdr_test(&s);
#else
return 0;
#endif
}
#endif // !STBI_NO_STDIO
extern int stbi_is_hdr_from_callbacks(stbi_io_callbacks const *clbk, void *user)
{
#ifndef STBI_NO_HDR
stbi s;
start_callbacks(&s, (stbi_io_callbacks *) clbk, user);
return stbi_hdr_test(&s);
#else
return 0;
#endif
}
#ifndef STBI_NO_HDR
static float h2l_gamma_i=1.0f/2.2f, h2l_scale_i=1.0f;
static float l2h_gamma=2.2f, l2h_scale=1.0f;
void stbi_hdr_to_ldr_gamma(float gamma) { h2l_gamma_i = 1/gamma; }
void stbi_hdr_to_ldr_scale(float scale) { h2l_scale_i = 1/scale; }
void stbi_ldr_to_hdr_gamma(float gamma) { l2h_gamma = gamma; }
void stbi_ldr_to_hdr_scale(float scale) { l2h_scale = scale; }
#endif
//////////////////////////////////////////////////////////////////////////////
//
// Common code used by all image loaders
//
enum
{
SCAN_load=0,
SCAN_type,
SCAN_header
};
static void refill_buffer(stbi *s)
{
int n = (s->io.read)(s->io_user_data,(char*)s->buffer_start,s->buflen);
if (n == 0) {
// at end of file, treat same as if from memory
s->read_from_callbacks = 0;
s->img_buffer = s->img_buffer_end-1;
*s->img_buffer = 0;
} else {
s->img_buffer = s->buffer_start;
s->img_buffer_end = s->buffer_start + n;
}
}
stbi_inline static int get8(stbi *s)
{
if (s->img_buffer < s->img_buffer_end)
return *s->img_buffer++;
if (s->read_from_callbacks) {
refill_buffer(s);
return *s->img_buffer++;
}
return 0;
}
stbi_inline static int at_eof(stbi *s)
{
if (s->io.read) {
if (!(s->io.eof)(s->io_user_data)) return 0;
// if feof() is true, check if buffer = end
// special case: we've only got the special 0 character at the end
if (s->read_from_callbacks == 0) return 1;
}
return s->img_buffer >= s->img_buffer_end;
}
stbi_inline static uint8 get8u(stbi *s)
{
return (uint8) get8(s);
}
static void skip(stbi *s, int n)
{
if (s->io.read) {
int blen = s->img_buffer_end - s->img_buffer;
if (blen < n) {
s->img_buffer = s->img_buffer_end;
(s->io.skip)(s->io_user_data, n - blen);
return;
}
}
s->img_buffer += n;
}
static int getn(stbi *s, stbi_uc *buffer, int n)
{
if (s->io.read) {
int blen = s->img_buffer_end - s->img_buffer;
if (blen < n) {
int res, count;
memcpy(buffer, s->img_buffer, blen);
count = (s->io.read)(s->io_user_data, (char*) buffer + blen, n - blen);
res = (count == (n-blen));
s->img_buffer = s->img_buffer_end;
return res;
}
}
if (s->img_buffer+n <= s->img_buffer_end) {
memcpy(buffer, s->img_buffer, n);
s->img_buffer += n;
return 1;
} else
return 0;
}
static int get16(stbi *s)
{
int z = get8(s);
return (z << 8) + get8(s);
}
static uint32 get32(stbi *s)
{
uint32 z = get16(s);
return (z << 16) + get16(s);
}
static int get16le(stbi *s)
{
int z = get8(s);
return z + (get8(s) << 8);
}
static uint32 get32le(stbi *s)
{
uint32 z = get16le(s);
return z + (get16le(s) << 16);
}
//////////////////////////////////////////////////////////////////////////////
//
// generic converter from built-in img_n to req_comp
// individual types do this automatically as much as possible (e.g. jpeg
// does all cases internally since it needs to colorspace convert anyway,
// and it never has alpha, so very few cases ). png can automatically
// interleave an alpha=255 channel, but falls back to this for other cases
//
// assume data buffer is malloced, so malloc a new one and free that one
// only failure mode is malloc failing
static uint8 compute_y(int r, int g, int b)
{
return (uint8) (((r*77) + (g*150) + (29*b)) >> 8);
}
static unsigned char *convert_format(unsigned char *data, int img_n, int req_comp, uint x, uint y)
{
int i,j;
unsigned char *good;
if (req_comp == img_n) return data;
assert(req_comp >= 1 && req_comp <= 4);
good = (unsigned char *) malloc(req_comp * x * y);
if (good == NULL) {
free(data);
return epuc("outofmem", "Out of memory");
}
for (j=0; j < (int) y; ++j) {
unsigned char *src = data + j * x * img_n ;
unsigned char *dest = good + j * x * req_comp;
#define COMBO(a,b) ((a)*8+(b))
#define CASE(a,b) case COMBO(a,b): for(i=x-1; i >= 0; --i, src += a, dest += b)
// convert source image with img_n components to one with req_comp components;
// avoid switch per pixel, so use switch per scanline and massive macros
switch (COMBO(img_n, req_comp)) {
CASE(1,2) dest[0]=src[0], dest[1]=255; break;
CASE(1,3) dest[0]=dest[1]=dest[2]=src[0]; break;
CASE(1,4) dest[0]=dest[1]=dest[2]=src[0], dest[3]=255; break;
CASE(2,1) dest[0]=src[0]; break;
CASE(2,3) dest[0]=dest[1]=dest[2]=src[0]; break;
CASE(2,4) dest[0]=dest[1]=dest[2]=src[0], dest[3]=src[1]; break;
CASE(3,4) dest[0]=src[0],dest[1]=src[1],dest[2]=src[2],dest[3]=255; break;
CASE(3,1) dest[0]=compute_y(src[0],src[1],src[2]); break;
CASE(3,2) dest[0]=compute_y(src[0],src[1],src[2]), dest[1] = 255; break;
CASE(4,1) dest[0]=compute_y(src[0],src[1],src[2]); break;
CASE(4,2) dest[0]=compute_y(src[0],src[1],src[2]), dest[1] = src[3]; break;
CASE(4,3) dest[0]=src[0],dest[1]=src[1],dest[2]=src[2]; break;
default: assert(0);
}
#undef CASE
}
free(data);
return good;
}
#ifndef STBI_NO_HDR
static float *ldr_to_hdr(stbi_uc *data, int x, int y, int comp)
{
int i,k,n;
float *output = (float *) malloc(x * y * comp * sizeof(float));
if (output == NULL) { free(data); return epf("outofmem", "Out of memory"); }
// compute number of non-alpha components
if (comp & 1) n = comp; else n = comp-1;
for (i=0; i < x*y; ++i) {
for (k=0; k < n; ++k) {
output[i*comp + k] = (float) pow(data[i*comp+k]/255.0f, l2h_gamma) * l2h_scale;
}
if (k < comp) output[i*comp + k] = data[i*comp+k]/255.0f;
}
free(data);
return output;
}
#define float2int(x) ((int) (x))
static stbi_uc *hdr_to_ldr(float *data, int x, int y, int comp)
{
int i,k,n;
stbi_uc *output = (stbi_uc *) malloc(x * y * comp);
if (output == NULL) { free(data); return epuc("outofmem", "Out of memory"); }
// compute number of non-alpha components
if (comp & 1) n = comp; else n = comp-1;
for (i=0; i < x*y; ++i) {
for (k=0; k < n; ++k) {
float z = (float) pow(data[i*comp+k]*h2l_scale_i, h2l_gamma_i) * 255 + 0.5f;
if (z < 0) z = 0;
if (z > 255) z = 255;
output[i*comp + k] = (uint8) float2int(z);
}
if (k < comp) {
float z = data[i*comp+k] * 255 + 0.5f;
if (z < 0) z = 0;
if (z > 255) z = 255;
output[i*comp + k] = (uint8) float2int(z);
}
}
free(data);
return output;
}
#endif
//////////////////////////////////////////////////////////////////////////////
//
// "baseline" JPEG/JFIF decoder (not actually fully baseline implementation)
//
// simple implementation
// - channel subsampling of at most 2 in each dimension
// - doesn't support delayed output of y-dimension
// - simple interface (only one output format: 8-bit interleaved RGB)
// - doesn't try to recover corrupt jpegs
// - doesn't allow partial loading, loading multiple at once
// - still fast on x86 (copying globals into locals doesn't help x86)
// - allocates lots of intermediate memory (full size of all components)
// - non-interleaved case requires this anyway
// - allows good upsampling (see next)
// high-quality
// - upsampled channels are bilinearly interpolated, even across blocks
// - quality integer IDCT derived from IJG's 'slow'
// performance
// - fast huffman; reasonable integer IDCT
// - uses a lot of intermediate memory, could cache poorly
// - load http://nothings.org/remote/anemones.jpg 3 times on 2.8Ghz P4
// stb_jpeg: 1.34 seconds (MSVC6, default release build)
// stb_jpeg: 1.06 seconds (MSVC6, processor = Pentium Pro)
// IJL11.dll: 1.08 seconds (compiled by intel)
// IJG 1998: 0.98 seconds (MSVC6, makefile provided by IJG)
// IJG 1998: 0.95 seconds (MSVC6, makefile + proc=PPro)
// huffman decoding acceleration
#define FAST_BITS 9 // larger handles more cases; smaller stomps less cache
typedef struct
{
uint8 fast[1 << FAST_BITS];
// weirdly, repacking this into AoS is a 10% speed loss, instead of a win
uint16 code[256];
uint8 values[256];
uint8 size[257];
unsigned int maxcode[18];
int delta[17]; // old 'firstsymbol' - old 'firstcode'
} huffman;
typedef struct
{
#ifdef STBI_SIMD
unsigned short dequant2[4][64];
#endif
stbi *s;
huffman huff_dc[4];
huffman huff_ac[4];
uint8 dequant[4][64];
// sizes for components, interleaved MCUs
int img_h_max, img_v_max;
int img_mcu_x, img_mcu_y;
int img_mcu_w, img_mcu_h;
// definition of jpeg image component
struct
{
int id;
int h,v;
int tq;
int hd,ha;
int dc_pred;
int x,y,w2,h2;
uint8 *data;
void *raw_data;
uint8 *linebuf;
} img_comp[4];
uint32 code_buffer; // jpeg entropy-coded buffer
int code_bits; // number of valid bits
unsigned char marker; // marker seen while filling entropy buffer
int nomore; // flag if we saw a marker so must stop
int scan_n, order[4];
int restart_interval, todo;
} jpeg;
static int build_huffman(huffman *h, int *count)
{
int i,j,k=0,code;
// build size list for each symbol (from JPEG spec)
for (i=0; i < 16; ++i)
for (j=0; j < count[i]; ++j)
h->size[k++] = (uint8) (i+1);
h->size[k] = 0;
// compute actual symbols (from jpeg spec)
code = 0;
k = 0;
for(j=1; j <= 16; ++j) {
// compute delta to add to code to compute symbol id
h->delta[j] = k - code;
if (h->size[k] == j) {
while (h->size[k] == j)
h->code[k++] = (uint16) (code++);
if (code-1 >= (1 << j)) return e("bad code lengths","Corrupt JPEG");
}
// compute largest code + 1 for this size, preshifted as needed later
h->maxcode[j] = code << (16-j);
code <<= 1;
}
h->maxcode[j] = 0xffffffff;
// build non-spec acceleration table; 255 is flag for not-accelerated
memset(h->fast, 255, 1 << FAST_BITS);
for (i=0; i < k; ++i) {
int s = h->size[i];
if (s <= FAST_BITS) {
int c = h->code[i] << (FAST_BITS-s);
int m = 1 << (FAST_BITS-s);
for (j=0; j < m; ++j) {
h->fast[c+j] = (uint8) i;
}
}
}
return 1;
}
static void grow_buffer_unsafe(jpeg *j)
{
do {
int b = j->nomore ? 0 : get8(j->s);
if (b == 0xff) {
int c = get8(j->s);
if (c != 0) {
j->marker = (unsigned char) c;
j->nomore = 1;
return;
}
}
j->code_buffer |= b << (24 - j->code_bits);
j->code_bits += 8;
} while (j->code_bits <= 24);
}
// (1 << n) - 1
static uint32 bmask[17]={0,1,3,7,15,31,63,127,255,511,1023,2047,4095,8191,16383,32767,65535};
// decode a jpeg huffman value from the bitstream
stbi_inline static int decode(jpeg *j, huffman *h)
{
unsigned int temp;
int c,k;
if (j->code_bits < 16) grow_buffer_unsafe(j);
// look at the top FAST_BITS and determine what symbol ID it is,
// if the code is <= FAST_BITS
c = (j->code_buffer >> (32 - FAST_BITS)) & ((1 << FAST_BITS)-1);
k = h->fast[c];
if (k < 255) {
int s = h->size[k];
if (s > j->code_bits)
return -1;
j->code_buffer <<= s;
j->code_bits -= s;
return h->values[k];
}
// naive test is to shift the code_buffer down so k bits are
// valid, then test against maxcode. To speed this up, we've
// preshifted maxcode left so that it has (16-k) 0s at the
// end; in other words, regardless of the number of bits, it
// wants to be compared against something shifted to have 16;
// that way we don't need to shift inside the loop.
temp = j->code_buffer >> 16;
for (k=FAST_BITS+1 ; ; ++k)
if (temp < h->maxcode[k])
break;
if (k == 17) {
// error! code not found
j->code_bits -= 16;
return -1;
}
if (k > j->code_bits)
return -1;
// convert the huffman code to the symbol id
c = ((j->code_buffer >> (32 - k)) & bmask[k]) + h->delta[k];
assert((((j->code_buffer) >> (32 - h->size[c])) & bmask[h->size[c]]) == h->code[c]);
// convert the id to a symbol
j->code_bits -= k;
j->code_buffer <<= k;
return h->values[c];
}
// combined JPEG 'receive' and JPEG 'extend', since baseline
// always extends everything it receives.
stbi_inline static int extend_receive(jpeg *j, int n)
{
unsigned int m = 1 << (n-1);
unsigned int k;
if (j->code_bits < n) grow_buffer_unsafe(j);
#if 1
k = stbi_lrot(j->code_buffer, n);
j->code_buffer = k & ~bmask[n];
k &= bmask[n];
j->code_bits -= n;
#else
k = (j->code_buffer >> (32 - n)) & bmask[n];
j->code_bits -= n;
j->code_buffer <<= n;
#endif
// the following test is probably a random branch that won't
// predict well. I tried to table accelerate it but failed.
// maybe it's compiling as a conditional move?
if (k < m)
return (-1 << n) + k + 1;
else
return k;
}
// given a value that's at position X in the zigzag stream,
// where does it appear in the 8x8 matrix coded as row-major?
static uint8 dezigzag[64+15] =
{
0, 1, 8, 16, 9, 2, 3, 10,
17, 24, 32, 25, 18, 11, 4, 5,
12, 19, 26, 33, 40, 48, 41, 34,
27, 20, 13, 6, 7, 14, 21, 28,
35, 42, 49, 56, 57, 50, 43, 36,
29, 22, 15, 23, 30, 37, 44, 51,
58, 59, 52, 45, 38, 31, 39, 46,
53, 60, 61, 54, 47, 55, 62, 63,
// let corrupt input sample past end
63, 63, 63, 63, 63, 63, 63, 63,
63, 63, 63, 63, 63, 63, 63
};
// decode one 64-entry block--
static int decode_block(jpeg *j, short data[64], huffman *hdc, huffman *hac, int b)
{
int diff,dc,k;
int t = decode(j, hdc);
if (t < 0) return e("bad huffman code","Corrupt JPEG");
// 0 all the ac values now so we can do it 32-bits at a time
memset(data,0,64*sizeof(data[0]));
diff = t ? extend_receive(j, t) : 0;
dc = j->img_comp[b].dc_pred + diff;
j->img_comp[b].dc_pred = dc;
data[0] = (short) dc;
// decode AC components, see JPEG spec
k = 1;
do {
int r,s;
int rs = decode(j, hac);
if (rs < 0) return e("bad huffman code","Corrupt JPEG");
s = rs & 15;
r = rs >> 4;
if (s == 0) {
if (rs != 0xf0) break; // end block
k += 16;
} else {
k += r;
// decode into unzigzag'd location
data[dezigzag[k++]] = (short) extend_receive(j,s);
}
} while (k < 64);
return 1;
}
// take a -128..127 value and clamp it and convert to 0..255
stbi_inline static uint8 clamp(int x)
{
// trick to use a single test to catch both cases
if ((unsigned int) x > 255) {
if (x < 0) return 0;
if (x > 255) return 255;
}
return (uint8) x;
}
#define f2f(x) (int) (((x) * 4096 + 0.5))
#define fsh(x) ((x) << 12)
// derived from jidctint -- DCT_ISLOW
#define IDCT_1D(s0,s1,s2,s3,s4,s5,s6,s7) \
int t0,t1,t2,t3,p1,p2,p3,p4,p5,x0,x1,x2,x3; \
p2 = s2; \
p3 = s6; \
p1 = (p2+p3) * f2f(0.5411961f); \
t2 = p1 + p3*f2f(-1.847759065f); \
t3 = p1 + p2*f2f( 0.765366865f); \
p2 = s0; \
p3 = s4; \
t0 = fsh(p2+p3); \
t1 = fsh(p2-p3); \
x0 = t0+t3; \
x3 = t0-t3; \
x1 = t1+t2; \
x2 = t1-t2; \
t0 = s7; \
t1 = s5; \
t2 = s3; \
t3 = s1; \
p3 = t0+t2; \
p4 = t1+t3; \
p1 = t0+t3; \
p2 = t1+t2; \
p5 = (p3+p4)*f2f( 1.175875602f); \
t0 = t0*f2f( 0.298631336f); \
t1 = t1*f2f( 2.053119869f); \
t2 = t2*f2f( 3.072711026f); \
t3 = t3*f2f( 1.501321110f); \
p1 = p5 + p1*f2f(-0.899976223f); \
p2 = p5 + p2*f2f(-2.562915447f); \
p3 = p3*f2f(-1.961570560f); \
p4 = p4*f2f(-0.390180644f); \
t3 += p1+p4; \
t2 += p2+p3; \
t1 += p2+p4; \
t0 += p1+p3;
#ifdef STBI_SIMD
typedef unsigned short stbi_dequantize_t;
#else
typedef uint8 stbi_dequantize_t;
#endif
// .344 seconds on 3*anemones.jpg
static void idct_block(uint8 *out, int out_stride, short data[64], stbi_dequantize_t *dequantize)
{
int i,val[64],*v=val;
stbi_dequantize_t *dq = dequantize;
uint8 *o;
short *d = data;
// columns
for (i=0; i < 8; ++i,++d,++dq, ++v) {
// if all zeroes, shortcut -- this avoids dequantizing 0s and IDCTing
if (d[ 8]==0 && d[16]==0 && d[24]==0 && d[32]==0
&& d[40]==0 && d[48]==0 && d[56]==0) {
// no shortcut 0 seconds
// (1|2|3|4|5|6|7)==0 0 seconds
// all separate -0.047 seconds
// 1 && 2|3 && 4|5 && 6|7: -0.047 seconds
int dcterm = d[0] * dq[0] << 2;
v[0] = v[8] = v[16] = v[24] = v[32] = v[40] = v[48] = v[56] = dcterm;
} else {
IDCT_1D(d[ 0]*dq[ 0],d[ 8]*dq[ 8],d[16]*dq[16],d[24]*dq[24],
d[32]*dq[32],d[40]*dq[40],d[48]*dq[48],d[56]*dq[56])
// constants scaled things up by 1<<12; let's bring them back
// down, but keep 2 extra bits of precision
x0 += 512; x1 += 512; x2 += 512; x3 += 512;
v[ 0] = (x0+t3) >> 10;
v[56] = (x0-t3) >> 10;
v[ 8] = (x1+t2) >> 10;
v[48] = (x1-t2) >> 10;
v[16] = (x2+t1) >> 10;
v[40] = (x2-t1) >> 10;
v[24] = (x3+t0) >> 10;
v[32] = (x3-t0) >> 10;
}
}
for (i=0, v=val, o=out; i < 8; ++i,v+=8,o+=out_stride) {
// no fast case since the first 1D IDCT spread components out
IDCT_1D(v[0],v[1],v[2],v[3],v[4],v[5],v[6],v[7])
// constants scaled things up by 1<<12, plus we had 1<<2 from first
// loop, plus horizontal and vertical each scale by sqrt(8) so together
// we've got an extra 1<<3, so 1<<17 total we need to remove.
