From ba3c0cbce1485bab9e6fc6d4f2a63b15043a52eb Mon Sep 17 00:00:00 2001 From: Mikko Mononen Date: Tue, 25 Feb 2014 22:53:46 +0200 Subject: [PATCH] Fix for issue #40: Added support to render using premultiplied alpha - added support to take screen shots - added support to render to frame buffer using premultiplied alpha - fixed spelling NVGaling -> NVGalign --- example/demo.c | 104 ++++++++ example/demo.h | 2 + example/example_gl2.c | 18 +- example/example_gl3.c | 18 +- example/example_gles2.c | 18 +- example/example_gles3.c | 18 +- example/stb_image_write.h | 511 ++++++++++++++++++++++++++++++++++++++ src/nanovg.c | 10 +- src/nanovg.h | 16 +- src/nanovg_gl2.h | 11 +- src/nanovg_gl3.h | 9 +- src/nanovg_gl3buf.h | 9 +- 12 files changed, 723 insertions(+), 21 deletions(-) create mode 100644 example/stb_image_write.h diff --git a/example/demo.c b/example/demo.c index 951b281..ea53cba 100644 --- a/example/demo.c +++ b/example/demo.c @@ -7,6 +7,8 @@ #endif #include #include "nanovg.h" +#define STB_IMAGE_WRITE_IMPLEMENTATION +#include "stb_image_write.h" #ifdef _MSC_VER @@ -850,3 +852,105 @@ void renderDemo(struct NVGcontext* vg, float mx, float my, float width, float he nvgRestore(vg); } +static int mini(int a, int b) { return a < b ? a : b; } + +static void unpremultiplyAlpha(unsigned char* image, int w, int h, int stride) +{ + int x,y; + + // Unpremultiply + for (y = 0; y < h; y++) { + unsigned char *row = &image[y*stride]; + for (x = 0; x < w; x++) { + int r = row[0], g = row[1], b = row[2], a = row[3]; + if (a != 0) { + row[0] = (int)mini(r*255/a, 255); + row[1] = (int)mini(g*255/a, 255); + row[2] = (int)mini(b*255/a, 255); + } + row += 4; + } + } + + // Defringe + for (y = 0; y < h; y++) { + unsigned char *row = &image[y*stride]; + for (x = 0; x < w; x++) { + int r = 0, g = 0, b = 0, a = row[3], n = 0; + if (a == 0) { + if (x-1 > 0 && row[-1] != 0) { + r += row[-4]; + g += row[-3]; + b += row[-2]; + n++; + } + if (x+1 < w && row[7] != 0) { + r += row[4]; + g += row[5]; + b += row[6]; + n++; + } + if (y-1 > 0 && row[-stride+3] != 0) { + r += row[-stride]; + g += row[-stride+1]; + b += row[-stride+2]; + n++; + } + if (y+1 < h && row[stride+3] != 0) { + r += row[stride]; + g += row[stride+1]; + b += row[stride+2]; + n++; + } + if (n > 0) { + row[0] = r/n; + row[1] = g/n; + row[2] = b/n; + } + } + row += 4; + } + } +} + +static void setAlpha(unsigned char* image, int w, int h, int stride, unsigned char a) +{ + int x, y; + for (y = 0; y < h; y++) { + unsigned char* row = &image[y*stride]; + for (x = 0; x < w; x++) + row[x*4+3] = a; + } +} + +static void flipHorizontal(unsigned char* image, int w, int h, int stride) +{ + int i = 0, j = h-1, k; + while (i < j) { + unsigned char* ri = &image[i * stride]; + unsigned char* rj = &image[j * stride]; + for (k = 0; k < w*4; k++) { + unsigned char t = ri[k]; + ri[k] = rj[k]; + rj[k] = t; + } + i++; + j--; + } +} + +void saveScreenShot(int w, int h, int premult, const char* name) +{ + unsigned char* image = (unsigned char*)malloc(w*h*4); + if (image == NULL) + return; + glReadPixels(0, 0, w, h, GL_RGBA, GL_UNSIGNED_BYTE, image); + if (premult) + unpremultiplyAlpha(image, w, h, w*4); + else + setAlpha(image, w, h, w*4, 