// so we want to round that, which means adding 0.5 * 1<<17,
// aka 65536. Also, we'll end up with -128 to 127 that we want
// to encode as 0..255 by adding 128, so we'll add that before the shift
x0 += 65536 + (128<<17);
x1 += 65536 + (128<<17);
x2 += 65536 + (128<<17);
x3 += 65536 + (128<<17);
// tried computing the shifts into temps, or'ing the temps to see
// if any were out of range, but that was slower
o[0] = clamp((x0+t3) >> 17);
o[7] = clamp((x0-t3) >> 17);
o[1] = clamp((x1+t2) >> 17);
o[6] = clamp((x1-t2) >> 17);
o[2] = clamp((x2+t1) >> 17);
o[5] = clamp((x2-t1) >> 17);
o[3] = clamp((x3+t0) >> 17);
o[4] = clamp((x3-t0) >> 17);
}
}
#ifdef STBI_SIMD
static stbi_idct_8x8 stbi_idct_installed = idct_block;
void stbi_install_idct(stbi_idct_8x8 func)
{
stbi_idct_installed = func;
}
#endif
#define MARKER_none 0xff
// if there's a pending marker from the entropy stream, return that
// otherwise, fetch from the stream and get a marker. if there's no
// marker, return 0xff, which is never a valid marker value
static uint8 get_marker(jpeg *j)
{
uint8 x;
if (j->marker != MARKER_none) { x = j->marker; j->marker = MARKER_none; return x; }
x = get8u(j->s);
if (x != 0xff) return MARKER_none;
while (x == 0xff)
x = get8u(j->s);
return x;
}
// in each scan, we'll have scan_n components, and the order
// of the components is specified by order[]
#define RESTART(x) ((x) >= 0xd0 && (x) <= 0xd7)
// after a restart interval, reset the entropy decoder and
// the dc prediction
static void reset(jpeg *j)
{
j->code_bits = 0;
j->code_buffer = 0;
j->nomore = 0;
j->img_comp[0].dc_pred = j->img_comp[1].dc_pred = j->img_comp[2].dc_pred = 0;
j->marker = MARKER_none;
j->todo = j->restart_interval ? j->restart_interval : 0x7fffffff;
// no more than 1<<31 MCUs if no restart_interal? that's plenty safe,
// since we don't even allow 1<<30 pixels
}
static int parse_entropy_coded_data(jpeg *z)
{
reset(z);
if (z->scan_n == 1) {
int i,j;
#ifdef STBI_SIMD
__declspec(align(16))
#endif
short data[64];
int n = z->order[0];
// non-interleaved data, we just need to process one block at a time,
// in trivial scanline order
// number of blocks to do just depends on how many actual "pixels" this
// component has, independent of interleaved MCU blocking and such
int w = (z->img_comp[n].x+7) >> 3;
int h = (z->img_comp[n].y+7) >> 3;
for (j=0; j < h; ++j) {
for (i=0; i < w; ++i) {
if (!decode_block(z, data, z->huff_dc+z->img_comp[n].hd, z->huff_ac+z->img_comp[n].ha, n)) return 0;
#ifdef STBI_SIMD
stbi_idct_installed(z->img_comp[n].data+z->img_comp[n].w2*j*8+i*8, z->img_comp[n].w2, data, z->dequant2[z->img_comp[n].tq]);
#else
idct_block(z->img_comp[n].data+z->img_comp[n].w2*j*8+i*8, z->img_comp[n].w2, data, z->dequant[z->img_comp[n].tq]);
#endif
// every data block is an MCU, so countdown the restart interval
if (--z->todo <= 0) {
if (z->code_bits < 24) grow_buffer_unsafe(z);
// if it's NOT a restart, then just bail, so we get corrupt data
// rather than no data
if (!RESTART(z->marker)) return 1;
reset(z);
}
}
}
} else { // interleaved!
int i,j,k,x,y;
short data[64];
for (j=0; j < z->img_mcu_y; ++j) {
for (i=0; i < z->img_mcu_x; ++i) {
// scan an interleaved mcu... process scan_n components in order
for (k=0; k < z->scan_n; ++k) {
int n = z->order[k];
// scan out an mcu's worth of this component; that's just determined
// by the basic H and V specified for the component
for (y=0; y < z->img_comp[n].v; ++y) {
for (x=0; x < z->img_comp[n].h; ++x) {
int x2 = (i*z->img_comp[n].h + x)*8;
int y2 = (j*z->img_comp[n].v + y)*8;
if (!decode_block(z, data, z->huff_dc+z->img_comp[n].hd, z->huff_ac+z->img_comp[n].ha, n)) return 0;
#ifdef STBI_SIMD
stbi_idct_installed(z->img_comp[n].data+z->img_comp[n].w2*y2+x2, z->img_comp[n].w2, data, z->dequant2[z->img_comp[n].tq]);
#else
idct_block(z->img_comp[n].data+z->img_comp[n].w2*y2+x2, z->img_comp[n].w2, data, z->dequant[z->img_comp[n].tq]);
#endif
}
}
}
// after all interleaved components, that's an interleaved MCU,
// so now count down the restart interval
if (--z->todo <= 0) {
if (z->code_bits < 24) grow_buffer_unsafe(z);
// if it's NOT a restart, then just bail, so we get corrupt data
// rather than no data
if (!RESTART(z->marker)) return 1;
reset(z);
}
}
}
}
return 1;
}
static int process_marker(jpeg *z, int m)
{
int L;
switch (m) {
case MARKER_none: // no marker found
return e("expected marker","Corrupt JPEG");
case 0xC2: // SOF - progressive
return e("progressive jpeg","JPEG format not supported (progressive)");
case 0xDD: // DRI - specify restart interval
if (get16(z->s) != 4) return e("bad DRI len","Corrupt JPEG");
z->restart_interval = get16(z->s);
return 1;
case 0xDB: // DQT - define quantization table
L = get16(z->s)-2;
while (L > 0) {
int q = get8(z->s);
int p = q >> 4;
int t = q & 15,i;
if (p != 0) return e("bad DQT type","Corrupt JPEG");
if (t > 3) return e("bad DQT table","Corrupt JPEG");
for (i=0; i < 64; ++i)
z->dequant[t][dezigzag[i]] = get8u(z->s);
#ifdef STBI_SIMD
for (i=0; i < 64; ++i)
z->dequant2[t][i] = z->dequant[t][i];
#endif
L -= 65;
}
return L==0;
case 0xC4: // DHT - define huffman table
L = get16(z->s)-2;
while (L > 0) {
uint8 *v;
int sizes[16],i,m=0;
int q = get8(z->s);
int tc = q >> 4;
int th = q & 15;
if (tc > 1 || th > 3) return e("bad DHT header","Corrupt JPEG");
for (i=0; i < 16; ++i) {
sizes[i] = get8(z->s);
m += sizes[i];
}
L -= 17;
if (tc == 0) {
if (!build_huffman(z->huff_dc+th, sizes)) return 0;
v = z->huff_dc[th].values;
} else {
if (!build_huffman(z->huff_ac+th, sizes)) return 0;
v = z->huff_ac[th].values;
}
for (i=0; i < m; ++i)
v[i] = get8u(z->s);
L -= m;
}
return L==0;
}
// check for comment block or APP blocks
if ((m >= 0xE0 && m <= 0xEF) || m == 0xFE) {
skip(z->s, get16(z->s)-2);
return 1;
}
return 0;
}
// after we see SOS
static int process_scan_header(jpeg *z)
{
int i;
int Ls = get16(z->s);
z->scan_n = get8(z->s);
if (z->scan_n < 1 || z->scan_n > 4 || z->scan_n > (int) z->s->img_n) return e("bad SOS component count","Corrupt JPEG");
if (Ls != 6+2*z->scan_n) return e("bad SOS len","Corrupt JPEG");
for (i=0; i < z->scan_n; ++i) {
int id = get8(z->s), which;
int q = get8(z->s);
for (which = 0; which < z->s->img_n; ++which)
if (z->img_comp[which].id == id)
break;
if (which == z->s->img_n) return 0;
z->img_comp[which].hd = q >> 4; if (z->img_comp[which].hd > 3) return e("bad DC huff","Corrupt JPEG");
z->img_comp[which].ha = q & 15; if (z->img_comp[which].ha > 3) return e("bad AC huff","Corrupt JPEG");
z->order[i] = which;
}
if (get8(z->s) != 0) return e("bad SOS","Corrupt JPEG");
get8(z->s); // should be 63, but might be 0
if (get8(z->s) != 0) return e("bad SOS","Corrupt JPEG");
return 1;
}
static int process_frame_header(jpeg *z, int scan)
{
stbi *s = z->s;
int Lf,p,i,q, h_max=1,v_max=1,c;
Lf = get16(s); if (Lf < 11) return e("bad SOF len","Corrupt JPEG"); // JPEG
p = get8(s); if (p != 8) return e("only 8-bit","JPEG format not supported: 8-bit only"); // JPEG baseline
s->img_y = get16(s); if (s->img_y == 0) return e("no header height", "JPEG format not supported: delayed height"); // Legal, but we don't handle it--but neither does IJG
s->img_x = get16(s); if (s->img_x == 0) return e("0 width","Corrupt JPEG"); // JPEG requires
c = get8(s);
if (c != 3 && c != 1) return e("bad component count","Corrupt JPEG"); // JFIF requires
s->img_n = c;
for (i=0; i < c; ++i) {
z->img_comp[i].data = NULL;
z->img_comp[i].linebuf = NULL;
}
if (Lf != 8+3*s->img_n) return e("bad SOF len","Corrupt JPEG");
for (i=0; i < s->img_n; ++i) {
z->img_comp[i].id = get8(s);
if (z->img_comp[i].id != i+1) // JFIF requires
if (z->img_comp[i].id != i) // some version of jpegtran outputs non-JFIF-compliant files!
return e("bad component ID","Corrupt JPEG");
q = get8(s);
z->img_comp[i].h = (q >> 4); if (!z->img_comp[i].h || z->img_comp[i].h > 4) return e("bad H","Corrupt JPEG");
z->img_comp[i].v = q & 15; if (!z->img_comp[i].v || z->img_comp[i].v > 4) return e("bad V","Corrupt JPEG");
z->img_comp[i].tq = get8(s); if (z->img_comp[i].tq > 3) return e("bad TQ","Corrupt JPEG");
}
if (scan != SCAN_load) return 1;
if ((1 << 30) / s->img_x / s->img_n < s->img_y) return e("too large", "Image too large to decode");
for (i=0; i < s->img_n; ++i) {
if (z->img_comp[i].h > h_max) h_max = z->img_comp[i].h;
if (z->img_comp[i].v > v_max) v_max = z->img_comp[i].v;
}
// compute interleaved mcu info
z->img_h_max = h_max;
z->img_v_max = v_max;
z->img_mcu_w = h_max * 8;
z->img_mcu_h = v_max * 8;
z->img_mcu_x = (s->img_x + z->img_mcu_w-1) / z->img_mcu_w;
z->img_mcu_y = (s->img_y + z->img_mcu_h-1) / z->img_mcu_h;
for (i=0; i < s->img_n; ++i) {
// number of effective pixels (e.g. for non-interleaved MCU)
z->img_comp[i].x = (s->img_x * z->img_comp[i].h + h_max-1) / h_max;
z->img_comp[i].y = (s->img_y * z->img_comp[i].v + v_max-1) / v_max;
// to simplify generation, we'll allocate enough memory to decode
// the bogus oversized data from using interleaved MCUs and their
// big blocks (e.g. a 16x16 iMCU on an image of width 33); we won't
// discard the extra data until colorspace conversion
z->img_comp[i].w2 = z->img_mcu_x * z->img_comp[i].h * 8;
z->img_comp[i].h2 = z->img_mcu_y * z->img_comp[i].v * 8;
z->img_comp[i].raw_data = malloc(z->img_comp[i].w2 * z->img_comp[i].h2+15);
if (z->img_comp[i].raw_data == NULL) {
for(--i; i >= 0; --i) {
free(z->img_comp[i].raw_data);
z->img_comp[i].data = NULL;
}
return e("outofmem", "Out of memory");
}
// align blocks for installable-idct using mmx/sse
z->img_comp[i].data = (uint8*) (((size_t) z->img_comp[i].raw_data + 15) & ~15);
z->img_comp[i].linebuf = NULL;
}
return 1;
}
// use comparisons since in some cases we handle more than one case (e.g. SOF)
#define DNL(x) ((x) == 0xdc)
#define SOI(x) ((x) == 0xd8)
#define EOI(x) ((x) == 0xd9)
#define SOF(x) ((x) == 0xc0 || (x) == 0xc1)
#define SOS(x) ((x) == 0xda)
static int decode_jpeg_header(jpeg *z, int scan)
{
int m;
z->marker = MARKER_none; // initialize cached marker to empty
m = get_marker(z);
if (!SOI(m)) return e("no SOI","Corrupt JPEG");
if (scan == SCAN_type) return 1;
m = get_marker(z);
while (!SOF(m)) {
if (!process_marker(z,m)) return 0;
m = get_marker(z);
while (m == MARKER_none) {
// some files have extra padding after their blocks, so ok, we'll scan
if (at_eof(z->s)) return e("no SOF", "Corrupt JPEG");
m = get_marker(z);
}
}
if (!process_frame_header(z, scan)) return 0;
return 1;
}
static int decode_jpeg_image(jpeg *j)
{
int m;
j->restart_interval = 0;
if (!decode_jpeg_header(j, SCAN_load)) return 0;
m = get_marker(j);
while (!EOI(m)) {
if (SOS(m)) {
if (!process_scan_header(j)) return 0;
if (!parse_entropy_coded_data(j)) return 0;
if (j->marker == MARKER_none ) {
// handle 0s at the end of image data from IP Kamera 9060
while (!at_eof(j->s)) {
int x = get8(j->s);
if (x == 255) {
j->marker = get8u(j->s);
break;
} else if (x != 0) {
return 0;
}
}
// if we reach eof without hitting a marker, get_marker() below will fail and we'll eventually return 0
}
} else {
if (!process_marker(j, m)) return 0;
}
m = get_marker(j);
}
return 1;
}
// static jfif-centered resampling (across block boundaries)
typedef uint8 *(*resample_row_func)(uint8 *out, uint8 *in0, uint8 *in1,
int w, int hs);
#define div4(x) ((uint8) ((x) >> 2))
static uint8 *resample_row_1(uint8 *out, uint8 *in_near, uint8 *in_far, int w, int hs)
{
STBI_NOTUSED(out);
STBI_NOTUSED(in_far);
STBI_NOTUSED(w);
STBI_NOTUSED(hs);
return in_near;
}
static uint8* resample_row_v_2(uint8 *out, uint8 *in_near, uint8 *in_far, int w, int hs)
{
// need to generate two samples vertically for every one in input
int i;
STBI_NOTUSED(hs);
for (i=0; i < w; ++i)
out[i] = div4(3*in_near[i] + in_far[i] + 2);
return out;
}
static uint8* resample_row_h_2(uint8 *out, uint8 *in_near, uint8 *in_far, int w, int hs)
{
// need to generate two samples horizontally for every one in input
int i;
uint8 *input = in_near;
if (w == 1) {
// if only one sample, can't do any interpolation
out[0] = out[1] = input[0];
return out;
}
out[0] = input[0];
out[1] = div4(input[0]*3 + input[1] + 2);
for (i=1; i < w-1; ++i) {
int n = 3*input[i]+2;
out[i*2+0] = div4(n+input[i-1]);
out[i*2+1] = div4(n+input[i+1]);
}
out[i*2+0] = div4(input[w-2]*3 + input[w-1] + 2);
out[i*2+1] = input[w-1];
STBI_NOTUSED(in_far);
STBI_NOTUSED(hs);
return out;
}
#define div16(x) ((uint8) ((x) >> 4))
static uint8 *resample_row_hv_2(uint8 *out, uint8 *in_near, uint8 *in_far, int w, int hs)
{
// need to generate 2x2 samples for every one in input
int i,t0,t1;
if (w == 1) {
out[0] = out[1] = div4(3*in_near[0] + in_far[0] + 2);
return out;
}
t1 = 3*in_near[0] + in_far[0];
out[0] = div4(t1+2);
for (i=1; i < w; ++i) {
t0 = t1;
t1 = 3*in_near[i]+in_far[i];
out[i*2-1] = div16(3*t0 + t1 + 8);
out[i*2 ] = div16(3*t1 + t0 + 8);
}
out[w*2-1] = div4(t1+2);
STBI_NOTUSED(hs);
return out;
}
static uint8 *resample_row_generic(uint8 *out, uint8 *in_near, uint8 *in_far, int w, int hs)
{
// resample with nearest-neighbor
int i,j;
in_far = in_far;
for (i=0; i < w; ++i)
for (j=0; j < hs; ++j)
out[i*hs+j] = in_near[i];
return out;
}
#define float2fixed(x) ((int) ((x) * 65536 + 0.5))
// 0.38 seconds on 3*anemones.jpg (0.25 with processor = Pro)
// VC6 without processor=Pro is generating multiple LEAs per multiply!
static void YCbCr_to_RGB_row(uint8 *out, const uint8 *y, const uint8 *pcb, const uint8 *pcr, int count, int step)
{
int i;
for (i=0; i < count; ++i) {
int y_fixed = (y[i] << 16) + 32768; // rounding
int r,g,b;
int cr = pcr[i] - 128;
int cb = pcb[i] - 128;
r = y_fixed + cr*float2fixed(1.40200f);
g = y_fixed - cr*float2fixed(0.71414f) - cb*float2fixed(0.34414f);
b = y_fixed + cb*float2fixed(1.77200f);
r >>= 16;
g >>= 16;
b >>= 16;
if ((unsigned) r > 255) { if (r < 0) r = 0; else r = 255; }
if ((unsigned) g > 255) { if (g < 0) g = 0; else g = 255; }
if ((unsigned) b > 255) { if (b < 0) b = 0; else b = 255; }
out[0] = (uint8)r;
out[1] = (uint8)g;
out[2] = (uint8)b;
out[3] = 255;
out += step;
}
}
#ifdef STBI_SIMD
static stbi_YCbCr_to_RGB_run stbi_YCbCr_installed = YCbCr_to_RGB_row;
void stbi_install_YCbCr_to_RGB(stbi_YCbCr_to_RGB_run func)
{
stbi_YCbCr_installed = func;
}
#endif
// clean up the temporary component buffers
static void cleanup_jpeg(jpeg *j)
{
int i;
for (i=0; i < j->s->img_n; ++i) {
if (j->img_comp[i].data) {
free(j->img_comp[i].raw_data);
j->img_comp[i].data = NULL;
}
if (j->img_comp[i].linebuf) {
free(j->img_comp[i].linebuf);
j->img_comp[i].linebuf = NULL;
}
}
}
typedef struct
{
resample_row_func resample;
uint8 *line0,*line1;
int hs,vs; // expansion factor in each axis
int w_lores; // horizontal pixels pre-expansion
int ystep; // how far through vertical expansion we are
int ypos; // which pre-expansion row we're on
} stbi_resample;
static uint8 *load_jpeg_image(jpeg *z, int *out_x, int *out_y, int *comp, int req_comp)
{
int n, decode_n;
// validate req_comp
if (req_comp < 0 || req_comp > 4) return epuc("bad req_comp", "Internal error");
z->s->img_n = 0;
// load a jpeg image from whichever source
if (!decode_jpeg_image(z)) { cleanup_jpeg(z); return NULL; }
// determine actual number of components to generate
n = req_comp ? req_comp : z->s->img_n;
if (z->s->img_n == 3 && n < 3)
decode_n = 1;
else
decode_n = z->s->img_n;
// resample and color-convert
{
int k;
uint i,j;
uint8 *output;
uint8 *coutput[4];
stbi_resample res_comp[4];
for (k=0; k < decode_n; ++k) {
stbi_resample *r = &res_comp[k];
// allocate line buffer big enough for upsampling off the edges
// with upsample factor of 4
z->img_comp[k].linebuf = (uint8 *) malloc(z->s->img_x + 3);
if (!z->img_comp[k].linebuf) { cleanup_jpeg(z); return epuc("outofmem", "Out of memory"); }
r->hs = z->img_h_max / z->img_comp[k].h;
r->vs = z->img_v_max / z->img_comp[k].v;
r->ystep = r->vs >> 1;
r->w_lores = (z->s->img_x + r->hs-1) / r->hs;
r->ypos = 0;
r->line0 = r->line1 = z->img_comp[k].data;
if (r->hs == 1 && r->vs == 1) r->resample = resample_row_1;
else if (r->hs == 1 && r->vs == 2) r->resample = resample_row_v_2;
else if (r->hs == 2 && r->vs == 1) r->resample = resample_row_h_2;
else if (r->hs == 2 && r->vs == 2) r->resample = resample_row_hv_2;
else r->resample = resample_row_generic;
}
// can't error after this so, this is safe
output = (uint8 *) malloc(n * z->s->img_x * z->s->img_y + 1);
if (!output) { cleanup_jpeg(z); return epuc("outofmem", "Out of memory"); }
// now go ahead and resample
for (j=0; j < z->s->img_y; ++j) {
uint8 *out = output + n * z->s->img_x * j;
for (k=0; k < decode_n; ++k) {
stbi_resample *r = &res_comp[k];
int y_bot = r->ystep >= (r->vs >> 1);
coutput[k] = r->resample(z->img_comp[k].linebuf,
y_bot ? r->line1 : r->line0,
y_bot ? r->line0 : r->line1,
r->w_lores, r->hs);
if (++r->ystep >= r->vs) {
r->ystep = 0;
r->line0 = r->line1;
if (++r->ypos < z->img_comp[k].y)
r->line1 += z->img_comp[k].w2;
}
}
if (n >= 3) {
uint8 *y = coutput[0];
if (z->s->img_n == 3) {
#ifdef STBI_SIMD
stbi_YCbCr_installed(out, y, coutput[1], coutput[2], z->s.img_x, n);
#else
YCbCr_to_RGB_row(out, y, coutput[1], coutput[2], z->s->img_x, n);
#endif
} else
for (i=0; i < z->s->img_x; ++i) {
out[0] = out[1] = out[2] = y[i];
out[3] = 255; // not used if n==3
out += n;
}
} else {
uint8 *y = coutput[0];
if (n == 1)
for (i=0; i < z->s->img_x; ++i) out[i] = y[i];
else
for (i=0; i < z->s->img_x; ++i) *out++ = y[i], *out++ = 255;
}
}
cleanup_jpeg(z);
*out_x = z->s->img_x;
*out_y = z->s->img_y;
if (comp) *comp = z->s->img_n; // report original components, not output
return output;
}
}
static unsigned char *stbi_jpeg_load(stbi *s, int *x, int *y, int *comp, int req_comp)
{
jpeg j;
j.s = s;
return load_jpeg_image(&j, x,y,comp,req_comp);
}
static int stbi_jpeg_test(stbi *s)
{
int r;
jpeg j;
j.s = s;
r = decode_jpeg_header(&j, SCAN_type);
stbi_rewind(s);
return r;
}
static int stbi_jpeg_info_raw(jpeg *j, int *x, int *y, int *comp)
{
if (!decode_jpeg_header(j, SCAN_header)) {
stbi_rewind( j->s );
return 0;
}
if (x) *x = j->s->img_x;
if (y) *y = j->s->img_y;
if (comp) *comp = j->s->img_n;
return 1;
}
static int stbi_jpeg_info(stbi *s, int *x, int *y, int *comp)
{
jpeg j;
j.s = s;
return stbi_jpeg_info_raw(&j, x, y, comp);
}
// public domain zlib decode v0.2 Sean Barrett 2006-11-18
// simple implementation
// - all input must be provided in an upfront buffer
// - all output is written to a single output buffer (can malloc/realloc)
// performance
// - fast huffman
// fast-way is faster to check than jpeg huffman, but slow way is slower
#define ZFAST_BITS 9 // accelerate all cases in default tables
#define ZFAST_MASK ((1 << ZFAST_BITS) - 1)
// zlib-style huffman encoding
// (jpegs packs from left, zlib from right, so can't share code)
typedef struct
{
uint16 fast[1 << ZFAST_BITS];
uint16 firstcode[16];
int maxcode[17];
uint16 firstsymbol[16];
uint8 size[288];
uint16 value[288];
} zhuffman;
stbi_inline static int bitreverse16(int n)
{
n = ((n & 0xAAAA) >> 1) | ((n & 0x5555) << 1);
n = ((n & 0xCCCC) >> 2) | ((n & 0x3333) << 2);
n = ((n & 0xF0F0) >> 4) | ((n & 0x0F0F) << 4);
n = ((n & 0xFF00) >> 8) | ((n & 0x00FF) << 8);
return n;
}
stbi_inline static int bit_reverse(int v, int bits)
{
assert(bits <= 16);
// to bit reverse n bits, reverse 16 and shift
// e.g. 11 bits, bit reverse and shift away 5
return bitreverse16(v) >> (16-bits);
}
static int zbuild_huffman(zhuffman *z, uint8 *sizelist, int num)
{
int i,k=0;
int code, next_code[16], sizes[17];
// DEFLATE spec for generating codes
memset(sizes, 0, sizeof(sizes));
memset(z->fast, 255, sizeof(z->fast));
for (i=0; i < num; ++i)
++sizes[sizelist[i]];
sizes[0] = 0;
for (i=1; i < 16; ++i)
assert(sizes[i] <= (1 << i));
code = 0;
for (i=1; i < 16; ++i) {
next_code[i] = code;
z->firstcode[i] = (uint16) code;
z->firstsymbol[i] = (uint16) k;
code = (code + sizes[i]);
if (sizes[i])
if (code-1 >= (1 << i)) return e("bad codelengths","Corrupt JPEG");
z->maxcode[i] = code << (16-i); // preshift for inner loop
code <<= 1;
k += sizes[i];
}
z->maxcode[16] = 0x10000; // sentinel
for (i=0; i < num; ++i) {
int s = sizelist[i];
if (s) {
int c = next_code[s] - z->firstcode[s] + z->firstsymbol[s];
z->size[c] = (uint8)s;
z->value[c] = (uint16)i;
if (s <= ZFAST_BITS) {
int k = bit_reverse(next_code[s],s);
while (k < (1 << ZFAST_BITS)) {
z->fast[k] = (uint16) c;
k += (1 << s);
}
}
++next_code[s];
}
}
return 1;
}
// zlib-from-memory implementation for PNG reading
// because PNG allows splitting the zlib stream arbitrarily,
// and it's annoying structurally to have PNG call ZLIB call PNG,
// we require PNG read all the IDATs and combine them into a single
// memory buffer
typedef struct
{
uint8 *zbuffer, *zbuffer_end;
int num_bits;
uint32 code_buffer;
char *zout;
char *zout_start;
char *zout_end;
int z_expandable;
zhuffman z_length, z_distance;
} zbuf;
stbi_inline static int zget8(zbuf *z)
{
if (z->zbuffer >= z->zbuffer_end) return 0;
return *z->zbuffer++;
}
static void fill_bits(zbuf *z)
{
do {
assert(z->code_buffer < (1U << z->num_bits));
z->code_buffer |= zget8(z) << z->num_bits;
z->num_bits += 8;
} while (z->num_bits <= 24);
}
stbi_inline static unsigned int zreceive(zbuf *z, int n)
{
unsigned int k;
if (z->num_bits < n) fill_bits(z);
k = z->code_buffer & ((1 << n) - 1);
z->code_buffer >>= n;
z->num_bits -= n;
return k;
}
stbi_inline static int zhuffman_decode(zbuf *a, zhuffman *z)
{
int b,s,k;
if (a->num_bits < 16) fill_bits(a);
b = z->fast[a->code_buffer & ZFAST_MASK];
if (b < 0xffff) {
s = z->size[b];
a->code_buffer >>= s;
a->num_bits -= s;
return z->value[b];
}
// not resolved by fast table, so compute it the slow way
// use jpeg approach, which requires MSbits at top
k = bit_reverse(a->code_buffer, 16);
for (s=ZFAST_BITS+1; ; ++s)
if (k < z->maxcode[s])
break;
if (s == 16) return -1; // invalid code!
// code size is s, so:
b = (k >> (16-s)) - z->firstcode[s] + z->firstsymbol[s];
assert(z->size[b] == s);
a->code_buffer >>= s;
a->num_bits -= s;
return z->value[b];
}
static int expand(zbuf *z, int n) // need to make room for n bytes
{
char *q;
int cur, limit;
if (!z->z_expandable) return e("output buffer limit","Corrupt PNG");
cur = (int) (z->zout - z->zout_start);
limit = (int) (z->zout_end - z->zout_start);
while (cur + n > limit)
limit *= 2;
q = (char *) realloc(z->zout_start, limit);
if (q == NULL) return e("outofmem", "Out of memory");
z->zout_start = q;
z->zout = q + cur;
z->zout_end = q + limit;
return 1;
}
static int length_base[31] = {
3,4,5,6,7,8,9,10,11,13,
15,17,19,23,27,31,35,43,51,59,
67,83,99,115,131,163,195,227,258,0,0 };
static int length_extra[31]=
{ 0,0,0,0,0,0,0,0,1,1,1,1,2,2,2,2,3,3,3,3,4,4,4,4,5,5,5,5,0,0,0 };
static int dist_base[32] = { 1,2,3,4,5,7,9,13,17,25,33,49,65,97,129,193,
257,385,513,769,1025,1537,2049,3073,4097,6145,8193,12289,16385,24577,0,0};
static int dist_extra[32] =
{ 0,0,0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,11,11,12,12,13,13};
static int parse_huffman_block(zbuf *a)
{
for(;;) {
int z = zhuffman_decode(a, &a->z_length);
if (z < 256) {
if (z < 0) return e("bad huffman code","Corrupt PNG"); // error in huffman codes
if (a->zout >= a->zout_end) if (!expand(a, 1)) return 0;
*a->zout++ = (char) z;
} else {
uint8 *p;
int len,dist;
if (z == 256) return 1;
z -= 257;
len = length_base[z];
if (length_extra[z]) len += zreceive(a, length_extra[z]);
z = zhuffman_decode(a, &a->z_distance);
if (z < 0) return e("bad huffman code","Corrupt PNG");
dist = dist_base[z];
if (dist_extra[z]) dist += zreceive(a, dist_extra[z]);
if (a->zout - a->zout_start < dist) return e("bad dist","Corrupt PNG");
if (a->zout + len > a->zout_end) if (!expand(a, len)) return 0;
p = (uint8 *) (a->zout - dist);
while (len--)
*a->zout++ = *p++;
}
}
}
static int compute_huffman_codes(zbuf *a)
{
static uint8 length_dezigzag[19] = { 16,17,18,0,8,7,9,6,10,5,11,4,12,3,13,2,14,1,15 };
zhuffman z_codelength;
uint8 lencodes[286+32+137];//padding for maximum single op
uint8 codelength_sizes[19];
int i,n;
int hlit = zreceive(a,5) + 257;
int hdist = zreceive(a,5) + 1;
int hclen = zreceive(a,4) + 4;
memset(codelength_sizes, 0, sizeof(codelength_sizes));
for (i=0; i < hclen; ++i) {
int s = zreceive(a,3);
codelength_sizes[length_dezigzag[i]] = (uint8) s;
}
if (!zbuild_huffman(&z_codelength, codelength_sizes, 19)) return 0;
n = 0;
while (n < hlit + hdist) {
int c = zhuffman_decode(a, &z_codelength);
assert(c >= 0 && c < 19);
if (c < 16)
lencodes[n++] = (uint8) c;
else if (c == 16) {
c = zreceive(a,2)+3;
memset(lencodes+n, lencodes[n-1], c);
n += c;
} else if (c == 17) {
c = zreceive(a,3)+3;
memset(lencodes+n, 0, c);
n += c;
} else {
assert(c == 18);
c = zreceive(a,7)+11;
memset(lencodes+n, 0, c);
n += c;
}
}
if (n != hlit+hdist) return e("bad codelengths","Corrupt PNG");
if (!zbuild_huffman(&a->z_length, lencodes, hlit)) return 0;
if (!zbuild_huffman(&a->z_distance, lencodes+hlit, hdist)) return 0;
return 1;
}
static int parse_uncompressed_block(zbuf *a)
{
uint8 header[4];
int len,nlen,k;
if (a->num_bits & 7)
zreceive(a, a->num_bits & 7); // discard
// drain the bit-packed data into header
k = 0;
while (a->num_bits > 0) {
header[k++] = (uint8) (a->code_buffer & 255); // wtf this warns?
a->code_buffer >>= 8;
a->num_bits -= 8;
}
assert(a->num_bits == 0);
// now fill header the normal way
while (k < 4)
header[k++] = (uint8) zget8(a);
len = header[1] * 256 + header[0];
nlen = header[3] * 256 + header[2];
if (nlen != (len ^ 0xffff)) return e("zlib corrupt","Corrupt PNG");
if (a->zbuffer + len > a->zbuffer_end) return e("read past buffer","Corrupt PNG");
if (a->zout + len > a->zout_end)
if (!expand(a, len)) return 0;
memcpy(a->zout, a->zbuffer, len);
a->zbuffer += len;
a->zout += len;
return 1;
}
static int parse_zlib_header(zbuf *a)
{
int cmf = zget8(a);
int cm = cmf & 15;
/* int cinfo = cmf >> 4; */
int flg = zget8(a);
if ((cmf*256+flg) % 31 != 0) return e("bad zlib header","Corrupt PNG"); // zlib spec
if (flg & 32) return e("no preset dict","Corrupt PNG"); // preset dictionary not allowed in png
if (cm != 8) return e("bad compression","Corrupt PNG"); // DEFLATE required for png
// window = 1 << (8 + cinfo)... but who cares, we fully buffer output
return 1;
}
// @TODO: should statically initialize these for optimal thread safety
static uint8 default_length[288], default_distance[32];
static void init_defaults(void)
{
int i; // use <= to match clearly with spec
for (i=0; i <= 143; ++i) default_length[i] = 8;
for ( ; i <= 255; ++i) default_length[i] = 9;
for ( ; i <= 279; ++i) default_length[i] = 7;
for ( ; i <= 287; ++i) default_length[i] = 8;
for (i=0; i <= 31; ++i) default_distance[i] = 5;
}
int stbi_png_partial; // a quick hack to only allow decoding some of a PNG... I should implement real streaming support instead
static int parse_zlib(zbuf *a, int parse_header)
{
int final, type;
if (parse_header)
if (!parse_zlib_header(a)) return 0;
a->num_bits = 0;
a->code_buffer = 0;
do {
final = zreceive(a,1);
type = zreceive(a,2);
if (type == 0) {
if (!parse_uncompressed_block(a)) return 0;
} else if (type == 3) {
return 0;
} else {
if (type == 1) {
// use fixed code lengths
if (!default_distance[31]) init_defaults();
if (!zbuild_huffman(&a->z_length , default_length , 288)) return 0;
if (!zbuild_huffman(&a->z_distance, default_distance, 32)) return 0;
} else {
if (!compute_huffman_codes(a)) return 0;
}
if (!parse_huffman_block(a)) return 0;
}
if (stbi_png_partial && a->zout - a->zout_start > 65536)
break;
} while (!final);
return 1;
}
static int do_zlib(zbuf *a, char *obuf, int olen, int exp, int parse_header)
{
a->zout_start = obuf;
a->zout = obuf;
a->zout_end = obuf + olen;
a->z_expandable = exp;
return parse_zlib(a, parse_header);
}
char *stbi_zlib_decode_malloc_guesssize(const char *buffer, int len, int initial_size, int *outlen)
{
zbuf a;
char *p = (char *) malloc(initial_size);
if (p == NULL) return NULL;
a.zbuffer = (uint8 *) buffer;
a.zbuffer_end = (uint8 *) buffer + len;
if (do_zlib(&a, p, initial_size, 1, 1)) {
if (outlen) *outlen = (int) (a.zout - a.zout_start);
return a.zout_start;
} else {
free(a.zout_start);
return NULL;
}
}
char *stbi_zlib_decode_malloc(char const *buffer, int len, int *outlen)
{
return stbi_zlib_decode_malloc_guesssize(buffer, len, 16384, outlen);
}
char *stbi_zlib_decode_malloc_guesssize_headerflag(const char *buffer, int len, int initial_size, int *outlen, int parse_header)
{
zbuf a;
char *p = (char *) malloc(initial_size);
if (p == NULL) return NULL;
a.zbuffer = (uint8 *) buffer;
a.zbuffer_end = (uint8 *) buffer + len;
if (do_zlib(&a, p, initial_size, 1, parse_header)) {
if (outlen) *outlen = (int) (a.zout - a.zout_start);
return a.zout_start;
} else {
free(a.zout_start);
return NULL;
}
}
int stbi_zlib_decode_buffer(char *obuffer, int olen, char const *ibuffer, int ilen)
{
zbuf a;
a.zbuffer = (uint8 *) ibuffer;
a.zbuffer_end = (uint8 *) ibuffer + ilen;
if (do_zlib(&a, obuffer, olen, 0, 1))
return (int) (a.zout - a.zout_start);
else
return -1;
}
char *stbi_zlib_decode_noheader_malloc(char const *buffer, int len, int *outlen)
{
zbuf a;
char *p = (char *) malloc(16384);
if (p == NULL) return NULL;
a.zbuffer = (uint8 *) buffer;
a.zbuffer_end = (uint8 *) buffer+len;
if (do_zlib(&a, p, 16384, 1, 0)) {
if (outlen) *outlen = (int) (a.zout - a.zout_start);
return a.zout_start;
} else {
free(a.zout_start);
return NULL;
}
}
int stbi_zlib_decode_noheader_buffer(char *obuffer, int olen, const char *ibuffer, int ilen)
{
zbuf a;
a.zbuffer = (uint8 *) ibuffer;
a.zbuffer_end = (uint8 *) ibuffer + ilen;
if (do_zlib(&a, obuffer, olen, 0, 0))
return (int) (a.zout - a.zout_start);
else
return -1;
}
// public domain "baseline" PNG decoder v0.10 Sean Barrett 2006-11-18
// simple implementation
// - only 8-bit samples
// - no CRC checking
// - allocates lots of intermediate memory
// - avoids problem of streaming data between subsystems
// - avoids explicit window management
// performance
// - uses stb_zlib, a PD zlib implementation with fast huffman decoding
typedef struct
{
uint32 length;
uint32 type;
} chunk;
#define PNG_TYPE(a,b,c,d) (((a) << 24) + ((b) << 16) + ((c) << 8) + (d))
static chunk get_chunk_header(stbi *s)
{
chunk c;
c.length = get32(s);
c.type = get32(s);
return c;
}
static int check_png_header(stbi *s)
{
static uint8 png_sig[8] = { 137,80,78,71,13,10,26,10 };
int i;
for (i=0; i < 8; ++i)
if (get8u(s) != png_sig[i]) return e("bad png sig","Not a PNG");
return 1;
}
typedef struct
{
stbi *s;
uint8 *idata, *expanded, *out;
} png;
enum {
F_none=0, F_sub=1, F_up=2, F_avg=3, F_paeth=4,
F_avg_first, F_paeth_first
};
static uint8 first_row_filter[5] =
{
F_none, F_sub, F_none, F_avg_first, F_paeth_first
};
static int paeth(int a, int b, int c)
{
int p = a + b - c;
int pa = abs(p-a);
int pb = abs(p-b);
int pc = abs(p-c);
if (pa <= pb && pa <= pc) return a;
if (pb <= pc) return b;
return c;
}
// create the png data from post-deflated data
static int create_png_image_raw(png *a, uint8 *raw, uint32 raw_len, int out_n, uint32 x, uint32 y)
{
stbi *s = a->s;
uint32 i,j,stride = x*out_n;
int k;
int img_n = s->img_n; // copy it into a local for later
assert(out_n == s->img_n || out_n == s->img_n+1);
if (stbi_png_partial) y = 1;
a->out = (uint8 *) malloc(x * y * out_n);
if (!a->out) return e("outofmem", "Out of memory");
if (!stbi_png_partial) {
if (s->img_x == x && s->img_y == y) {
if (raw_len != (img_n * x + 1) * y) return e("not enough pixels","Corrupt PNG");
} else { // interlaced:
if (raw_len < (img_n * x + 1) * y) return e("not enough pixels","Corrupt PNG");
}
}
for (j=0; j < y; ++j) {
uint8 *cur = a->out + stride*j;
uint8 *prior = cur - stride;
int filter = *raw++;
if (filter > 4) return e("invalid filter","Corrupt PNG");
// if first row, use special filter that doesn't sample previous row
if (j == 0) filter = first_row_filter[filter];
// handle first pixel explicitly
for (k=0; k < img_n; ++k) {
switch (filter) {
case F_none : cur[k] = raw[k]; break;
case F_sub : cur[k] = raw[k]; break;
case F_up : cur[k] = raw[k] + prior[k]; break;
case F_avg : cur[k] = raw[k] + (prior[k]>>1); break;
case F_paeth : cur[k] = (uint8) (raw[k] + paeth(0,prior[k],0)); break;
case F_avg_first : cur[k] = raw[k]; break;
case F_paeth_first: cur[k] = raw[k]; break;
}
}
if (img_n != out_n) cur[img_n] = 255;
raw += img_n;
cur += out_n;
prior += out_n;
// this is a little gross, so that we don't switch per-pixel or per-component
if (img_n == out_n) {
#define CASE(f) \
case f: \
for (i=x-1; i >= 1; --i, raw+=img_n,cur+=img_n,prior+=img_n) \
for (k=0; k < img_n; ++k)
switch (filter) {
CASE(F_none) cur[k] = raw[k]; break;
CASE(F_sub) cur[k] = raw[k] + cur[k-img_n]; break;
CASE(F_up) cur[k] = raw[k] + prior[k]; break;
CASE(F_avg) cur[k] = raw[k] + ((prior[k] + cur[k-img_n])>>1); break;
CASE(F_paeth) cur[k] = (uint8) (raw[k] + paeth(cur[k-img_n],prior[k],prior[k-img_n])); break;
CASE(F_avg_first) cur[k] = raw[k] + (cur[k-img_n] >> 1); break;
CASE(F_paeth_first) cur[k] = (uint8) (raw[k] + paeth(cur[k-img_n],0,0)); break;
}
#undef CASE
} else {
assert(img_n+1 == out_n);
#define CASE(f) \
case f: \
for (i=x-1; i >= 1; --i, cur[img_n]=255,raw+=img_n,cur+=out_n,prior+=out_n) \
for (k=0; k < img_n; ++k)
switch (filter) {
CASE(F_none) cur[k] = raw[k]; break;
CASE(F_sub) cur[k] = raw[k] + cur[k-out_n]; break;
CASE(F_up) cur[k] = raw[k] + prior[k]; break;
CASE(F_avg) cur[k] = raw[k] + ((prior[k] + cur[k-out_n])>>1); break;
CASE(F_paeth) cur[k] = (uint8) (raw[k] + paeth(cur[k-out_n],prior[k],prior[k-out_n])); break;
CASE(F_avg_first) cur[k] = raw[k] + (cur[k-out_n] >> 1); break;
CASE(F_paeth_first) cur[k] = (uint8) (raw[k] + paeth(cur[k-out_n],0,0)); break;
}
#undef CASE
}
}
return 1;
}
static int create_png_image(png *a, uint8 *raw, uint32 raw_len, int out_n, int interlaced)
{
uint8 *final;
int p;
int save;
if (!interlaced)
return create_png_image_raw(a, raw, raw_len, out_n, a->s->img_x, a->s->img_y);
save = stbi_png_partial;
stbi_png_partial = 0;
// de-interlacing
final = (uint8 *) malloc(a->s->img_x * a->s->img_y * out_n);
for (p=0; p < 7; ++p) {
int xorig[] = { 0,4,0,2,0,1,0 };
int yorig[] = { 0,0,4,0,2,0,1 };
int xspc[] = { 8,8,4,4,2,2,1 };
int yspc[] = { 8,8,8,4,4,2,2 };
int i,j,x,y;
// pass1_x[4] = 0, pass1_x[5] = 1, pass1_x[12] = 1
x = (a->s->img_x - xorig[p] + xspc[p]-1) / xspc[p];
y = (a->s->img_y - yorig[p] + yspc[p]-1) / yspc[p];
if (x && y) {
if (!create_png_image_raw(a, raw, raw_len, out_n, x, y)) {
free(final);
return 0;
}
for (j=0; j < y; ++j)
for (i=0; i < x; ++i)
memcpy(final + (j*yspc[p]+yorig[p])*a->s->img_x*out_n + (i*xspc[p]+xorig[p])*out_n,
a->out + (j*x+i)*out_n, out_n);
free(a->out);
raw += (x*out_n+1)*y;
raw_len -= (x*out_n+1)*y;
}
}
a->out = final;
stbi_png_partial = save;
return 1;
}
static int compute_transparency(png *z, uint8 tc[3], int out_n)
{
stbi *s = z->s;
uint32 i, pixel_count = s->img_x * s->img_y;
uint8 *p = z->out;
// compute color-based transparency, assuming we've
// already got 255 as the alpha value in the output
assert(out_n == 2 || out_n == 4);
if (out_n == 2) {
for (i=0; i < pixel_count; ++i) {
p[1] = (p[0] == tc[0] ? 0 : 255);
p += 2;
}
} else {
for (i=0; i < pixel_count; ++i) {
if (p[0] == tc[0] && p[1] == tc[1] && p[2] == tc[2])
p[3] = 0;
p += 4;
}
}
return 1;
}
static int expand_palette(png *a, uint8 *palette, int len, int pal_img_n)
{
uint32 i, pixel_count = a->s->img_x * a->s->img_y;
uint8 *p, *temp_out, *orig = a->out;
p = (uint8 *) malloc(pixel_count * pal_img_n);
if (p == NULL) return e("outofmem", "Out of memory");
// between here and free(out) below, exitting would leak
temp_out = p;
if (pal_img_n == 3) {
for (i=0; i < pixel_count; ++i) {
int n = orig[i]*4;
p[0] = palette[n ];
p[1] = palette[n+1];
p[2] = palette[n+2];
p += 3;
}
} else {
for (i=0; i < pixel_count; ++i) {
int n = orig[i]*4;
p[0] = palette[n ];
p[1] = palette[n+1];
p[2] = palette[n+2];
p[3] = palette[n+3];
p += 4;
}
}
free(a->out);
a->out = temp_out;
STBI_NOTUSED(len);
return 1;
}
static int stbi_unpremultiply_on_load = 0;
static int stbi_de_iphone_flag = 0;
void stbi_set_unpremultiply_on_load(int flag_true_if_should_unpremultiply)
{
stbi_unpremultiply_on_load = flag_true_if_should_unpremultiply;
}
void stbi_convert_iphone_png_to_rgb(int flag_true_if_should_convert)
{
stbi_de_iphone_flag = flag_true_if_should_convert;
}
static void stbi_de_iphone(png *z)
{
stbi *s = z->s;
uint32 i, pixel_count = s->img_x * s->img_y;
uint8 *p = z->out;
if (s->img_out_n == 3) { // convert bgr to rgb
for (i=0; i < pixel_count; ++i) {
uint8 t = p[0];
p[0] = p[2];
p[2] = t;
p += 3;
}
} else {
assert(s->img_out_n == 4);
if (stbi_unpremultiply_on_load) {
// convert bgr to rgb and unpremultiply
for (i=0; i < pixel_count; ++i) {
uint8 a = p[3];
uint8 t = p[0];
if (a) {
p[0] = p[2] * 255 / a;
p[1] = p[1] * 255 / a;
p[2] = t * 255 / a;
} else {
p[0] = p[2];
p[2] = t;
}
p += 4;
}
} else {
// convert bgr to rgb
for (i=0; i < pixel_count; ++i) {
uint8 t = p[0];
p[0] = p[2];
p[2] = t;
p += 4;
}
}
}
}
static int parse_png_file(png *z, int scan, int req_comp)
{
uint8 palette[1024], pal_img_n=0;
uint8 has_trans=0, tc[3];
uint32 ioff=0, idata_limit=0, i, pal_len=0;
int first=1,k,interlace=0, iphone=0;
stbi *s = z->s;
z->expanded = NULL;
z->idata = NULL;
z->out = NULL;
if (!check_png_header(s)) return 0;
if (scan == SCAN_type) return 1;
for (;;) {
chunk c = get_chunk_header(s);
switch (c.type) {
case PNG_TYPE('C','g','B','I'):
iphone = stbi_de_iphone_flag;
skip(s, c.length);
break;
case PNG_TYPE('I','H','D','R'): {
int depth,color,comp,filter;
if (!first) return e("multiple IHDR","Corrupt PNG");
first = 0;
if (c.length != 13) return e("bad IHDR len","Corrupt PNG");
s->img_x = get32(s); if (s->img_x > (1 << 24)) return e("too large","Very large image (corrupt?)");
s->img_y = get32(s); if (s->img_y > (1 << 24)) return e("too large","Very large image (corrupt?)");
depth = get8(s); if (depth != 8) return e("8bit only","PNG not supported: 8-bit only");
color = get8(s); if (color > 6) return e("bad ctype","Corrupt PNG");
if (color == 3) pal_img_n = 3; else if (color & 1) return e("bad ctype","Corrupt PNG");
comp = get8(s); if (comp) return e("bad comp method","Corrupt PNG");
filter= get8(s); if (filter) return e("bad filter method","Corrupt PNG");
interlace = get8(s); if (interlace>1) return e("bad interlace method","Corrupt PNG");
if (!s->img_x || !s->img_y) return e("0-pixel image","Corrupt PNG");
if (!pal_img_n) {
s->img_n = (color & 2 ? 3 : 1) + (color & 4 ? 1 : 0);
if ((1 << 30) / s->img_x / s->img_n < s->img_y) return e("too large", "Image too large to decode");
if (scan == SCAN_header) return 1;
} else {
// if paletted, then pal_n is our final components, and
// img_n is # components to decompress/filter.
s->img_n = 1;
if ((1 << 30) / s->img_x / 4 < s->img_y) return e("too large","Corrupt PNG");
// if SCAN_header, have to scan to see if we have a tRNS
}
break;
}
case PNG_TYPE('P','L','T','E'): {
if (first) return e("first not IHDR", "Corrupt PNG");
if (c.length > 256*3) return e("invalid PLTE","Corrupt PNG");
pal_len = c.length / 3;
if (pal_len * 3 != c.length) return e("invalid PLTE","Corrupt PNG");
for (i=0; i < pal_len; ++i) {
palette[i*4+0] = get8u(s);
palette[i*4+1] = get8u(s);
palette[i*4+2] = get8u(s);
palette[i*4+3] = 255;
}
break;
}
case PNG_TYPE('t','R','N','S'): {
if (first) return e("first not IHDR", "Corrupt PNG");
if (z->idata) return e("tRNS after IDAT","Corrupt PNG");
if (pal_img_n) {
if (scan == SCAN_header) { s->img_n = 4; return 1; }
if (pal_len == 0) return e("tRNS before PLTE","Corrupt PNG");
if (c.length > pal_len) return e("bad tRNS len","Corrupt PNG");
pal_img_n = 4;
for (i=0; i < c.length; ++i)
palette[i*4+3] = get8u(s);
} else {
if (!(s->img_n & 1)) return e("tRNS with alpha","Corrupt PNG");
if (c.length != (uint32) s->img_n*2) return e("bad tRNS len","Corrupt PNG");
has_trans = 1;
for (k=0; k < s->img_n; ++k)
tc[k] = (uint8) get16(s); // non 8-bit images will be larger
}
break;
}
case PNG_TYPE('I','D','A','T'): {
if (first) return e("first not IHDR", "Corrupt PNG");
if (pal_img_n && !pal_len) return e("no PLTE","Corrupt PNG");
if (scan == SCAN_header) { s->img_n = pal_img_n; return 1; }
if (ioff + c.length > idata_limit) {
uint8 *p;
if (idata_limit == 0) idata_limit = c.length > 4096 ? c.length : 4096;
while (ioff + c.length > idata_limit)
idata_limit *= 2;
p = (uint8 *) realloc(z->idata, idata_limit); if (p == NULL) return e("outofmem", "Out of memory");
z->idata = p;
}
if (!getn(s, z->idata+ioff,c.length)) return e("outofdata","Corrupt PNG");
ioff += c.length;
break;
}
case PNG_TYPE('I','E','N','D'): {
uint32 raw_len;
if (first) return e("first not IHDR", "Corrupt PNG");
if (scan != SCAN_load) return 1;
if (z->idata == NULL) return e("no IDAT","Corrupt PNG");
z->expanded = (uint8 *) stbi_zlib_decode_malloc_guesssize_headerflag((char *) z->idata, ioff, 16384, (int *) &raw_len, !iphone);
if (z->expanded == NULL) return 0; // zlib should set error
free(z->idata); z->idata = NULL;
if ((req_comp == s->img_n+1 && req_comp != 3 && !pal_img_n) || has_trans)
s->img_out_n = s->img_n+1;
else
s->img_out_n = s->img_n;
if (!create_png_image(z, z->expanded, raw_len, s->img_out_n, interlace)) return 0;
if (has_trans)
if (!compute_transparency(z, tc, s->img_out_n)) return 0;
if (iphone && s->img_out_n > 2)
stbi_de_iphone(z);
if (pal_img_n) {
// pal_img_n == 3 or 4
s->img_n = pal_img_n; // record the actual colors we had
s->img_out_n = pal_img_n;
if (req_comp >= 3) s->img_out_n = req_comp;
if (!expand_palette(z, palette, pal_len, s->img_out_n))
return 0;
}
free(z->expanded); z->expanded = NULL;
return 1;
}
default:
// if critical, fail
if (first) return e("first not IHDR", "Corrupt PNG");
if ((c.type & (1 << 29)) == 0) {
#ifndef STBI_NO_FAILURE_STRINGS
// not threadsafe
static char invalid_chunk[] = "XXXX chunk not known";
invalid_chunk[0] = (uint8) (c.type >> 24);
invalid_chunk[1] = (uint8) (c.type >> 16);
invalid_chunk[2] = (uint8) (c.type >> 8);
invalid_chunk[3] = (uint8) (c.type >> 0);
#endif
return e(invalid_chunk, "PNG not supported: unknown chunk type");
}
skip(s, c.length);
break;
}
// end of chunk, read and skip CRC
get32(s);
}
}
static unsigned char *do_png(png *p, int *x, int *y, int *n, int req_comp)
{
unsigned char *result=NULL;
if (req_comp < 0 || req_comp > 4) return epuc("bad req_comp", "Internal error");
if (parse_png_file(p, SCAN_load, req_comp)) {
result = p->out;
p->out = NULL;
if (req_comp && req_comp != p->s->img_out_n) {
result = convert_format(result, p->s->img_out_n, req_comp, p->s->img_x, p->s->img_y);
p->s->img_out_n = req_comp;
if (result == NULL) return result;
}
*x = p->s->img_x;
*y = p->s->img_y;
if (n) *n = p->s->img_n;
}
free(p->out); p->out = NULL;
free(p->expanded); p->expanded = NULL;
free(p->idata); p->idata = NULL;
return result;
}
static unsigned char *stbi_png_load(stbi *s, int *x, int *y, int *comp, int req_comp)
{
png p;
p.s = s;
return do_png(&p, x,y,comp,req_comp);
}
static int stbi_png_test(stbi *s)
{
int r;
r = check_png_header(s);
stbi_rewind(s);
return r;
}
static int stbi_png_info_raw(png *p, int *x, int *y, int *comp)
{
if (!parse_png_file(p, SCAN_header, 0)) {
stbi_rewind( p->s );
return 0;
}
if (x) *x = p->s->img_x;
if (y) *y = p->s->img_y;
if (comp) *comp = p->s->img_n;
return 1;
}
static int stbi_png_info(stbi *s, int *x, int *y, int *comp)
{
png p;
p.s = s;
return stbi_png_info_raw(&p, x, y, comp);
}
// Microsoft/Windows BMP image
static int bmp_test(stbi *s)
{
int sz;
if (get8(s) != 'B') return 0;
if (get8(s) != 'M') return 0;
get32le(s); // discard filesize
get16le(s); // discard reserved
get16le(s); // discard reserved
get32le(s); // discard data offset
sz = get32le(s);
if (sz == 12 || sz == 40 || sz == 56 || sz == 108) return 1;
return 0;
}
static int stbi_bmp_test(stbi *s)
{
int r = bmp_test(s);
stbi_rewind(s);
return r;
}
// returns 0..31 for the highest set bit
static int high_bit(unsigned int z)
{
int n=0;
if (z == 0) return -1;
if (z >= 0x10000) n += 16, z >>= 16;
if (z >= 0x00100) n += 8, z >>= 8;
if (z >= 0x00010) n += 4, z >>= 4;
if (z >= 0x00004) n += 2, z >>= 2;
if (z >= 0x00002) n += 1, z >>= 1;
return n;
}
static int bitcount(unsigned int a)
{
a = (a & 0x55555555) + ((a >> 1) & 0x55555555); // max 2
a = (a & 0x33333333) + ((a >> 2) & 0x33333333); // max 4
a = (a + (a >> 4)) & 0x0f0f0f0f; // max 8 per 4, now 8 bits
a = (a + (a >> 8)); // max 16 per 8 bits
a = (a + (a >> 16)); // max 32 per 8 bits
return a & 0xff;
}
static int shiftsigned(int v, int shift, int bits)
{
int result;
int z=0;
if (shift < 0) v <<= -shift;
else v >>= shift;
result = v;
z = bits;
while (z < 8) {
result += v >> z;
z += bits;
}
return result;
}
static stbi_uc *bmp_load(stbi *s, int *x, int *y, int *comp, int req_comp)
{
uint8 *out;
unsigned int mr=0,mg=0,mb=0,ma=0, fake_a=0;
stbi_uc pal[256][4];
int psize=0,i,j,compress=0,width;
int bpp, flip_vertically, pad, target, offset, hsz;
if (get8(s) != 'B' || get8(s) != 'M') return epuc("not BMP", "Corrupt BMP");
get32le(s); // discard filesize
get16le(s); // discard reserved
get16le(s); // discard reserved
offset = get32le(s);
hsz = get32le(s);
if (hsz != 12 && hsz != 40 && hsz != 56 && hsz != 108) return epuc("unknown BMP", "BMP type not supported: unknown");
if (hsz == 12) {
s->img_x = get16le(s);
s->img_y = get16le(s);
} else {
s->img_x = get32le(s);
s->img_y = get32le(s);
}
if (get16le(s) != 1) return epuc("bad BMP", "bad BMP");
bpp = get16le(s);
if (bpp == 1) return epuc("monochrome", "BMP type not supported: 1-bit");
flip_vertically = ((int) s->img_y) > 0;
s->img_y = abs((int) s->img_y);
if (hsz == 12) {
if (bpp < 24)
psize = (offset - 14 - 24) / 3;
} else {
compress = get32le(s);
if (compress == 1 || compress == 2) return epuc("BMP RLE", "BMP type not supported: RLE");
get32le(s); // discard sizeof
get32le(s); // discard hres
get32le(s); // discard vres
get32le(s); // discard colorsused
get32le(s); // discard max important
if (hsz == 40 || hsz == 56) {
if (hsz == 56) {
get32le(s);
get32le(s);
get32le(s);
get32le(s);
}
if (bpp == 16 || bpp == 32) {
mr = mg = mb = 0;
if (compress == 0) {
if (bpp == 32) {
mr = 0xffu << 16;
mg = 0xffu << 8;
mb = 0xffu << 0;
ma = 0xffu << 24;
fake_a = 1; // @TODO: check for cases like alpha value is all 0 and switch it to 255
} else {
mr = 31u << 10;
mg = 31u << 5;
mb = 31u << 0;
}
} else if (compress == 3) {
mr = get32le(s);
mg = get32le(s);
mb = get32le(s);
// not documented, but generated by photoshop and handled by mspaint
if (mr == mg && mg == mb) {
// ?!?!?
return epuc("bad BMP", "bad BMP");
}
} else
return epuc("bad BMP", "bad BMP");
}
} else {
assert(hsz == 108);
mr = get32le(s);
mg = get32le(s);
mb = get32le(s);
ma = get32le(s);
get32le(s); // discard color space
for (i=0; i < 12; ++i)
get32le(s); // discard color space parameters
}
if (bpp < 16)
psize = (offset - 14 - hsz) >> 2;
}
s->img_n = ma ? 4 : 3;
if (req_comp && req_comp >= 3) // we can directly decode 3 or 4
target = req_comp;
else
target = s->img_n; // if they want monochrome, we'll post-convert
out = (stbi_uc *) malloc(target * s->img_x * s->img_y);
if (!out) return epuc("outofmem", "Out of memory");
if (bpp < 16) {
int z=0;
if (psize == 0 || psize > 256) { free(out); return epuc("invalid", "Corrupt BMP"); }
for (i=0; i < psize; ++i) {
pal[i][2] = get8u(s);
pal[i][1] = get8u(s);
pal[i][0] = get8u(s);
if (hsz != 12) get8(s);
pal[i][3] = 255;
}
skip(s, offset - 14 - hsz - psize * (hsz == 12 ? 3 : 4));
if (bpp == 4) width = (s->img_x + 1) >> 1;
else if (bpp == 8) width = s->img_x;
else { free(out); return epuc("bad bpp", "Corrupt BMP"); }
pad = (-width)&3;
for (j=0; j < (int) s->img_y; ++j) {
for (i=0; i < (int) s->img_x; i += 2) {
int v=get8(s),v2=0;
if (bpp == 4) {
v2 = v & 15;
v >>= 4;
}
out[z++] = pal[v][0];
out[z++] = pal[v][1];
out[z++] = pal[v][2];
if (target == 4) out[z++] = 255;
if (i+1 == (int) s->img_x) break;
v = (bpp == 8) ? get8(s) : v2;
out[z++] = pal[v][0];
out[z++] = pal[v][1];
out[z++] = pal[v][2];
if (target == 4) out[z++] = 255;
}
skip(s, pad);
}
} else {
int rshift=0,gshift=0,bshift=0,ashift=0,rcount=0,gcount=0,bcount=0,acount=0;
int z = 0;
int easy=0;
skip(s, offset - 14 - hsz);
if (bpp == 24) width = 3 * s->img_x;
else if (bpp == 16) width = 2*s->img_x;
else /* bpp = 32 and pad = 0 */ width=0;
pad = (-width) & 3;
if (bpp == 24) {
easy = 1;
} else if (bpp == 32) {
if (mb == 0xff && mg == 0xff00 && mr == 0x00ff0000 && ma == 0xff000000)
easy = 2;
}
if (!easy) {
if (!mr || !mg || !mb) { free(out); return epuc("bad masks", "Corrupt BMP"); }
// right shift amt to put high bit in position #7
rshift = high_bit(mr)-7; rcount = bitcount(mr);
gshift = high_bit(mg)-7; gcount = bitcount(mr);
bshift = high_bit(mb)-7; bcount = bitcount(mr);
ashift = high_bit(ma)-7; acount = bitcount(mr);
}
for (j=0; j < (int) s->img_y; ++j) {
if (easy) {
for (i=0; i < (int) s->img_x; ++i) {
int a;
out[z+2] = get8u(s);
out[z+1] = get8u(s);
out[z+0] = get8u(s);
z += 3;
a = (easy == 2 ? get8(s) : 255);
if (target == 4) out[z++] = (uint8) a;
}
} else {
for (i=0; i < (int) s->img_x; ++i) {
uint32 v = (bpp == 16 ? get16le(s) : get32le(s));
int a;
out[z++] = (uint8) shiftsigned(v & mr, rshift, rcount);
out[z++] = (uint8) shiftsigned(v & mg, gshift, gcount);
out[z++] = (uint8) shiftsigned(v & mb, bshift, bcount);
a = (ma ? shiftsigned(v & ma, ashift, acount) : 255);
if (target == 4) out[z++] = (uint8) a;
}
}
skip(s, pad);
}
}
if (flip_vertically) {
stbi_uc t;
for (j=0; j < (int) s->img_y>>1; ++j) {
stbi_uc *p1 = out + j *s->img_x*target;
stbi_uc *p2 = out + (s->img_y-1-j)*s->img_x*target;
for (i=0; i < (int) s->img_x*target; ++i) {
t = p1[i], p1[i] = p2[i], p2[i] = t;
}
}
}
if (req_comp && req_comp != target) {
out = convert_format(out, target, req_comp, s->img_x, s->img_y);
if (out == NULL) return out; // convert_format frees input on failure
}
*x = s->img_x;
*y = s->img_y;
if (comp) *comp = s->img_n;
return out;
}
static stbi_uc *stbi_bmp_load(stbi *s,int *x, int *y, int *comp, int req_comp)
{
return bmp_load(s, x,y,comp,req_comp);
}
// Targa Truevision - TGA
// by Jonathan Dummer
static int tga_info(stbi *s, int *x, int *y, int *comp)
{
int tga_w, tga_h, tga_comp;
int sz;
get8u(s); // discard Offset
sz = get8u(s); // color type
if( sz > 1 ) {
stbi_rewind(s);
return 0; // only RGB or indexed allowed
}
sz = get8u(s); // image type
// only RGB or grey allowed, +/- RLE
if ((sz != 1) && (sz != 2) && (sz != 3) && (sz != 9) && (sz != 10) && (sz != 11)) return 0;
skip(s,9);
tga_w = get16le(s);
if( tga_w < 1 ) {
stbi_rewind(s);
return 0; // test width
}
tga_h = get16le(s);
if( tga_h < 1 ) {
stbi_rewind(s);
return 0; // test height
}
sz = get8(s); // bits per pixel
// only RGB or RGBA or grey allowed
if ((sz != 8) && (sz != 16) && (sz != 24) && (sz != 32)) {
stbi_rewind(s);
return 0;
}
tga_comp = sz;
if (x) *x = tga_w;
if (y) *y = tga_h;
if (comp) *comp = tga_comp / 8;
return 1; // seems to have passed everything
}
int stbi_tga_info(stbi *s, int *x, int *y, int *comp)
{
return tga_info(s, x, y, comp);
}
static int tga_test(stbi *s)
{
int sz;
get8u(s); // discard Offset
sz = get8u(s); // color type
if ( sz > 1 ) return 0; // only RGB or indexed allowed
sz = get8u(s); // image type
if ( (sz != 1) && (sz != 2) && (sz != 3) && (sz != 9) && (sz != 10) && (sz != 11) ) return 0; // only RGB or grey allowed, +/- RLE
get16(s); // discard palette start
get16(s); // discard palette length
get8(s); // discard bits per palette color entry
get16(s); // discard x origin
get16(s); // discard y origin
if ( get16(s) < 1 ) return 0; // test width
if ( get16(s) < 1 ) return 0; // test height
sz = get8(s); // bits per pixel
if ( (sz != 8) && (sz != 16) && (sz != 24) && (sz != 32) ) return 0; // only RGB or RGBA or grey allowed
return 1; // seems to have passed everything
}
static int stbi_tga_test(stbi *s)
{
int res = tga_test(s);
stbi_rewind(s);
return res;
}
static stbi_uc *tga_load(stbi *s, int *x, int *y, int *comp, int req_comp)
{
// read in the TGA header stuff
int tga_offset = get8u(s);
int tga_indexed = get8u(s);
int tga_image_type = get8u(s);
int tga_is_RLE = 0;
int tga_palette_start = get16le(s);
int tga_palette_len = get16le(s);
int tga_palette_bits = get8u(s);
int tga_x_origin = get16le(s);
int tga_y_origin = get16le(s);
int tga_width = get16le(s);
int tga_height = get16le(s);
int tga_bits_per_pixel = get8u(s);
int tga_inverted = get8u(s);
// image data
unsigned char *tga_data;
unsigned char *tga_palette = NULL;
int i, j;
unsigned char raw_data[4];
unsigned char trans_data[4];
int RLE_count = 0;
int RLE_repeating = 0;
int read_next_pixel = 1;
// do a tiny bit of precessing
if ( tga_image_type >= 8 )
{
tga_image_type -= 8;
tga_is_RLE = 1;
}
/* int tga_alpha_bits = tga_inverted & 15; */
tga_inverted = 1 - ((tga_inverted >> 5) & 1);
// error check
if ( //(tga_indexed) ||
(tga_width < 1) || (tga_height < 1) ||
(tga_image_type < 1) || (tga_image_type > 3) ||
((tga_bits_per_pixel != 8) && (tga_bits_per_pixel != 16) &&
(tga_bits_per_pixel != 24) && (tga_bits_per_pixel != 32))
)
{
return NULL; // we don't report this as a bad TGA because we don't even know if it's TGA
}
// If I'm paletted, then I'll use the number of bits from the palette
if ( tga_indexed )
{
tga_bits_per_pixel = tga_palette_bits;
}
// tga info
*x = tga_width;
*y = tga_height;
if ( (req_comp < 1) || (req_comp > 4) )
{
// just use whatever the file was
req_comp = tga_bits_per_pixel / 8;
*comp = req_comp;
} else
{
// force a new number of components
*comp = tga_bits_per_pixel/8;
}
tga_data = (unsigned char*)malloc( tga_width * tga_height * req_comp );
if (!tga_data) return epuc("outofmem", "Out of memory");
// skip to the data's starting position (offset usually = 0)
skip(s, tga_offset );
// do I need to load a palette?
if ( tga_indexed )
{
// any data to skip? (offset usually = 0)
skip(s, tga_palette_start );
// load the palette
tga_palette = (unsigned char*)malloc( tga_palette_len * tga_palette_bits / 8 );
if (!tga_palette) return epuc("outofmem", "Out of memory");
if (!getn(s, tga_palette, tga_palette_len * tga_palette_bits / 8 )) {
free(tga_data);
free(tga_palette);
return epuc("bad palette", "Corrupt TGA");
}
}
// load the data
trans_data[0] = trans_data[1] = trans_data[2] = trans_data[3] = 0;
for (i=0; i < tga_width * tga_height; ++i)
{
// if I'm in RLE mode, do I need to get a RLE chunk?
if ( tga_is_RLE )
{
if ( RLE_count == 0 )
{
// yep, get the next byte as a RLE command
int RLE_cmd = get8u(s);
RLE_count = 1 + (RLE_cmd & 127);
RLE_repeating = RLE_cmd >> 7;
read_next_pixel = 1;
} else if ( !RLE_repeating )
{
read_next_pixel = 1;
}
} else
{
read_next_pixel = 1;
}
// OK, if I need to read a pixel, do it now
if ( read_next_pixel )
{
// load however much data we did have
if ( tga_indexed )
{
// read in 1 byte, then perform the lookup
int pal_idx = get8u(s);
if ( pal_idx >= tga_palette_len )
{
// invalid index
pal_idx = 0;
}
pal_idx *= tga_bits_per_pixel / 8;
for (j = 0; j*8 < tga_bits_per_pixel; ++j)
{
raw_data[j] = tga_palette[pal_idx+j];
}
} else
{
// read in the data raw
for (j = 0; j*8 < tga_bits_per_pixel; ++j)
{
raw_data[j] = get8u(s);
}
}
// convert raw to the intermediate format
switch (tga_bits_per_pixel)
{
case 8:
// Luminous => RGBA
trans_data[0] = raw_data[0];
trans_data[1] = raw_data[0];
trans_data[2] = raw_data[0];
trans_data[3] = 255;
break;
case 16:
// Luminous,Alpha => RGBA
trans_data[0] = raw_data[0];
trans_data[1] = raw_data[0];
trans_data[2] = raw_data[0];
trans_data[3] = raw_data[1];
break;
case 24:
// BGR => RGBA
trans_data[0] = raw_data[2];
trans_data[1] = raw_data[1];
trans_data[2] = raw_data[0];
trans_data[3] = 255;
break;
case 32:
// BGRA => RGBA
trans_data[0] = raw_data[2];
trans_data[1] = raw_data[1];
trans_data[2] = raw_data[0];
trans_data[3] = raw_data[3];
break;
}
// clear the reading flag for the next pixel
read_next_pixel = 0;
} // end of reading a pixel
// convert to final format
switch (req_comp)
{
case 1:
// RGBA => Luminance
tga_data[i*req_comp+0] = compute_y(trans_data[0],trans_data[1],trans_data[2]);
break;
case 2:
// RGBA => Luminance,Alpha
tga_data[i*req_comp+0] = compute_y(trans_data[0],trans_data[1],trans_data[2]);
tga_data[i*req_comp+1] = trans_data[3];
break;
case 3:
// RGBA => RGB
tga_data[i*req_comp+0] = trans_data[0];
tga_data[i*req_comp+1] = trans_data[1];
tga_data[i*req_comp+2] = trans_data[2];
break;
case 4:
// RGBA => RGBA
tga_data[i*req_comp+0] = trans_data[0];
tga_data[i*req_comp+1] = trans_data[1];
tga_data[i*req_comp+2] = trans_data[2];
tga_data[i*req_comp+3] = trans_data[3];
break;
}
// in case we're in RLE mode, keep counting down
--RLE_count;
}
// do I need to invert the image?
if ( tga_inverted )
{
for (j = 0; j*2 < tga_height; ++j)
{
int index1 = j * tga_width * req_comp;
int index2 = (tga_height - 1 - j) * tga_width * req_comp;
for (i = tga_width * req_comp; i > 0; --i)
{
unsigned char temp = tga_data[index1];
tga_data[index1] = tga_data[index2];
tga_data[index2] = temp;
++index1;
++index2;
}
}
}
// clear my palette, if I had one
if ( tga_palette != NULL )
{
free( tga_palette );
}
// the things I do to get rid of an error message, and yet keep
// Microsoft's C compilers happy... [8^(
tga_palette_start = tga_palette_len = tga_palette_bits =
tga_x_origin = tga_y_origin = 0;
// OK, done
return tga_data;
}
static stbi_uc *stbi_tga_load(stbi *s, int *x, int *y, int *comp, int req_comp)
{
return tga_load(s,x,y,comp,req_comp);
}
// *************************************************************************************************
// Photoshop PSD loader -- PD by Thatcher Ulrich, integration by Nicolas Schulz, tweaked by STB
static int psd_test(stbi *s)
{
if (get32(s) != 0x38425053) return 0; // "8BPS"
else return 1;
}
static int stbi_psd_test(stbi *s)
{
int r = psd_test(s);
stbi_rewind(s);
return r;
}
static stbi_uc *psd_load(stbi *s, int *x, int *y, int *comp, int req_comp)
{
int pixelCount;
int channelCount, compression;
int channel, i, count, len;
int w,h;
uint8 *out;
// Check identifier
if (get32(s) != 0x38425053) // "8BPS"
return epuc("not PSD", "Corrupt PSD image");
// Check file type version.
if (get16(s) != 1)
return epuc("wrong version", "Unsupported version of PSD image");
// Skip 6 reserved bytes.
skip(s, 6 );
// Read the number of channels (R, G, B, A, etc).
channelCount = get16(s);
if (channelCount < 0 || channelCount > 16)
return epuc("wrong channel count", "Unsupported number of channels in PSD image");
// Read the rows and columns of the image.
h = get32(s);
w = get32(s);
// Make sure the depth is 8 bits.
if (get16(s) != 8)
return epuc("unsupported bit depth", "PSD bit depth is not 8 bit");
// Make sure the color mode is RGB.
// Valid options are:
// 0: Bitmap
// 1: Grayscale
// 2: Indexed color
// 3: RGB color
// 4: CMYK color
// 7: Multichannel
// 8: Duotone
// 9: Lab color
if (get16(s) != 3)
return epuc("wrong color format", "PSD is not in RGB color format");
// Skip the Mode Data. (It's the palette for indexed color; other info for other modes.)
skip(s,get32(s) );
// Skip the image resources. (resolution, pen tool paths, etc)
skip(s, get32(s) );
// Skip the reserved data.
skip(s, get32(s) );
// Find out if the data is compressed.
// Known values:
// 0: no compression
// 1: RLE compressed
compression = get16(s);
if (compression > 1)
return epuc("bad compression", "PSD has an unknown compression format");
// Create the destination image.
out = (stbi_uc *) malloc(4 * w*h);
if (!out) return epuc("outofmem", "Out of memory");
pixelCount = w*h;
// Initialize the data to zero.
//memset( out, 0, pixelCount * 4 );
// Finally, the image data.
if (compression) {
// RLE as used by .PSD and .TIFF
// Loop until you get the number of unpacked bytes you are expecting:
// Read the next source byte into n.
// If n is between 0 and 127 inclusive, copy the next n+1 bytes literally.
// Else if n is between -127 and -1 inclusive, copy the next byte -n+1 times.
// Else if n is 128, noop.
// Endloop
// The RLE-compressed data is preceeded by a 2-byte data count for each row in the data,
// which we're going to just skip.
skip(s, h * channelCount * 2 );
// Read the RLE data by channel.
for (channel = 0; channel < 4; channel++) {
uint8 *p;
p = out+channel;
if (channel >= channelCount) {
// Fill this channel with default data.
for (i = 0; i < pixelCount; i++) *p = (channel == 3 ? 255 : 0), p += 4;
} else {
// Read the RLE data.
count = 0;
while (count < pixelCount) {
len = get8(s);
if (len == 128) {
// No-op.
} else if (len < 128) {
// Copy next len+1 bytes literally.
len++;
count += len;
while (len) {
*p = get8u(s);
p += 4;
len--;
}
} else if (len > 128) {
uint8 val;
// Next -len+1 bytes in the dest are replicated from next source byte.
// (Interpret len as a negative 8-bit int.)
len ^= 0x0FF;
len += 2;
val = get8u(s);
count += len;
while (len) {
*p = val;
p += 4;
len--;
}
}
}
}
}
} else {
// We're at the raw image data. It's each channel in order (Red, Green, Blue, Alpha, ...)
// where each channel consists of an 8-bit value for each pixel in the image.
// Read the data by channel.
for (channel = 0; channel < 4; channel++) {
uint8 *p;
p = out + channel;
if (channel > channelCount) {
// Fill this channel with default data.
for (i = 0; i < pixelCount; i++) *p = channel == 3 ? 255 : 0, p += 4;
} else {
// Read the data.
for (i = 0; i < pixelCount; i++)
*p = get8u(s), p += 4;
}
}
}
if (req_comp && req_comp != 4) {
out = convert_format(out, 4, req_comp, w, h);
if (out == NULL) return out; // convert_format frees input on failure
}
if (comp) *comp = channelCount;
*y = h;
*x = w;
return out;
}
static stbi_uc *stbi_psd_load(stbi *s, int *x, int *y, int *comp, int req_comp)
{
return psd_load(s,x,y,comp,req_comp);
}
// *************************************************************************************************
// Softimage PIC loader
// by Tom Seddon
//
// See http://softimage.wiki.softimage.com/index.php/INFO:_PIC_file_format
// See http://ozviz.wasp.uwa.edu.au/~pbourke/dataformats/softimagepic/
static int pic_is4(stbi *s,const char *str)
{
int i;
for (i=0; i<4; ++i)
if (get8(s) != (stbi_uc)str[i])
return 0;
return 1;
}
static int pic_test(stbi *s)
{
int i;
if (!pic_is4(s,"\x53\x80\xF6\x34"))
return 0;
for(i=0;i<84;++i)
get8(s);
if (!pic_is4(s,"PICT"))
return 0;
return 1;
}
typedef struct
{
stbi_uc size,type,channel;
} pic_packet_t;
static stbi_uc *pic_readval(stbi *s, int channel, stbi_uc *dest)
{
int mask=0x80, i;
for (i=0; i<4; ++i, mask>>=1) {
if (channel & mask) {
if (at_eof(s)) return epuc("bad file","PIC file too short");
dest[i]=get8u(s);
}
}
return dest;
}
static void pic_copyval(int channel,stbi_uc *dest,const stbi_uc *src)
{
int mask=0x80,i;
for (i=0;i<4; ++i, mask>>=1)
if (channel&mask)
dest[i]=src[i];
}
static stbi_uc *pic_load2(stbi *s,int width,int height,int *comp, stbi_uc *result)
{
int act_comp=0,num_packets=0,y,chained;
pic_packet_t packets[10];
// this will (should...) cater for even some bizarre stuff like having data
// for the same channel in multiple packets.
do {
pic_packet_t *packet;
if (num_packets==sizeof(packets)/sizeof(packets[0]))
return epuc("bad format","too many packets");
packet = &packets[num_packets++];
chained = get8(s);
packet->size = get8u(s);
packet->type = get8u(s);
packet->channel = get8u(s);
act_comp |= packet->channel;
if (at_eof(s)) return epuc("bad file","file too short (reading packets)");
if (packet->size != 8) return epuc("bad format","packet isn't 8bpp");
} while (chained);
*comp = (act_comp & 0x10 ? 4 : 3); // has alpha channel?
for(y=0; y<height; ++y) {
int packet_idx;
for(packet_idx=0; packet_idx < num_packets; ++packet_idx) {
pic_packet_t *packet = &packets[packet_idx];
stbi_uc *dest = result+y*width*4;
switch (packet->type) {
default:
return epuc("bad format","packet has bad compression type");
case 0: {//uncompressed
int x;
for(x=0;x<width;++x, dest+=4)
if (!pic_readval(s,packet->channel,dest))
return 0;
break;
}
case 1://Pure RLE
{
int left=width, i;
while (left>0) {
stbi_uc count,value[4];
count=get8u(s);
if (at_eof(s)) return epuc("bad file","file too short (pure read count)");
if (count > left)
count = (uint8) left;
if (!pic_readval(s,packet->channel,value)) return 0;
for(i=0; i<count; ++i,dest+=4)
pic_copyval(packet->channel,dest,value);
left -= count;
}
}
break;
case 2: {//Mixed RLE
int left=width;
while (left>0) {
int count = get8(s), i;
if (at_eof(s)) return epuc("bad file","file too short (mixed read count)");
if (count >= 128) { // Repeated
stbi_uc value[4];
int i;
if (count==128)
count = get16(s);
else
count -= 127;
if (count > left)
return epuc("bad file","scanline overrun");
if (!pic_readval(s,packet->channel,value))
return 0;
for(i=0;i<count;++i, dest += 4)
pic_copyval(packet->channel,dest,value);
} else { // Raw
++count;
if (count>left) return epuc("bad file","scanline overrun");
for(i=0;i<count;++i, dest+=4)
if (!pic_readval(s,packet->channel,dest))
return 0;
}
left-=count;
}
break;
}
}
}
}
return result;
}
static stbi_uc *pic_load(stbi *s,int *px,int *py,int *comp,int req_comp)
{
stbi_uc *result;
int i, x,y;
for (i=0; i<92; ++i)
get8(s);
x = get16(s);
y = get16(s);
if (at_eof(s)) return epuc("bad file","file too short (pic header)");
if ((1 << 28) / x < y) return epuc("too large", "Image too large to decode");
get32(s); //skip `ratio'
get16(s); //skip `fields'
get16(s); //skip `pad'
// intermediate buffer is RGBA
result = (stbi_uc *) malloc(x*y*4);
memset(result, 0xff, x*y*4);
if (!pic_load2(s,x,y,comp, result)) {
free(result);
result=0;
}
*px = x;
*py = y;
if (req_comp == 0) req_comp = *comp;
result=convert_format(result,4,req_comp,x,y);
return result;
}
static int stbi_pic_test(stbi *s)
{
int r = pic_test(s);
stbi_rewind(s);
return r;
}
static stbi_uc *stbi_pic_load(stbi *s, int *x, int *y, int *comp, int req_comp)
{
return pic_load(s,x,y,comp,req_comp);
}
// *************************************************************************************************
// GIF loader -- public domain by Jean-Marc Lienher -- simplified/shrunk by stb
typedef struct stbi_gif_lzw_struct {
int16 prefix;
uint8 first;
uint8 suffix;
} stbi_gif_lzw;
typedef struct stbi_gif_struct
{
int w,h;
stbi_uc *out; // output buffer (always 4 components)
int flags, bgindex, ratio, transparent, eflags;
uint8 pal[256][4];
uint8 lpal[256][4];
stbi_gif_lzw codes[4096];
uint8 *color_table;
int parse, step;
int lflags;
int start_x, start_y;
int max_x, max_y;
int cur_x, cur_y;
int line_size;
} stbi_gif;
static int gif_test(stbi *s)
{
int sz;
if (get8(s) != 'G' || get8(s) != 'I' || get8(s) != 'F' || get8(s) != '8') return 0;
sz = get8(s);
if (sz != '9' && sz != '7') return 0;
if (get8(s) != 'a') return 0;
return 1;
}
static int stbi_gif_test(stbi *s)
{
int r = gif_test(s);
stbi_rewind(s);
return r;
}
static void stbi_gif_parse_colortable(stbi *s, uint8 pal[256][4], int num_entries, int transp)
{
int i;
for (i=0; i < num_entries; ++i) {
pal[i][2] = get8u(s);
pal[i][1] = get8u(s);
pal[i][0] = get8u(s);
pal[i][3] = transp ? 0 : 255;
}
}
static int stbi_gif_header(stbi *s, stbi_gif *g, int *comp, int is_info)
{
uint8 version;
if (get8(s) != 'G' || get8(s) != 'I' || get8(s) != 'F' || get8(s) != '8')
return e("not GIF", "Corrupt GIF");
version = get8u(s);
if (version != '7' && version != '9') return e("not GIF", "Corrupt GIF");
if (get8(s) != 'a') return e("not GIF", "Corrupt GIF");
failure_reason = "";
g->w = get16le(s);
g->h = get16le(s);
g->flags = get8(s);
g->bgindex = get8(s);
g->ratio = get8(s);
g->transparent = -1;
if (comp != 0) *comp = 4; // can't actually tell whether it's 3 or 4 until we parse the comments
if (is_info) return 1;
if (g->flags & 0x80)
stbi_gif_parse_colortable(s,g->pal, 2 << (g->flags & 7), -1);
return 1;
}
static int stbi_gif_info_raw(stbi *s, int *x, int *y, int *comp)
{
stbi_gif g;
if (!stbi_gif_header(s, &g, comp, 1)) {
stbi_rewind( s );
return 0;
}
if (x) *x = g.w;
if (y) *y = g.h;
return 1;
}
static void stbi_out_gif_code(stbi_gif *g, uint16 code)
{
uint8 *p, *c;
// recurse to decode the prefixes, since the linked-list is backwards,
// and working backwards through an interleaved image would be nasty
if (g->codes[code].prefix >= 0)
stbi_out_gif_code(g, g->codes[code].prefix);
if (g->cur_y >= g->max_y) return;
p = &g->out[g->cur_x + g->cur_y];
c = &g->color_table[g->codes[code].suffix * 4];
if (c[3] >= 128) {
p[0] = c[2];
p[1] = c[1];
p[2] = c[0];
p[3] = c[3];
}
g->cur_x += 4;
if (g->cur_x >= g->max_x) {
g->cur_x = g->start_x;
g->cur_y += g->step;
while (g->cur_y >= g->max_y && g->parse > 0) {
g->step = (1 << g->parse) * g->line_size;
g->cur_y = g->start_y + (g->step >> 1);
--g->parse;
}
}
}
static uint8 *stbi_process_gif_raster(stbi *s, stbi_gif *g)
{
uint8 lzw_cs;
int32 len, code;
uint32 first;
int32 codesize, codemask, avail, oldcode, bits, valid_bits, clear;
stbi_gif_lzw *p;
lzw_cs = get8u(s);
clear = 1 << lzw_cs;
first = 1;
codesize = lzw_cs + 1;
codemask = (1 << codesize) - 1;
bits = 0;
valid_bits = 0;
for (code = 0; code < clear; code++) {
g->codes[code].prefix = -1;
g->codes[code].first = (uint8) code;
g->codes[code].suffix = (uint8) code;
}
// support no starting clear code
avail = clear+2;
oldcode = -1;
len = 0;
for(;;) {
if (valid_bits < codesize) {
if (len == 0) {
len = get8(s); // start new block
if (len == 0)
return g->out;
}
--len;
bits |= (int32) get8(s) << valid_bits;
valid_bits += 8;
} else {
int32 code = bits & codemask;
bits >>= codesize;
valid_bits -= codesize;
// @OPTIMIZE: is there some way we can accelerate the non-clear path?
if (code == clear) { // clear code
codesize = lzw_cs + 1;
codemask = (1 << codesize) - 1;
avail = clear + 2;
oldcode = -1;
first = 0;
} else if (code == clear + 1) { // end of stream code
skip(s, len);
while ((len = get8(s)) > 0)
skip(s,len);
return g->out;
} else if (code <= avail) {
if (first) return epuc("no clear code", "Corrupt GIF");
if (oldcode >= 0) {
p = &g->codes[avail++];
if (avail > 4096) return epuc("too many codes", "Corrupt GIF");
p->prefix = (int16) oldcode;
p->first = g->codes[oldcode].first;
p->suffix = (code == avail) ? p->first : g->codes[code].first;
} else if (code == avail)
return epuc("illegal code in raster", "Corrupt GIF");
stbi_out_gif_code(g, (uint16) code);
if ((avail & codemask) == 0 && avail <= 0x0FFF) {
codesize++;
codemask = (1 << codesize) - 1;
}
oldcode = code;
} else {
return epuc("illegal code in raster", "Corrupt GIF");
}
}
}
}
static void stbi_fill_gif_background(stbi_gif *g)
{
int i;
uint8 *c = g->pal[g->bgindex];
// @OPTIMIZE: write a dword at a time
for (i = 0; i < g->w * g->h * 4; i += 4) {
uint8 *p = &g->out[i];
p[0] = c[2];
p[1] = c[1];
p[2] = c[0];
p[3] = c[3];
}
}
// this function is designed to support animated gifs, although stb_image doesn't support it
static uint8 *stbi_gif_load_next(stbi *s, stbi_gif *g, int *comp, int req_comp)
{
int i;
uint8 *old_out = 0;
if (g->out == 0) {
if (!stbi_gif_header(s, g, comp,0)) return 0; // failure_reason set by stbi_gif_header
g->out = (uint8 *) malloc(4 * g->w * g->h);
if (g->out == 0) return epuc("outofmem", "Out of memory");
stbi_fill_gif_background(g);
} else {
// animated-gif-only path
if (((g->eflags & 0x1C) >> 2) == 3) {
old_out = g->out;
g->out = (uint8 *) malloc(4 * g->w * g->h);
if (g->out == 0) return epuc("outofmem", "Out of memory");
memcpy(g->out, old_out, g->w*g->h*4);
}
}
for (;;) {
switch (get8(s)) {
case 0x2C: /* Image Descriptor */
{
int32 x, y, w, h;
uint8 *o;
x = get16le(s);
y = get16le(s);
w = get16le(s);
h = get16le(s);
if (((x + w) > (g->w)) || ((y + h) > (g->h)))
return epuc("bad Image Descriptor", "Corrupt GIF");
g->line_size = g->w * 4;
g->start_x = x * 4;
g->start_y = y * g->line_size;
g->max_x = g->start_x + w * 4;
g->max_y = g->start_y + h * g->line_size;
g->cur_x = g->start_x;
g->cur_y = g->start_y;
g->lflags = get8(s);
if (g->lflags & 0x40) {
g->step = 8 * g->line_size; // first interlaced spacing
g->parse = 3;
} else {
g->step = g->line_size;
g->parse = 0;
}
if (g->lflags & 0x80) {
stbi_gif_parse_colortable(s,g->lpal, 2 << (g->lflags & 7), g->eflags & 0x01 ? g->transparent : -1);
g->color_table = (uint8 *) g->lpal;
} else if (g->flags & 0x80) {
for (i=0; i < 256; ++i) // @OPTIMIZE: reset only the previous transparent
g->pal[i][3] = 255;
if (g->transparent >= 0 && (g->eflags & 0x01))
g->pal[g->transparent][3] = 0;
g->color_table = (uint8 *) g->pal;
} else
return epuc("missing color table", "Corrupt GIF");
o = stbi_process_gif_raster(s, g);
if (o == NULL) return NULL;
if (req_comp && req_comp != 4)
o = convert_format(o, 4, req_comp, g->w, g->h);
return o;
}
case 0x21: // Comment Extension.
{
int len;
if (get8(s) == 0xF9) { // Graphic Control Extension.
len = get8(s);
if (len == 4) {
g->eflags = get8(s);
get16le(s); // delay
g->transparent = get8(s);
} else {
skip(s, len);
break;
}
}
while ((len = get8(s)) != 0)
skip(s, len);
break;
}
case 0x3B: // gif stream termination code
return (uint8 *) 1;
default:
return epuc("unknown code", "Corrupt GIF");
}
}
}
static stbi_uc *stbi_gif_load(stbi *s, int *x, int *y, int *comp, int req_comp)
{
uint8 *u = 0;
stbi_gif g={0};
u = stbi_gif_load_next(s, &g, comp, req_comp);
if (u == (void *) 1) u = 0; // end of animated gif marker
if (u) {
*x = g.w;
*y = g.h;
}
return u;
}
static int stbi_gif_info(stbi *s, int *x, int *y, int *comp)
{
return stbi_gif_info_raw(s,x,y,comp);
}
// *************************************************************************************************
// Radiance RGBE HDR loader
// originally by Nicolas Schulz
#ifndef STBI_NO_HDR
static int hdr_test(stbi *s)
{
const char *signature = "#?RADIANCE\n";
int i;
for (i=0; signature[i]; ++i)
if (get8(s) != signature[i])
return 0;
return 1;
}
static int stbi_hdr_test(stbi* s)
{
int r = hdr_test(s);
stbi_rewind(s);
return r;
}
#define HDR_BUFLEN 1024
static char *hdr_gettoken(stbi *z, char *buffer)
{
int len=0;
char c = '\0';
c = (char) get8(z);
while (!at_eof(z) && c != '\n') {
buffer[len++] = c;
if (len == HDR_BUFLEN-1) {
// flush to end of line
while (!at_eof(z) && get8(z) != '\n')
;
break;
}
c = (char) get8(z);
}
buffer[len] = 0;
return buffer;
}
static void hdr_convert(float *output, stbi_uc *input, int req_comp)
{
if ( input[3] != 0 ) {
float f1;
// Exponent
f1 = (float) ldexp(1.0f, input[3] - (int)(128 + 8));
if (req_comp <= 2)
output[0] = (input[0] + input[1] + input[2]) * f1 / 3;
else {
output[0] = input[0] * f1;
output[1] = input[1] * f1;
output[2] = input[2] * f1;
}
if (req_comp == 2) output[1] = 1;
if (req_comp == 4) output[3] = 1;
} else {
switch (req_comp) {
case 4: output[3] = 1; /* fallthrough */
case 3: output[0] = output[1] = output[2] = 0;
break;
case 2: output[1] = 1; /* fallthrough */
case 1: output[0] = 0;
break;
}
}
}
static float *hdr_load(stbi *s, int *x, int *y, int *comp, int req_comp)
{
char buffer[HDR_BUFLEN];
char *token;
int valid = 0;
int width, height;
stbi_uc *scanline;
float *hdr_data;
int len;
unsigned char count, value;
int i, j, k, c1,c2, z;
// Check identifier
if (strcmp(hdr_gettoken(s,buffer), "#?RADIANCE") != 0)
return epf("not HDR", "Corrupt HDR image");
// Parse header
for(;;) {
token = hdr_gettoken(s,buffer);
if (token[0] == 0) break;
if (strcmp(token, "FORMAT=32-bit_rle_rgbe") == 0) valid = 1;
}
if (!valid) return epf("unsupported format", "Unsupported HDR format");
// Parse width and height
// can't use sscanf() if we're not using stdio!
token = hdr_gettoken(s,buffer);
if (strncmp(token, "-Y ", 3)) return epf("unsupported data layout", "Unsupported HDR format");
token += 3;
height = strtol(token, &token, 10);
while (*token == ' ') ++token;
if (strncmp(token, "+X ", 3)) return epf("unsupported data layout", "Unsupported HDR format");
token += 3;
width = strtol(token, NULL, 10);
*x = width;
*y = height;
*comp = 3;
if (req_comp == 0) req_comp = 3;
// Read data
hdr_data = (float *) malloc(height * width * req_comp * sizeof(float));
// Load image data
// image data is stored as some number of sca
if ( width < 8 || width >= 32768) {
// Read flat data
for (j=0; j < height; ++j) {
for (i=0; i < width; ++i) {
stbi_uc rgbe[4];
main_decode_loop:
getn(s, rgbe, 4);
hdr_convert(hdr_data + j * width * req_comp + i * req_comp, rgbe, req_comp);
}
}
} else {
// Read RLE-encoded data
scanline = NULL;
for (j = 0; j < height; ++j) {
c1 = get8(s);
c2 = get8(s);
len = get8(s);
if (c1 != 2 || c2 != 2 || (len & 0x80)) {
// not run-length encoded, so we have to actually use THIS data as a decoded
// pixel (note this can't be a valid pixel--one of RGB must be >= 128)
uint8 rgbe[4];
rgbe[0] = (uint8) c1;
rgbe[1] = (uint8) c2;
rgbe[2] = (uint8) len;
rgbe[3] = (uint8) get8u(s);
hdr_convert(hdr_data, rgbe, req_comp);
i = 1;
j = 0;
free(scanline);
goto main_decode_loop; // yes, this makes no sense
}
len <<= 8;
len |= get8(s);
if (len != width) { free(hdr_data); free(scanline); return epf("invalid decoded scanline length", "corrupt HDR"); }
if (scanline == NULL) scanline = (stbi_uc *) malloc(width * 4);
for (k = 0; k < 4; ++k) {
i = 0;
while (i < width) {
count = get8u(s);
if (count > 128) {
// Run
value = get8u(s);
count -= 128;
for (z = 0; z < count; ++z)
scanline[i++ * 4 + k] = value;
} else {
// Dump
for (z = 0; z < count; ++z)
scanline[i++ * 4 + k] = get8u(s);
}
}
}
for (i=0; i < width; ++i)
hdr_convert(hdr_data+(j*width + i)*req_comp, scanline + i*4, req_comp);
}
free(scanline);
}
return hdr_data;
}
static float *stbi_hdr_load(stbi *s, int *x, int *y, int *comp, int req_comp)
{
return hdr_load(s,x,y,comp,req_comp);
}
static int stbi_hdr_info(stbi *s, int *x, int *y, int *comp)
{
char buffer[HDR_BUFLEN];
char *token;
int valid = 0;
if (strcmp(hdr_gettoken(s,buffer), "#?RADIANCE") != 0) {
stbi_rewind( s );
return 0;
}
for(;;) {
token = hdr_gettoken(s,buffer);
if (token[0] == 0) break;
if (strcmp(token, "FORMAT=32-bit_rle_rgbe") == 0) valid = 1;
}
if (!valid) {
stbi_rewind( s );
return 0;
}
token = hdr_gettoken(s,buffer);
if (strncmp(token, "-Y ", 3)) {
stbi_rewind( s );
return 0;
}
token += 3;
*y = strtol(token, &token, 10);
while (*token == ' ') ++token;
if (strncmp(token, "+X ", 3)) {
stbi_rewind( s );
return 0;
}
token += 3;
*x = strtol(token, NULL, 10);
*comp = 3;
return 1;
}
#endif // STBI_NO_HDR
static int stbi_bmp_info(stbi *s, int *x, int *y, int *comp)
{
int hsz;
if (get8(s) != 'B' || get8(s) != 'M') {
stbi_rewind( s );
return 0;
}
skip(s,12);
hsz = get32le(s);
if (hsz != 12 && hsz != 40 && hsz != 56 && hsz != 108) {
stbi_rewind( s );
return 0;
}
if (hsz == 12) {
*x = get16le(s);
*y = get16le(s);
} else {
*x = get32le(s);
*y = get32le(s);
}
if (get16le(s) != 1) {
stbi_rewind( s );
return 0;
}
*comp = get16le(s) / 8;
return 1;
}
static int stbi_psd_info(stbi *s, int *x, int *y, int *comp)
{
int channelCount;
if (get32(s) != 0x38425053) {
stbi_rewind( s );
return 0;
}
if (get16(s) != 1) {
stbi_rewind( s );
return 0;
}
skip(s, 6);
channelCount = get16(s);
if (channelCount < 0 || channelCount > 16) {
stbi_rewind( s );
return 0;
}
*y = get32(s);
*x = get32(s);
if (get16(s) != 8) {
stbi_rewind( s );
return 0;
}
if (get16(s) != 3) {
stbi_rewind( s );
return 0;
}
*comp = 4;
return 1;
}
static int stbi_pic_info(stbi *s, int *x, int *y, int *comp)
{
int act_comp=0,num_packets=0,chained;
pic_packet_t packets[10];
skip(s, 92);
*x = get16(s);
*y = get16(s);
if (at_eof(s)) return 0;
if ( (*x) != 0 && (1 << 28) / (*x) < (*y)) {
stbi_rewind( s );
return 0;
}
skip(s, 8);
do {
pic_packet_t *packet;
if (num_packets==sizeof(packets)/sizeof(packets[0]))
return 0;
packet = &packets[num_packets++];
chained = get8(s);
packet->size = get8u(s);
packet->type = get8u(s);
packet->channel = get8u(s);
act_comp |= packet->channel;
if (at_eof(s)) {
stbi_rewind( s );
return 0;
}
if (packet->size != 8) {
stbi_rewind( s );
return 0;
}
} while (chained);
*comp = (act_comp & 0x10 ? 4 : 3);
return 1;
}
static int stbi_info_main(stbi *s, int *x, int *y, int *comp)
{
if (stbi_jpeg_info(s, x, y, comp))
return 1;
if (stbi_png_info(s, x, y, comp))
return 1;
if (stbi_gif_info(s, x, y, comp))
return 1;
if (stbi_bmp_info(s, x, y, comp))
return 1;
if (stbi_psd_info(s, x, y, comp))
return 1;
if (stbi_pic_info(s, x, y, comp))
return 1;
#ifndef STBI_NO_HDR
if (stbi_hdr_info(s, x, y, comp))
return 1;
#endif
// test tga last because it's a crappy test!
if (stbi_tga_info(s, x, y, comp))
return 1;
return e("unknown image type", "Image not of any known type, or corrupt");
}
#ifndef STBI_NO_STDIO
int stbi_info(char const *filename, int *x, int *y, int *comp)
{
FILE *f = fopen(filename, "rb");
int result;
if (!f) return e("can't fopen", "Unable to open file");
result = stbi_info_from_file(f, x, y, comp);
fclose(f);
return result;
}
int stbi_info_from_file(FILE *f, int *x, int *y, int *comp)
{
int r;
stbi s;
long pos = ftell(f);
start_file(&s, f);
r = stbi_info_main(&s,x,y,comp);
fseek(f,pos,SEEK_SET);
return r;
}
#endif // !STBI_NO_STDIO
int stbi_info_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp)
{
stbi s;
start_mem(&s,buffer,len);
return stbi_info_main(&s,x,y,comp);
}
int stbi_info_from_callbacks(stbi_io_callbacks const *c, void *user, int *x, int *y, int *comp)
{
stbi s;
start_callbacks(&s, (stbi_io_callbacks *) c, user);
return stbi_info_main(&s,x,y,comp);
}
#endif // STBI_HEADER_FILE_ONLY
/*
revision history:
1.33 (2011-07-14)
make stbi_is_hdr work in STBI_NO_HDR (as specified), minor compiler-friendly improvements
1.32 (2011-07-13)
support for "info" function for all supported filetypes (SpartanJ)
1.31 (2011-06-20)
a few more leak fixes, bug in PNG handling (SpartanJ)
1.30 (2011-06-11)
added ability to load files via callbacks to accomidate custom input streams (Ben Wenger)
removed deprecated format-specific test/load functions
removed support for installable file formats (stbi_loader) -- would have been broken for IO callbacks anyway
error cases in bmp and tga give messages and don't leak (Raymond Barbiero, grisha)
fix inefficiency in decoding 32-bit BMP (David Woo)
1.29 (2010-08-16)
various warning fixes from Aurelien Pocheville
1.28 (2010-08-01)
fix bug in GIF palette transparency (SpartanJ)
1.27 (2010-08-01)
cast-to-uint8 to fix warnings
1.26 (2010-07-24)
fix bug in file buffering for PNG reported by SpartanJ
1.25 (2010-07-17)
refix trans_data warning (Won Chun)
1.24 (2010-07-12)
perf improvements reading from files on platforms with lock-heavy fgetc()
minor perf improvements for jpeg
deprecated type-specific functions so we'll get feedback if they're needed
attempt to fix trans_data warning (Won Chun)
1.23 fixed bug in iPhone support
1.22 (2010-07-10)
removed image *writing* support
stbi_info support from Jetro Lauha
GIF support from Jean-Marc Lienher
iPhone PNG-extensions from James Brown
warning-fixes from Nicolas Schulz and Janez Zemva (i.e. Janez (U+017D)emva)
1.21 fix use of 'uint8' in header (reported by jon blow)
1.20 added support for Softimage PIC, by Tom Seddon
1.19 bug in interlaced PNG corruption check (found by ryg)
1.18 2008-08-02
fix a threading bug (local mutable static)
1.17 support interlaced PNG
1.16 major bugfix - convert_format converted one too many pixels
1.15 initialize some fields for thread safety
1.14 fix threadsafe conversion bug
header-file-only version (#define STBI_HEADER_FILE_ONLY before including)
1.13 threadsafe
1.12 const qualifiers in the API
1.11 Support installable IDCT, colorspace conversion routines
1.10 Fixes for 64-bit (don't use "unsigned long")
optimized upsampling by Fabian "ryg" Giesen
1.09 Fix format-conversion for PSD code (bad global variables!)
1.08 Thatcher Ulrich's PSD code integrated by Nicolas Schulz
1.07 attempt to fix C++ warning/errors again
1.06 attempt to fix C++ warning/errors again
1.05 fix TGA loading to return correct *comp and use good luminance calc
1.04 default float alpha is 1, not 255; use 'void *' for stbi_image_free
1.03 bugfixes to STBI_NO_STDIO, STBI_NO_HDR
1.02 support for (subset of) HDR files, float interface for preferred access to them
1.01 fix bug: possible bug in handling right-side up bmps... not sure
fix bug: the stbi_bmp_load() and stbi_tga_load() functions didn't work at all
1.00 interface to zlib that skips zlib header
0.99 correct handling of alpha in palette
0.98 TGA loader by lonesock; dynamically add loaders (untested)
0.97 jpeg errors on too large a file; also catch another malloc failure
0.96 fix detection of invalid v value - particleman@mollyrocket forum
0.95 during header scan, seek to markers in case of padding
0.94 STBI_NO_STDIO to disable stdio usage; rename all #defines the same
0.93 handle jpegtran output; verbose errors
0.92 read 4,8,16,24,32-bit BMP files of several formats
0.91 output 24-bit Windows 3.0 BMP files
0.90 fix a few more warnings; bump version number to approach 1.0
0.61 bugfixes due to Marc LeBlanc, Christopher Lloyd
0.60 fix compiling as c++
0.59 fix warnings: merge Dave Moore's -Wall fixes
0.58 fix bug: zlib uncompressed mode len/nlen was wrong endian
0.57 fix bug: jpg last huffman symbol before marker was >9 bits but less than 16 available
0.56 fix bug: zlib uncompressed mode len vs. nlen
0.55 fix bug: restart_interval not initialized to 0
0.54 allow NULL for 'int *comp'
0.53 fix bug in png 3->4; speedup png decoding
0.52 png handles req_comp=3,4 directly; minor cleanup; jpeg comments
0.51 obey req_comp requests, 1-component jpegs return as 1-component,
on 'test' only check type, not whether we support this variant
0.50 first released version
*/
================================================
FILE: example/stb_image_write.h
================================================
/* stbiw-0.92 - public domain - http://nothings.org/stb/stb_image_write.h
writes out PNG/BMP/TGA images to C stdio - Sean Barrett 2010
no warranty implied; use at your own risk
Before including,
#define STB_IMAGE_WRITE_IMPLEMENTATION
in the file that you want to have the implementation.
ABOUT:
This header file is a library for writing images to C stdio. It could be
adapted to write to memory or a general streaming interface; let me know.
The PNG output is not optimal; it is 20-50% larger than the file
written by a decent optimizing implementation. This library is designed
for source code compactness and simplicitly, not optimal image file size
or run-time performance.
USAGE:
There are three functions, one for each image file format:
int stbi_write_png(char const *filename, int w, int h, int comp, const void *data, int stride_in_bytes);
int stbi_write_bmp(char const *filename, int w, int h, int comp, const void *data);
int stbi_write_tga(char const *filename, int w, int h, int comp, const void *data);
Each function returns 0 on failure and non-0 on success.
The functions create an image file defined by the parameters. The image
is a rectangle of pixels stored from left-to-right, top-to-bottom.
Each pixel contains 'comp' channels of data stored interleaved with 8-bits
per channel, in the following order: 1=Y, 2=YA, 3=RGB, 4=RGBA. (Y is
monochrome color.) The rectangle is 'w' pixels wide and 'h' pixels tall.
The *data pointer points to the first byte of the top-left-most pixel.
For PNG, "stride_in_bytes" is the distance in bytes from the first byte of
a row of pixels to the first byte of the next row of pixels.
PNG creates output files with the same number of components as the input.
The BMP and TGA formats expand Y to RGB in the file format. BMP does not
output alpha.
PNG supports writing rectangles of data even when the bytes storing rows of
data are not consecutive in memory (e.g. sub-rectangles of a larger image),
by supplying the stride between the beginning of adjacent rows. The other
formats do not. (Thus you cannot write a native-format BMP through the BMP
writer, both because it is in BGR order and because it may have padding
at the end of the line.)
*/
#ifndef INCLUDE_STB_IMAGE_WRITE_H
#define INCLUDE_STB_IMAGE_WRITE_H
#ifdef __cplusplus
extern "C" {
#endif
extern int stbi_write_png(char const *filename, int w, int h, int comp, const void *data, int stride_in_bytes);
extern int stbi_write_bmp(char const *filename, int w, int h, int comp, const void *data);
extern int stbi_write_tga(char const *filename, int w, int h, int comp, const void *data);
#ifdef __cplusplus
}
#endif
#endif//INCLUDE_STB_IMAGE_WRITE_H
#ifdef STB_IMAGE_WRITE_IMPLEMENTATION
#include <stdarg.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <assert.h>
typedef unsigned int stbiw_uint32;
typedef int stb_image_write_test[sizeof(stbiw_uint32)==4 ? 1 : -1];
static void writefv(FILE *f, const char *fmt, va_list v)
{
while (*fmt) {
switch (*fmt++) {
case ' ': break;
case '1': { unsigned char x = (unsigned char) va_arg(v, int); fputc(x,f); break; }
case '2': { int x = va_arg(v,int); unsigned char b[2];
b[0] = (unsigned char) x; b[1] = (unsigned char) (x>>8);
fwrite(b,2,1,f); break; }
case '4': { stbiw_uint32 x = va_arg(v,int); unsigned char b[4];
b[0]=(unsigned char)x; b[1]=(unsigned char)(x>>8);
b[2]=(unsigned char)(x>>16); b[3]=(unsigned char)(x>>24);
fwrite(b,4,1,f); break; }
default:
assert(0);
return;
}
}
}
static void write3(FILE *f, unsigned char a, unsigned char b, unsigned char c)
{
unsigned char arr[3];
arr[0] = a, arr[1] = b, arr[2] = c;
fwrite(arr, 3, 1, f);
}
static void write_pixels(FILE *f, int rgb_dir, int vdir, int x, int y, int comp, void *data, int write_alpha, int scanline_pad)
{
unsigned char bg[3] = { 255, 0, 255}, px[3];
stbiw_uint32 zero = 0;
int i,j,k, j_end;
if (y <= 0)
return;
if (vdir < 0)
j_end = -1, j = y-1;
else
j_end = y, j = 0;
for (; j != j_end; j += vdir) {
for (i=0; i < x; ++i) {
unsigned char *d = (unsigned char *) data + (j*x+i)*comp;
if (write_alpha < 0)
fwrite(&d[comp-1], 1, 1, f);
switch (comp) {
case 1:
case 2: write3(f, d[0],d[0],d[0]);
break;
case 4:
if (!write_alpha) {
// composite against pink background
for (k=0; k < 3; ++k)
px[k] = bg[k] + ((d[k] - bg[k]) * d[3])/255;
write3(f, px[1-rgb_dir],px[1],px[1+rgb_dir]);
break;
}
/* FALLTHROUGH */
case 3:
write3(f, d[1-rgb_dir],d[1],d[1+rgb_dir]);
break;
}
if (write_alpha > 0)
fwrite(&d[comp-1], 1, 1, f);
}
fwrite(&zero,scanline_pad,1,f);
}
}
static int outfile(char const *filename, int rgb_dir, int vdir, int x, int y, int comp, void *data, int alpha, int pad, const char *fmt, ...)
{
FILE *f;
if (y < 0 || x < 0) return 0;
f = fopen(filename, "wb");
if (f) {
va_list v;
va_start(v, fmt);
writefv(f, fmt, v);
va_end(v);
write_pixels(f,rgb_dir,vdir,x,y,comp,data,alpha,pad);
fclose(f);
}
return f != NULL;
}
int stbi_write_bmp(char const *filename, int x, int y, int comp, const void *data)
{
int pad = (-x*3) & 3;
return outfile(filename,-1,-1,x,y,comp,(void *) data,0,pad,
"11 4 22 4" "4 44 22 444444",
'B', 'M', 14+40+(x*3+pad)*y, 0,0, 14+40, // file header
40, x,y, 1,24, 0,0,0,0,0,0); // bitmap header
}
int stbi_write_tga(char const *filename, int x, int y, int comp, const void *data)
{
int has_alpha = !(comp & 1);
return outfile(filename, -1,-1, x, y, comp, (void *) data, has_alpha, 0,
"111 221 2222 11", 0,0,2, 0,0,0, 0,0,x,y, 24+8*has_alpha, 8*has_alpha);
}
// stretchy buffer; stbi__sbpush() == vector<>::push_back() -- stbi__sbcount() == vector<>::size()
#define stbi__sbraw(a) ((int *) (a) - 2)
#define stbi__sbm(a) stbi__sbraw(a)[0]
#define stbi__sbn(a) stbi__sbraw(a)[1]
#define stbi__sbneedgrow(a,n) ((a)==0 || stbi__sbn(a)+n >= stbi__sbm(a))
#define stbi__sbmaybegrow(a,n) (stbi__sbneedgrow(a,(n)) ? stbi__sbgrow(a,n) : 0)
#define stbi__sbgrow(a,n) stbi__sbgrowf((void **) &(a), (n), sizeof(*(a)))
#define stbi__sbpush(a, v) (stbi__sbmaybegrow(a,1), (a)[stbi__sbn(a)++] = (v))
#define stbi__sbcount(a) ((a) ? stbi__sbn(a) : 0)
#define stbi__sbfree(a) ((a) ? free(stbi__sbraw(a)),0 : 0)
static void *stbi__sbgrowf(void **arr, int increment, int itemsize)
{
int m = *arr ? 2*stbi__sbm(*arr)+increment : increment+1;
void *p = realloc(*arr ? stbi__sbraw(*arr) : 0, itemsize * m + sizeof(int)*2);
assert(p);
if (p) {
if (!*arr) ((int *) p)[1] = 0;
*arr = (void *) ((int *) p + 2);
stbi__sbm(*arr) = m;
}
return *arr;
}
static unsigned char *stbi__zlib_flushf(unsigned char *data, unsigned int *bitbuffer, int *bitcount)
{
while (*bitcount >= 8) {
stbi__sbpush(data, (unsigned char) *bitbuffer);
*bitbuffer >>= 8;
*bitcount -= 8;
}
return data;
}
static int stbi__zlib_bitrev(int code, int codebits)
{
int res=0;
while (codebits--) {
res = (res << 1) | (code & 1);
code >>= 1;
}
return res;
}
static unsigned int stbi__zlib_countm(unsigned char *a, unsigned char *b, int limit)
{
int i;
for (i=0; i < limit && i < 258; ++i)
if (a[i] != b[i]) break;
return i;
}
static unsigned int stbi__zhash(unsigned char *data)
{
stbiw_uint32 hash = data[0] + (data[1] << 8) + (data[2] << 16);
hash ^= hash << 3;
hash += hash >> 5;
hash ^= hash << 4;
hash += hash >> 17;
hash ^= hash << 25;
hash += hash >> 6;
return hash;
}
#define stbi__zlib_flush() (out = stbi__zlib_flushf(out, &bitbuf, &bitcount))
#define stbi__zlib_add(code,codebits) \
(bitbuf |= (code) << bitcount, bitcount += (codebits), stbi__zlib_flush())
#define stbi__zlib_huffa(b,c) stbi__zlib_add(stbi__zlib_bitrev(b,c),c)
// default huffman tables
#define stbi__zlib_huff1(n) stbi__zlib_huffa(0x30 + (n), 8)
#define stbi__zlib_huff2(n) stbi__zlib_huffa(0x190 + (n)-144, 9)
#define stbi__zlib_huff3(n) stbi__zlib_huffa(0 + (n)-256,7)
#define stbi__zlib_huff4(n) stbi__zlib_huffa(0xc0 + (n)-280,8)
#define stbi__zlib_huff(n) ((n) <= 143 ? stbi__zlib_huff1(n) : (n) <= 255 ? stbi__zlib_huff2(n) : (n) <= 279 ? stbi__zlib_huff3(n) : stbi__zlib_huff4(n))
#define stbi__zlib_huffb(n) ((n) <= 143 ? stbi__zlib_huff1(n) : stbi__zlib_huff2(n))
#define stbi__ZHASH 16384
unsigned char * stbi_zlib_compress(unsigned char *data, int data_len, int *out_len, int quality)
{
static unsigned short lengthc[] = { 3,4,5,6,7,8,9,10,11,13,15,17,19,23,27,31,35,43,51,59,67,83,99,115,131,163,195,227,258, 259 };
static unsigned char lengtheb[]= { 0,0,0,0,0,0,0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 0 };
static unsigned short distc[] = { 1,2,3,4,5,7,9,13,17,25,33,49,65,97,129,193,257,385,513,769,1025,1537,2049,3073,4097,6145,8193,12289,16385,24577, 32768 };
static unsigned char disteb[] = { 0,0,0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,11,11,12,12,13,13 };
unsigned int bitbuf=0;
int i,j, bitcount=0;
unsigned char *out = NULL;
unsigned char **hash_table[stbi__ZHASH]; // 64KB on the stack!
if (quality < 5) quality = 5;
stbi__sbpush(out, 0x78); // DEFLATE 32K window
stbi__sbpush(out, 0x5e); // FLEVEL = 1
stbi__zlib_add(1,1); // BFINAL = 1
stbi__zlib_add(1,2); // BTYPE = 1 -- fixed huffman
for (i=0; i < stbi__ZHASH; ++i)
hash_table[i] = NULL;
i=0;
while (i < data_len-3) {
// hash next 3 bytes of data to be compressed
int h = stbi__zhash(data+i)&(stbi__ZHASH-1), best=3;
unsigned char *bestloc = 0;
unsigned char **hlist = hash_table[h];
int n = stbi__sbcount(hlist);
for (j=0; j < n; ++j) {
if (hlist[j]-data > i-32768) { // if entry lies within window
int d = stbi__zlib_countm(hlist[j], data+i, data_len-i);
if (d >= best) best=d,bestloc=hlist[j];
}
}
// when hash table entry is too long, delete half the entries
if (hash_table[h] && stbi__sbn(hash_table[h]) == 2*quality) {
memcpy(hash_table[h], hash_table[h]+quality, sizeof(hash_table[h][0])*quality);
stbi__sbn(hash_table[h]) = quality;
}
stbi__sbpush(hash_table[h],data+i);
if (bestloc) {
// "lazy matching" - check match at *next* byte, and if it's better, do cur byte as literal
h = stbi__zhash(data+i+1)&(stbi__ZHASH-1);
hlist = hash_table[h];
n = stbi__sbcount(hlist);
for (j=0; j < n; ++j) {
if (hlist[j]-data > i-32767) {
int e = stbi__zlib_countm(hlist[j], data+i+1, data_len-i-1);
if (e > best) { // if next match is better, bail on current match
bestloc = NULL;
break;
}
}
}
}
if (bestloc) {
int d = data+i - bestloc; // distance back
assert(d <= 32767 && best <= 258);
for (j=0; best > lengthc[j+1]-1; ++j);
stbi__zlib_huff(j+257);
if (lengtheb[j]) stbi__zlib_add(best - lengthc[j], lengtheb[j]);
for (j=0; d > distc[j+1]-1; ++j);
stbi__zlib_add(stbi__zlib_bitrev(j,5),5);
if (disteb[j]) stbi__zlib_add(d - distc[j], disteb[j]);
i += best;
} else {
stbi__zlib_huffb(data[i]);
++i;
}
}
// write out final bytes
for (;i < data_len; ++i)
stbi__zlib_huffb(data[i]);
stbi__zlib_huff(256); // end of block
// pad with 0 bits to byte boundary
while (bitcount)
stbi__zlib_add(0,1);
for (i=0; i < stbi__ZHASH; ++i)
(void) stbi__sbfree(hash_table[i]);
{
// compute adler32 on input
unsigned int i=0, s1=1, s2=0, blocklen = data_len % 5552;
int j=0;
while (j < data_len) {
for (i=0; i < blocklen; ++i) s1 += data[j+i], s2 += s1;
s1 %= 65521, s2 %= 65521;
j += blocklen;
blocklen = 5552;
}
stbi__sbpush(out, (unsigned char) (s2 >> 8));
stbi__sbpush(out, (unsigned char) s2);
stbi__sbpush(out, (unsigned char) (s1 >> 8));
stbi__sbpush(out, (unsigned char) s1);
}
*out_len = stbi__sbn(out);
// make returned pointer freeable
memmove(stbi__sbraw(out), out, *out_len);
return (unsigned char *) stbi__sbraw(out);
}
unsigned int stbi__crc32(unsigned char *buffer, int len)
{
static unsigned int crc_table[256];
unsigned int crc = ~0u;
int i,j;
if (crc_table[1] == 0)
for(i=0; i < 256; i++)
for (crc_table[i]=i, j=0; j < 8; ++j)
crc_table[i] = (crc_table[i] >> 1) ^ (crc_table[i] & 1 ? 0xedb88320 : 0);
for (i=0; i < len; ++i)
crc = (crc >> 8) ^ crc_table[buffer[i] ^ (crc & 0xff)];
return ~crc;
}
#define stbi__wpng4(o,a,b,c,d) ((o)[0]=(unsigned char)(a),(o)[1]=(unsigned char)(b),(o)[2]=(unsigned char)(c),(o)[3]=(unsigned char)(d),(o)+=4)
#define stbi__wp32(data,v) stbi__wpng4(data, (v)>>24,(v)>>16,(v)>>8,(v));
#define stbi__wptag(data,s) stbi__wpng4(data, s[0],s[1],s[2],s[3])
static void stbi__wpcrc(unsigned char **data, int len)
{
unsigned int crc = stbi__crc32(*data - len - 4, len+4);
stbi__wp32(*data, crc);
}
static unsigned char stbi__paeth(int a, int b, int c)
{
int p = a + b - c, pa = abs(p-a), pb = abs(p-b), pc = abs(p-c);
if (pa <= pb && pa <= pc) return (unsigned char) a;
if (pb <= pc) return (unsigned char) b;
return (unsigned char) c;
}
unsigned char *stbi_write_png_to_mem(unsigned char *pixels, int stride_bytes, int x, int y, int n, int *out_len)
{
int ctype[5] = { -1, 0, 4, 2, 6 };
unsigned char sig[8] = { 137,80,78,71,13,10,26,10 };
unsigned char *out,*o, *filt, *zlib;
signed char *line_buffer;
int i,j,k,p,zlen;
if (stride_bytes == 0)
stride_bytes = x * n;
filt = (unsigned char *) malloc((x*n+1) * y); if (!filt) return 0;
line_buffer = (signed char *) malloc(x * n); if (!line_buffer) { free(filt); return 0; }
for (j=0; j < y; ++j) {
static int mapping[] = { 0,1,2,3,4 };
static int firstmap[] = { 0,1,0,5,6 };
int *mymap = j ? mapping : firstmap;
int best = 0, bestval = 0x7fffffff;
for (p=0; p < 2; ++p) {
for (k= p?best:0; k < 5; ++k) {
int type = mymap[k],est=0;
unsigned char *z = pixels + stride_bytes*j;
for (i=0; i < n; ++i)
switch (type) {
case 0: line_buffer[i] = z[i]; break;
case 1: line_buffer[i] = z[i]; break;
case 2: line_buffer[i] = z[i] - z[i-stride_bytes]; break;
case 3: line_buffer[i] = z[i] - (z[i-stride_bytes]>>1); break;
case 4: line_buffer[i] = (signed char) (z[i] - stbi__paeth(0,z[i-stride_bytes],0)); break;
case 5: line_buffer[i] = z[i]; break;
case 6: line_buffer[i] = z[i]; break;
}
for (i=n; i < x*n; ++i) {
switch (type) {
case 0: line_buffer[i] = z[i]; break;
case 1: line_buffer[i] = z[i] - z[i-n]; break;
case 2: line_buffer[i] = z[i] - z[i-stride_bytes]; break;
case 3: line_buffer[i] = z[i] - ((z[i-n] + z[i-stride_bytes])>>1); break;
case 4: line_buffer[i] = z[i] - stbi__paeth(z[i-n], z[i-stride_bytes], z[i-stride_bytes-n]); break;
case 5: line_buffer[i] = z[i] - (z[i-n]>>1); break;
case 6: line_buffer[i] = z[i] - stbi__paeth(z[i-n], 0,0); break;
}
}
if (p) break;
for (i=0; i < x*n; ++i)
est += abs((signed char) line_buffer[i]);
if (est < bestval) { bestval = est; best = k; }
}
}
// when we get here, best contains the filter type, and line_buffer contains the data
filt[j*(x*n+1)] = (unsigned char) best;
memcpy(filt+j*(x*n+1)+1, line_buffer, x*n);
}
free(line_buffer);
zlib = stbi_zlib_compress(filt, y*( x*n+1), &zlen, 8); // increase 8 to get smaller but use more memory
free(filt);
if (!zlib) return 0;
// each tag requires 12 bytes of overhead
out = (unsigned char *) malloc(8 + 12+13 + 12+zlen + 12);
if (!out) return 0;
*out_len = 8 + 12+13 + 12+zlen + 12;
o=out;
memcpy(o,sig,8); o+= 8;
stbi__wp32(o, 13); // header length
stbi__wptag(o, "IHDR");
stbi__wp32(o, x);
stbi__wp32(o, y);
*o++ = 8;
*o++ = (unsigned char) ctype[n];
*o++ = 0;
*o++ = 0;
*o++ = 0;
stbi__wpcrc(&o,13);
stbi__wp32(o, zlen);
stbi__wptag(o, "IDAT");
memcpy(o, zlib, zlen); o += zlen; free(zlib);
stbi__wpcrc(&o, zlen);
stbi__wp32(o,0);
stbi__wptag(o, "IEND");
stbi__wpcrc(&o,0);
assert(o == out + *out_len);
return out;
}
int stbi_write_png(char const *filename, int x, int y, int comp, const void *data, int stride_bytes)
{
FILE *f;
int len;
unsigned char *png = stbi_write_png_to_mem((unsigned char *) data, stride_bytes, x, y, comp, &len);
if (!png) return 0;
f = fopen(filename, "wb");
if (!f) { free(png); return 0; }
fwrite(png, 1, len, f);
fclose(f);
free(png);
return 1;
}
#endif // STB_IMAGE_WRITE_IMPLEMENTATION
/* Revision history
0.92 (2010-08-01)
casts to unsigned char to fix warnings
0.91 (2010-07-17)
first public release
0.90 first internal release
*/
================================================
FILE: premake4.lua
================================================
local action = _ACTION or ""
solution "distancefield"
location ( "build" )
configurations { "Debug", "Release" }
platforms {"native", "x64", "x32"}
project "example"
kind "ConsoleApp"
language "C"
files { "example/example.c" }
includedirs { "src", "example" }
targetdir("build")
configuration { "linux" }
linkoptions { "`pkg-config --libs glfw3`" }
links { "GL", "GLU", "m" }
configuration { "windows" }
links { "glfw3", "gdi32", "winmm", "user32", "glu32","opengl32" }
configuration { "macosx" }
links { "glfw3" }
linkoptions { "-framework OpenGL", "-framework Cocoa", "-framework IOKit", "-framework CoreVideo" }
configuration "Debug"
defines { "DEBUG" }
flags { "Symbols", "ExtraWarnings"}
configuration "Release"
defines { "NDEBUG" }
flags { "Optimize", "ExtraWarnings"}
================================================
FILE: src/sdf.h
================================================
/*
Copyright (C) 2014 Mikko Mononen (memon@inside.org)
Copyright (C) 2009-2012 Stefan Gustavson (stefan.gustavson@gmail.com)
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE SOFTWARE.
*/
#ifndef SDF_H
#define SDF_H
// Sweep-and-update Euclidean distance transform of an antialised image for contour textures.
// Based on edtaa3func.c by Stefan Gustavson.
//
// White (255) pixels are treated as object pixels, zero pixels are treated as background.
// An attempt is made to treat antialiased edges correctly. The input image must have
// pixels in the range [0,255], and the antialiased image should be a box-filter
// sampling of the ideal, crisp edge. If the antialias region is more than 1 pixel wide,
// the result from this transform will be inaccurate.
// Pixels at image border are not calculated and are set to 0.
//
// The output distance field is encoded as bytes, where 0 = radius (outside) and 255 = -radius (inside).
// Input and output can be the same buffer.
// out - Output of the distance transform, one byte per pixel.
// outstride - Bytes per row on output image.
// radius - The radius of the distance field narrow band in pixels.
// img - Input image, one byte per pixel.
// width - Width if the image.
// height - Height if the image.
// stride - Bytes per row on input image.
int sdfBuildDistanceField(unsigned char* out, int outstride, float radius,
const unsigned char* img, int width, int height, int stride);
// Same as distXform, but does not allocate any memory.
// The 'temp' array should be enough to fit width * height * sizeof(float) * 3 bytes.
void sdfBuildDistanceFieldNoAlloc(unsigned char* out, int outstride, float radius,
const unsigned char* img, int width, int height, int stride,
unsigned char* temp);
// This function converts the antialiased image where each pixel represents coverage (box-filter
// sampling of the ideal, crisp edge) to a distance field with narrow band radius of sqrt(2).
// This is the fastest way to turn antialised image to contour texture. This function is good
// if you don't need the distance field for effects (i.e. fat outline or dropshadow).
// Input and output buffers must be different.
// out - Output of the distance transform, one byte per pixel.
// outstride - Bytes per row on output image.
// img - Input image, one byte per pixel.
// width - Width if the image.
// height - Height if the image.
// stride - Bytes per row on input image.
void sdfCoverageToDistanceField(unsigned char* out, int outstride,
const unsigned char* img, int width, int height, int stride);
#endif //SDF_H
#ifdef SDF_IMPLEMENTATION
#include <math.h>
#include <stdlib.h>
#define SDF_MAX_PASSES 10 // Maximum number of distance transform passes
#define SDF_SLACK 0.001f // Controls how much smaller the neighbour value must be to cosnider, too small slack increse iteration count.
#define SDF_SQRT2 1.4142136f // sqrt(2)
#define SDF_BIG 1e+37f // Big value used to initialize the distance field.
static float sdf__clamp01(float x)
{
return x < 0.0f ? 0.0f : (x > 1.0f ? 1.0f : x);
}
void sdfCoverageToDistanceField(unsigned char* out, int outstride,
const unsigned char* img, int width, int height, int stride)
{
int x, y;
// Zero out borders
for (x = 0; x < width; x++)
out[x] = 0;
for (y = 1; y < height; y++) {
out[y*outstride] = 0;
out[width-1+y*outstride] = 0;
}
for (x = 0; x < width; x++)
out[x+(height-1)*outstride] = 0;
for (y = 1; y < height-1; y++) {
for (x = 1; x < width-1; x++) {
int k = x + y * stride;
float d, gx, gy, glen, a, a1;
// Skip flat areas.
if (img[k] == 255) {
out[x+y*outstride] = 255;
continue;
}
if (img[k] == 0) {
// Special handling for cases where full opaque pixels are next to full transparent pixels.
// See: https://github.com/memononen/SDF/issues/2
int he = img[k-1] == 255 || img[k+1] == 255;
int ve = img[k-stride] == 255 || img[k+stride] == 255;
if (!he && !ve) {
out[x+y*outstride] = 0;
continue;
}
}
gx = -(float)img[k-stride-1] - SDF_SQRT2*(float)img[k-1] - (float)img[k+stride-1] + (float)img[k-stride+1] + SDF_SQRT2*(float)img[k+1] + (float)img[k+stride+1];
gy = -(float)img[k-stride-1] - SDF_SQRT2*(float)img[k-stride] - (float)img[k-stride+1] + (float)img[k+stride-1] + SDF_SQRT2*(float)img[k+stride] + (float)img[k+stride+1];
a = (float)img[k]/255.0f;
gx = fabsf(gx);
gy = fabsf(gy);
if (gx < 0.0001f || gy < 0.000f) {
d = (0.5f - a) * SDF_SQRT2;
} else {
glen = gx*gx + gy*gy;
glen = 1.0f / sqrtf(glen);
gx *= glen;
gy *= glen;
if (gx < gy) {
float temp = gx;
gx = gy;
gy = temp;
}
a1 = 0.5f*gy/gx;
if (a < a1) { // 0 <= a < a1
d = 0.5f*(gx + gy) - sqrtf(2.0f*gx*gy*a);
} else if (a < (1.0-a1)) { // a1 <= a <= 1-a1
d = (0.5f-a)*gx;
} else { // 1-a1 < a <= 1
d = -0.5f*(gx + gy) + sqrt(2.0f*gx*gy*(1.0f-a));
}
}
d *= 1.0f / SDF_SQRT2;
out[x+y*outstride] = (unsigned char)(sdf__clamp01(0.5f - d) * 255.0f);
}
}
}
static float sdf__edgedf(float gx, float gy, float a)
{
float df, a1;
if ((gx == 0) || (gy == 0)) {
// Either A) gu or gv are zero, or B) both
// Linear approximation is A) correct or B) a fair guess
df = 0.5f - a;
} else {
// Everything is symmetric wrt sign and transposition,
// so move to first octant (gx>=0, gy>=0, gx>=gy) to
// avoid handling all possible edge directions.
gx = fabsf(gx);
gy = fabsf(gy);
if (gx < gy) {
float temp = gx;
gx = gy;
gy = temp;
}
a1 = 0.5f*gy/gx;
if (a < a1) { // 0 <= a < a1
df = 0.5f*(gx + gy) - sqrtf(2.0f*gx*gy*a);
} else if (a < (1.0-a1)) { // a1 <= a <= 1-a1
df = (0.5f-a)*gx;
} else { // 1-a1 < a <= 1
df = -0.5f*(gx + gy) + sqrt(2.0f*gx*gy*(1.0f-a));
}
}
return df;
}
struct SDFpoint {
float x,y;
};
static float sdf__distsqr(struct SDFpoint* a, struct SDFpoint* b)
{
float dx = b->x - a->x, dy = b->y - a->y;
return dx*dx + dy*dy;
}
void sdfBuildDistanceFieldNoAlloc(unsigned char* out, int outstride, float radius,
const unsigned char* img, int width, int height, int stride,
unsigned char* temp)
{
int i, x, y, pass;
float scale;
float* tdist = (float*)&temp[0];
struct SDFpoint* tpt = (struct SDFpoint*)&temp[width * height * sizeof(float)];
// Initialize buffers
for (i = 0; i < width*height; i++) {
tpt[i].x = 0;
tpt[i].y = 0;
tdist[i] = SDF_BIG;
}
// Calculate position of the anti-aliased pixels and distance to the boundary of the shape.
for (y = 1; y < height-1; y++) {
for (x = 1; x < width-1; x++) {
int tk, k = x + y * stride;
struct SDFpoint c = { (float)x, (float)y };
float d, gx, gy, glen;
// Skip flat areas.
if (img[k] == 255) continue;
if (img[k] == 0) {
// Special handling for cases where full opaque pixels are next to full transparent pixels.
// See: https://github.com/memononen/SDF/issues/2
int he = img[k-1] == 255 || img[k+1] == 255;
int ve = img[k-stride] == 255 || img[k+stride] == 255;
if (!he && !ve) continue;
}
// Calculate gradient direction
gx = -(float)img[k-stride-1] - SDF_SQRT2*(float)img[k-1] - (float)img[k+stride-1] + (float)img[k-stride+1] + SDF_SQRT2*(float)img[k+1] + (float)img[k+stride+1];
gy = -(float)img[k-stride-1] - SDF_SQRT2*(float)img[k-stride] - (float)img[k-stride+1] + (float)img[k+stride-1] + SDF_SQRT2*(float)img[k+stride] + (float)img[k+stride+1];
if (fabsf(gx) < 0.001f && fabsf(gy) < 0.001f) continue;
glen = gx*gx + gy*gy;
if (glen > 0.0001f) {
glen = 1.0f / sqrtf(glen);
gx *= glen;
gy *= glen;
}
// Find nearest point on contour.
tk = x + y * width;
d = sdf__edgedf(gx, gy, (float)img[k]/255.0f);
tpt[tk].x = x + gx*d;
tpt[tk].y = y + gy*d;
tdist[tk] = sdf__distsqr(&c, &tpt[tk]);
}
}
// Calculate distance transform using sweep-and-update.
for (pass = 0; pass < SDF_MAX_PASSES; pass++){
int changed = 0;
// Bottom-left to top-right.
for (y = 1; y < height-1; y++) {
for (x = 1; x < width-1; x++) {
int k = x+y*width, kn, ch = 0;
struct SDFpoint c = { (float)x, (float)y }, pt;
float pd = tdist[k], d;
// (-1,-1)
kn = k - 1 - width;
if (tdist[kn] < pd) {
d = sdf__distsqr(&c, &tpt[kn]);
if (d + SDF_SLACK < pd) {
pt = tpt[kn];
pd = d;
ch = 1;
}
}
// (0,-1)
kn = k - width;
if (tdist[kn] < pd) {
d = sdf__distsqr(&c, &tpt[kn]);
if (d + SDF_SLACK < pd) {
pt = tpt[kn];
pd = d;
ch = 1;
}
}
// (1,-1)
kn = k + 1 - width;
if (tdist[kn] < pd) {
d = sdf__distsqr(&c, &tpt[kn]);
if (d + SDF_SLACK < pd) {
pt = tpt[kn];
pd = d;
ch = 1;
}
}
// (-1,0)
kn = k - 1;
if (tdist[kn] < tdist[k]) {
d = sdf__distsqr(&c, &tpt[kn]);
if (d + SDF_SLACK < pd) {
pt = tpt[kn];
pd = d;
ch = 1;
}
}
if (ch) {
tpt[k] = pt;
tdist[k] = pd;
changed++;
}
}
}
// Top-right to bottom-left.
for (y = height-2; y > 0 ; y--) {
for (x = width-2; x > 0; x--) {
int k = x+y*width, kn, ch = 0;
struct SDFpoint c = { (float)x, (float)y }, pt;
float pd = tdist[k], d;
// (1,0)
kn = k + 1;
if (tdist[kn] < pd) {
d = sdf__distsqr(&c, &tpt[kn]);
if (d + SDF_SLACK < pd) {
pt = tpt[kn];
pd = d;
ch = 1;
}
}
// (-1,1)
kn = k - 1 + width;
if (tdist[kn] < pd) {
d = sdf__distsqr(&c, &tpt[kn]);
if (d + SDF_SLACK < pd) {
pt = tpt[kn];
pd = d;
ch = 1;
}
}
// (0,1)
kn = k + width;
if (tdist[kn] < pd) {
d = sdf__distsqr(&c, &tpt[kn]);
if (d + SDF_SLACK < pd) {
pt = tpt[kn];
pd = d;
ch = 1;
}
}
// (1,1)
kn = k + 1 + width;
if (tdist[kn] < pd) {
d = sdf__distsqr(&c, &tpt[kn]);
if (d + SDF_SLACK < pd) {
pt = tpt[kn];
pd = d;
ch = 1;
}
}
if (ch) {
tpt[k] = pt;
tdist[k] = pd;
changed++;
}
}
}
if (changed == 0) break;
}
// Map to good range.
scale = 1.0f / radius;
for (y = 0; y < height; y++) {
for (x = 0; x < width; x++) {
float d = sqrtf(tdist[x+y*width]) * scale;
if (img[x+y*stride] > 127) d = -d;
out[x+y*outstride] = (unsigned char)(sdf__clamp01(0.5f - d*0.5f) * 255.0f);
}
}
}
int sdfBuildDistanceField(unsigned char* out, int outstride, float radius,
const unsigned char* img, int width, int height, int stride)
{
unsigned char* temp = (unsigned char*)malloc(width*height*sizeof(float)*3);
if (temp == NULL) return 0;
sdfBuildDistanceFieldNoAlloc(out, outstride, radius, img, width, height, stride, temp);
free(temp);
return 1;
}
#endif //SDF_IMPLEMENTATION
gitextract_bltkgvwp/
├── .gitignore
├── LICENSE.txt
├── README.md
├── example/
│ ├── example.c
│ ├── stb_image.c
│ └── stb_image_write.h
├── premake4.lua
└── src/
└── sdf.h
SYMBOL INDEX (195 symbols across 4 files)
FILE: example/example.c
function getPerfTime (line 13) | int64_t getPerfTime()
function deltaTimeUsec (line 20) | int deltaTimeUsec(int64_t start, int64_t end)
type Image (line 38) | struct Image {
type Image (line 43) | struct Image
type Image (line 45) | struct Image
type Image (line 45) | struct Image
type Image (line 59) | struct Image
type Image (line 61) | struct Image
type Image (line 61) | struct Image
function imgSave (line 78) | void imgSave(struct Image* img, const char* path)
function imgFree (line 83) | void imgFree(struct Image* img)
function imgInverse (line 90) | void imgInverse(struct Image* img)
function GLuint (line 97) | GLuint imgTexture(struct Image* img)
function loadImage (line 121) | int loadImage(const char* imageFile, float radius, float* imageAspect, G...
function key (line 166) | static void key(GLFWwindow* window, int key, int scancode, int action, i...
function main (line 206) | int main()
FILE: example/stb_image.c
type stbi_uc (line 208) | typedef unsigned char stbi_uc;
type stbi_io_callbacks (line 231) | typedef struct
type uint8 (line 361) | typedef unsigned char uint8;
type uint16 (line 362) | typedef unsigned short uint16;
type int16 (line 363) | typedef signed short int16;
type uint32 (line 364) | typedef unsigned int uint32;
type int32 (line 365) | typedef signed int int32;
type uint (line 366) | typedef unsigned int uint;
type stbi (line 393) | typedef struct
function start_mem (line 413) | static void start_mem(stbi *s, uint8 const *buffer, int len)
function start_callbacks (line 422) | static void start_callbacks(stbi *s, stbi_io_callbacks *c, void *user)
function stdio_read (line 434) | static int stdio_read(void *user, char *data, int size)
function stdio_skip (line 439) | static void stdio_skip(void *user, unsigned n)
function stdio_eof (line 444) | static int stdio_eof(void *user)
function start_file (line 456) | static void start_file(stbi *s, FILE *f)
function stbi_rewind (line 465) | static void stbi_rewind(stbi *s)
function e (line 503) | static int e(const char *str)
function stbi_image_free (line 524) | void stbi_image_free(void *retval_from_stbi_load)
function stbi_is_hdr_from_memory (line 643) | int stbi_is_hdr_from_memory(stbi_uc const *buffer, int len)
function stbi_is_hdr (line 657) | extern int stbi_is_hdr (char const *filename)
function stbi_is_hdr_from_file (line 668) | extern int stbi_is_hdr_from_file(FILE *f)
function stbi_is_hdr_from_callbacks (line 680) | extern int stbi_is_hdr_from_callbacks(stbi_io_callbacks const *clbk...
function stbi_hdr_to_ldr_gamma (line 695) | void stbi_hdr_to_ldr_gamma(float gamma) { h2l_gamma_i = 1/gamma; }
function stbi_hdr_to_ldr_scale (line 696) | void stbi_hdr_to_ldr_scale(float scale) { h2l_scale_i = 1/scale; }
function stbi_ldr_to_hdr_gamma (line 698) | void stbi_ldr_to_hdr_gamma(float gamma) { l2h_gamma = gamma; }
function stbi_ldr_to_hdr_scale (line 699) | void stbi_ldr_to_hdr_scale(float scale) { l2h_scale = scale; }
function refill_buffer (line 715) | static void refill_buffer(stbi *s)
function stbi_inline (line 729) | stbi_inline static int get8(stbi *s)
function stbi_inline (line 740) | stbi_inline static int at_eof(stbi *s)
function stbi_inline (line 752) | stbi_inline static uint8 get8u(stbi *s)
function skip (line 757) | static void skip(stbi *s, int n)
function getn (line 770) | static int getn(stbi *s, stbi_uc *buffer, int n)
function get16 (line 794) | static int get16(stbi *s)
function uint32 (line 800) | static uint32 get32(stbi *s)
function get16le (line 806) | static int get16le(stbi *s)
function uint32 (line 812) | static uint32 get32le(stbi *s)
function uint8 (line 829) | static uint8 compute_y(int r, int g, int b)
function stbi_uc (line 897) | static stbi_uc *hdr_to_ldr(float *data, int x, int y, int comp)
type huffman (line 953) | typedef struct
type jpeg (line 964) | typedef struct
function build_huffman (line 1003) | static int build_huffman(huffman *h, int *count)
function grow_buffer_unsafe (line 1044) | static void grow_buffer_unsafe(jpeg *j)
function stbi_inline (line 1065) | stbi_inline static int decode(jpeg *j, huffman *h)
function stbi_inline (line 1116) | stbi_inline static int extend_receive(jpeg *j, int n)
function decode_block (line 1159) | static int decode_block(jpeg *j, short data[64], huffman *hdc, huffman *...
function stbi_inline (line 1194) | stbi_inline static uint8 clamp(int x)
type stbi_dequantize_t (line 1246) | typedef unsigned short stbi_dequantize_t;
type uint8 (line 1248) | typedef uint8 stbi_dequantize_t;
function idct_block (line 1252) | static void idct_block(uint8 *out, int out_stride, short data[64], stbi_...
function stbi_install_idct (line 1316) | void stbi_install_idct(stbi_idct_8x8 func)
function uint8 (line 1326) | static uint8 get_marker(jpeg *j)
function reset (line 1343) | static void reset(jpeg *j)
function parse_entropy_coded_data (line 1355) | static int parse_entropy_coded_data(jpeg *z)
function process_marker (line 1427) | static int process_marker(jpeg *z, int m)
function process_scan_header (line 1496) | static int process_scan_header(jpeg *z)
function process_frame_header (line 1521) | static int process_frame_header(jpeg *z, int scan)
function decode_jpeg_header (line 1600) | static int decode_jpeg_header(jpeg *z, int scan)
function decode_jpeg_image (line 1621) | static int decode_jpeg_image(jpeg *j)
type uint8 (line 1654) | typedef uint8 *(*resample_row_func)(uint8 *out, uint8 *in0, uint8 *in1,
function uint8 (line 1659) | static uint8 *resample_row_1(uint8 *out, uint8 *in_near, uint8 *in_far, ...
function uint8 (line 1668) | static uint8* resample_row_v_2(uint8 *out, uint8 *in_near, uint8 *in_far...
function uint8 (line 1678) | static uint8* resample_row_h_2(uint8 *out, uint8 *in_near, uint8 *in_fa...
function uint8 (line 1708) | static uint8 *resample_row_hv_2(uint8 *out, uint8 *in_near, uint8 *in_fa...
function uint8 (line 1732) | static uint8 *resample_row_generic(uint8 *out, uint8 *in_near, uint8 *in...
function YCbCr_to_RGB_row (line 1747) | static void YCbCr_to_RGB_row(uint8 *out, const uint8 *y, const uint8 *pc...
function stbi_install_YCbCr_to_RGB (line 1775) | void stbi_install_YCbCr_to_RGB(stbi_YCbCr_to_RGB_run func)
function cleanup_jpeg (line 1783) | static void cleanup_jpeg(jpeg *j)
type stbi_resample (line 1798) | typedef struct
function uint8 (line 1808) | static uint8 *load_jpeg_image(jpeg *z, int *out_x, int *out_y, int *comp...
function stbi_jpeg_test (line 1915) | static int stbi_jpeg_test(stbi *s)
function stbi_jpeg_info_raw (line 1925) | static int stbi_jpeg_info_raw(jpeg *j, int *x, int *y, int *comp)
function stbi_jpeg_info (line 1937) | static int stbi_jpeg_info(stbi *s, int *x, int *y, int *comp)
type zhuffman (line 1957) | typedef struct
function stbi_inline (line 1967) | stbi_inline static int bitreverse16(int n)
function stbi_inline (line 1976) | stbi_inline static int bit_reverse(int v, int bits)
function zbuild_huffman (line 1984) | static int zbuild_huffman(zhuffman *z, uint8 *sizelist, int num)
type zbuf (line 2035) | typedef struct
function stbi_inline (line 2049) | stbi_inline static int zget8(zbuf *z)
function fill_bits (line 2055) | static void fill_bits(zbuf *z)
function zreceive (line 2064) | int zreceive(zbuf *z, int n)
function stbi_inline (line 2074) | stbi_inline static int zhuffman_decode(zbuf *a, zhuffman *z)
function expand (line 2101) | static int expand(zbuf *z, int n) // need to make room for n bytes
function parse_huffman_block (line 2132) | static int parse_huffman_block(zbuf *a)
function compute_huffman_codes (line 2160) | static int compute_huffman_codes(zbuf *a)
function parse_uncompressed_block (line 2206) | static int parse_uncompressed_block(zbuf *a)
function parse_zlib_header (line 2235) | static int parse_zlib_header(zbuf *a)
function init_defaults (line 2250) | static void init_defaults(void)
function parse_zlib (line 2262) | static int parse_zlib(zbuf *a, int parse_header)
function do_zlib (line 2293) | static int do_zlib(zbuf *a, char *obuf, int olen, int exp, int parse_hea...
function stbi_zlib_decode_buffer (line 2340) | int stbi_zlib_decode_buffer(char *obuffer, int olen, char const *ibuffer...
function stbi_zlib_decode_noheader_buffer (line 2367) | int stbi_zlib_decode_noheader_buffer(char *obuffer, int olen, const char...
type chunk (line 2389) | typedef struct
function chunk (line 2397) | static chunk get_chunk_header(stbi *s)
function check_png_header (line 2405) | static int check_png_header(stbi *s)
type png (line 2414) | typedef struct
function paeth (line 2431) | static int paeth(int a, int b, int c)
function create_png_image_raw (line 2443) | static int create_png_image_raw(png *a, uint8 *raw, uint32 raw_len, int ...
function create_png_image (line 2520) | static int create_png_image(png *a, uint8 *raw, uint32 raw_len, int out_...
function compute_transparency (line 2561) | static int compute_transparency(png *z, uint8 tc[3], int out_n)
function expand_palette (line 2586) | static int expand_palette(png *a, uint8 *palette, int len, int pal_img_n)
function stbi_set_unpremultiply_on_load (line 2626) | void stbi_set_unpremultiply_on_load(int flag_true_if_should_unpremultiply)
function stbi_convert_iphone_png_to_rgb (line 2630) | void stbi_convert_iphone_png_to_rgb(int flag_true_if_should_convert)
function stbi_de_iphone (line 2635) | static void stbi_de_iphone(png *z)
function parse_png_file (line 2677) | static int parse_png_file(png *z, int scan, int req_comp)
function stbi_png_test (line 2860) | static int stbi_png_test(stbi *s)
function stbi_png_info_raw (line 2868) | static int stbi_png_info_raw(png *p, int *x, int *y, int *comp)
function stbi_png_info (line 2880) | static int stbi_png_info(stbi *s, int *x, int *y, int *comp)
function bmp_test (line 2889) | static int bmp_test(stbi *s)
function stbi_bmp_test (line 2903) | static int stbi_bmp_test(stbi *s)
function high_bit (line 2912) | static int high_bit(unsigned int z)
function bitcount (line 2924) | static int bitcount(unsigned int a)
function shiftsigned (line 2934) | static int shiftsigned(int v, int shift, int bits)
function stbi_uc (line 2951) | static stbi_uc *bmp_load(stbi *s, int *x, int *y, int *comp, int req_comp)
function stbi_uc (line 3146) | static stbi_uc *stbi_bmp_load(stbi *s,int *x, int *y, int *comp, int req...
function tga_info (line 3155) | static int tga_info(stbi *s, int *x, int *y, int *comp)
function stbi_tga_info (line 3192) | int stbi_tga_info(stbi *s, int *x, int *y, int *comp)
function tga_test (line 3197) | static int tga_test(stbi *s)
function stbi_tga_test (line 3217) | static int stbi_tga_test(stbi *s)
function stbi_uc (line 3224) | static stbi_uc *tga_load(stbi *s, int *x, int *y, int *comp, int req_comp)
function stbi_uc (line 3450) | static stbi_uc *stbi_tga_load(stbi *s, int *x, int *y, int *comp, int re...
function psd_test (line 3459) | static int psd_test(stbi *s)
function stbi_psd_test (line 3465) | static int stbi_psd_test(stbi *s)
function stbi_uc (line 3472) | static stbi_uc *psd_load(stbi *s, int *x, int *y, int *comp, int req_comp)
function stbi_uc (line 3630) | static stbi_uc *stbi_psd_load(stbi *s, int *x, int *y, int *comp, int re...
function pic_is4 (line 3642) | static int pic_is4(stbi *s,const char *str)
function pic_test (line 3652) | static int pic_test(stbi *s)
type pic_packet_t (line 3668) | typedef struct
function stbi_uc (line 3673) | static stbi_uc *pic_readval(stbi *s, int channel, stbi_uc *dest)
function pic_copyval (line 3687) | static void pic_copyval(int channel,stbi_uc *dest,const stbi_uc *src)
function stbi_uc (line 3696) | static stbi_uc *pic_load2(stbi *s,int width,int height,int *comp, stbi_u...
function stbi_uc (line 3807) | static stbi_uc *pic_load(stbi *s,int *px,int *py,int *comp,int req_comp)
function stbi_pic_test (line 3840) | static int stbi_pic_test(stbi *s)
function stbi_uc (line 3847) | static stbi_uc *stbi_pic_load(stbi *s, int *x, int *y, int *comp, int re...
type stbi_gif_lzw (line 3854) | typedef struct stbi_gif_lzw_struct {
type stbi_gif (line 3860) | typedef struct stbi_gif_struct
function gif_test (line 3877) | static int gif_test(stbi *s)
function stbi_gif_test (line 3887) | static int stbi_gif_test(stbi *s)
function stbi_gif_parse_colortable (line 3894) | static void stbi_gif_parse_colortable(stbi *s, uint8 pal[256][4], int nu...
function stbi_gif_header (line 3905) | static int stbi_gif_header(stbi *s, stbi_gif *g, int *comp, int is_info)
function stbi_gif_info_raw (line 3933) | static int stbi_gif_info_raw(stbi *s, int *x, int *y, int *comp)
function stbi_out_gif_code (line 3945) | static void stbi_out_gif_code(stbi_gif *g, uint16 code)
function uint8 (line 3979) | static uint8 *stbi_process_gif_raster(stbi *s, stbi_gif *g)
function stbi_fill_gif_background (line 4058) | static void stbi_fill_gif_background(stbi_gif *g)
function uint8 (line 4073) | static uint8 *stbi_gif_load_next(stbi *s, stbi_gif *g, int *comp, int re...
function stbi_uc (line 4173) | static stbi_uc *stbi_gif_load(stbi *s, int *x, int *y, int *comp, int re...
function stbi_gif_info (line 4188) | static int stbi_gif_info(stbi *s, int *x, int *y, int *comp)
function hdr_test (line 4198) | static int hdr_test(stbi *s)
function stbi_hdr_test (line 4208) | static int stbi_hdr_test(stbi* s)
function hdr_convert (line 4238) | static void hdr_convert(float *output, stbi_uc *input, int req_comp)
function stbi_hdr_info (line 4381) | static int stbi_hdr_info(stbi *s, int *x, int *y, int *comp)
function stbi_bmp_info (line 4421) | static int stbi_bmp_info(stbi *s, int *x, int *y, int *comp)
function stbi_psd_info (line 4449) | static int stbi_psd_info(stbi *s, int *x, int *y, int *comp)
function stbi_pic_info (line 4480) | static int stbi_pic_info(stbi *s, int *x, int *y, int *comp)
function stbi_info_main (line 4525) | static int stbi_info_main(stbi *s, int *x, int *y, int *comp)
function stbi_info (line 4550) | int stbi_info(char const *filename, int *x, int *y, int *comp)
function stbi_info_from_file (line 4560) | int stbi_info_from_file(FILE *f, int *x, int *y, int *comp)
function stbi_info_from_memory (line 4572) | int stbi_info_from_memory(stbi_uc const *buffer, int len, int *x, int *y...
function stbi_info_from_callbacks (line 4579) | int stbi_info_from_callbacks(stbi_io_callbacks const *c, void *user, int...
FILE: example/stb_image_write.h
type stbiw_uint32 (line 79) | typedef unsigned int stbiw_uint32;
function writefv (line 82) | static void writefv(FILE *f, const char *fmt, va_list v)
function write3 (line 102) | static void write3(FILE *f, unsigned char a, unsigned char b, unsigned c...
function write_pixels (line 109) | static void write_pixels(FILE *f, int rgb_dir, int vdir, int x, int y, i...
function outfile (line 152) | static int outfile(char const *filename, int rgb_dir, int vdir, int x, i...
function stbi_write_bmp (line 168) | int stbi_write_bmp(char const *filename, int x, int y, int comp, const v...
function stbi_write_tga (line 177) | int stbi_write_tga(char const *filename, int x, int y, int comp, const v...
function stbi__zlib_bitrev (line 220) | static int stbi__zlib_bitrev(int code, int codebits)
function stbi__zlib_countm (line 230) | static unsigned int stbi__zlib_countm(unsigned char *a, unsigned char *b...
function stbi__zhash (line 238) | static unsigned int stbi__zhash(unsigned char *data)
function stbi__crc32 (line 367) | unsigned int stbi__crc32(unsigned char *buffer, int len)
function stbi__wpcrc (line 385) | static void stbi__wpcrc(unsigned char **data, int len)
function stbi__paeth (line 391) | static unsigned char stbi__paeth(int a, int b, int c)
function stbi_write_png (line 489) | int stbi_write_png(char const *filename, int x, int y, int comp, const v...
FILE: src/sdf.h
function sdf__clamp01 (line 82) | static float sdf__clamp01(float x)
function sdfCoverageToDistanceField (line 87) | void sdfCoverageToDistanceField(unsigned char* out, int outstride,
function sdf__edgedf (line 155) | static float sdf__edgedf(float gx, float gy, float a)
type SDFpoint (line 185) | struct SDFpoint {
function sdf__distsqr (line 189) | static float sdf__distsqr(struct SDFpoint* a, struct SDFpoint* b)
function sdfBuildDistanceFieldNoAlloc (line 195) | void sdfBuildDistanceFieldNoAlloc(unsigned char* out, int outstride, flo...
function sdfBuildDistanceField (line 375) | int sdfBuildDistanceField(unsigned char* out, int outstride, float radius,
Condensed preview — 8 files, each showing path, character count, and a content snippet. Download the .json file or copy for the full structured content (209K chars).
[
{
"path": ".gitignore",
"chars": 258,
"preview": "## Compiled source #\n*.com\n*.class\n*.dll\n*.exe\n*.o\n*.so\ntest\n\n## Logs and databases #\n*.log\n*.sql\n*.sqlite\n\n## OS genera"
},
{
"path": "LICENSE.txt",
"chars": 1168,
"preview": "The MIT License (MIT)\n\nCopyright (C) 2014 Mikko Mononen (memon@inside.org)\nCopyright (C) 2009-2012 Stefan Gustavson (ste"
},
{
"path": "README.md",
"chars": 3901,
"preview": "SDF\n===\n\nSweep-and-update Euclidean distance transform of an antialised image for contour texturing.\n\nThe code is based "
},
{
"path": "example/example.c",
"chars": 6936,
"preview": "#include <stdio.h>\n#include <string.h>\n#include <float.h>\n#include <GLFW/glfw3.h>\n#define SDF_IMPLEMENTATION\n#include \"s"
},
{
"path": "example/stb_image.c",
"chars": 151821,
"preview": "/* stbi-1.33 - public domain JPEG/PNG reader - http://nothings.org/stb_image.c\r\n when you control the images you're lo"
},
{
"path": "example/stb_image_write.h",
"chars": 18599,
"preview": "/* stbiw-0.92 - public domain - http://nothings.org/stb/stb_image_write.h\r\n writes out PNG/BMP/TGA images to C stdio -"
},
{
"path": "premake4.lua",
"chars": 845,
"preview": "\nlocal action = _ACTION or \"\"\n\nsolution \"distancefield\"\n\tlocation ( \"build\" )\n\tconfigurations { \"Debug\", \"Release\" }\n\tpl"
},
{
"path": "src/sdf.h",
"chars": 11845,
"preview": "/*\n Copyright (C) 2014 Mikko Mononen (memon@inside.org)\n Copyright (C) 2009-2012 Stefan Gustavson (stefan.gustavson@gmai"
}
]
About this extraction
This page contains the full source code of the memononen/SDF GitHub repository, extracted and formatted as plain text for AI agents and large language models (LLMs). The extraction includes 8 files (190.8 KB), approximately 60.9k tokens, and a symbol index with 195 extracted functions, classes, methods, constants, and types. Use this with OpenClaw, Claude, ChatGPT, Cursor, Windsurf, or any other AI tool that accepts text input. You can copy the full output to your clipboard or download it as a .txt file.
Extracted by GitExtract — free GitHub repo to text converter for AI. Built by Nikandr Surkov.