255); + flipHorizontal(image, w, h, w*4); + stbi_write_png(name, w, h, 4, image, w*4); + free(image); +} + diff --git a/example/demo.h b/example/demo.h index 5162716..999bb31 100644 --- a/example/demo.h +++ b/example/demo.h @@ -16,6 +16,8 @@ int loadDemoData(struct NVGcontext* vg, struct DemoData* data); void freeDemoData(struct NVGcontext* vg, struct DemoData* data); void renderDemo(struct NVGcontext* vg, float mx, float my, float width, float height, float t, int blowup, struct DemoData* data); +void saveScreenShot(int w, int h, int premult, const char* name); + #ifdef __cplusplus } #endif diff --git a/example/example_gl2.c b/example/example_gl2.c index 7707d2f..2ce42ca 100644 --- a/example/example_gl2.c +++ b/example/example_gl2.c @@ -34,6 +34,8 @@ void errorcb(int error, const char* desc) } int blowup = 0; +int screenshot = 0; +int premult = 0; static void key(GLFWwindow* window, int key, int scancode, int action, int mods) { @@ -43,6 +45,10 @@ static void key(GLFWwindow* window, int key, int scancode, int action, int mods) glfwSetWindowShouldClose(window, GL_TRUE); if (key == GLFW_KEY_SPACE && action == GLFW_PRESS) blowup = !blowup; + if (key == GLFW_KEY_S && action == GLFW_PRESS) + screenshot = 1; + if (key == GLFW_KEY_P && action == GLFW_PRESS) + premult = !premult; } int main() @@ -124,7 +130,10 @@ int main() // Update and render glViewport(0, 0, fbWidth, fbHeight); - glClearColor(0.3f, 0.3f, 0.32f, 1.0f); + if (premult) + glClearColor(0,0,0,0); + else + glClearColor(0.3f, 0.3f, 0.32f, 1.0f); glClear(GL_COLOR_BUFFER_BIT|GL_DEPTH_BUFFER_BIT|GL_STENCIL_BUFFER_BIT); glEnable(GL_BLEND); @@ -132,7 +141,7 @@ int main() glEnable(GL_CULL_FACE); glDisable(GL_DEPTH_TEST); - nvgBeginFrame(vg, winWidth, winHeight, pxRatio); + nvgBeginFrame(vg, winWidth, winHeight, pxRatio, premult ? NVG_PREMULTIPLIED_ALPHA : NVG_STRAIGHT_ALPHA); renderDemo(vg, mx,my, winWidth,winHeight, t, blowup, &data); renderGraph(vg, 5,5, &fps); @@ -141,6 +150,11 @@ int main() glEnable(GL_DEPTH_TEST); + if (screenshot) { + screenshot = 0; + saveScreenShot(fbWidth, fbHeight, premult, "dump.png"); + } + glfwSwapBuffers(window); glfwPollEvents(); } diff --git a/example/example_gl3.c b/example/example_gl3.c index 5aa460d..d5982a2 100644 --- a/example/example_gl3.c +++ b/example/example_gl3.c @@ -38,6 +38,8 @@ void errorcb(int error, const char* desc) } int blowup = 0; +int screenshot = 0; +int premult = 0; static void key(GLFWwindow* window, int key, int scancode, int action, int mods) { @@ -47,6 +49,10 @@ static void key(GLFWwindow* window, int key, int scancode, int action, int mods) glfwSetWindowShouldClose(window, GL_TRUE); if (key == GLFW_KEY_SPACE && action == GLFW_PRESS) blowup = !blowup; + if (key == GLFW_KEY_S && action == GLFW_PRESS) + screenshot = 1; + if (key == GLFW_KEY_P && action == GLFW_PRESS) + premult = !premult; } int main() @@ -139,7 +145,10 @@ int main() // Update and render glViewport(0, 0, fbWidth, fbHeight); - glClearColor(0.3f, 0.3f, 0.32f, 1.0f); + if (premult) + glClearColor(0,0,0,0); + else + glClearColor(0.3f, 0.3f, 0.32f, 1.0f); glClear(GL_COLOR_BUFFER_BIT|GL_DEPTH_BUFFER_BIT|GL_STENCIL_BUFFER_BIT); glEnable(GL_BLEND); @@ -147,7 +156,7 @@ int main() glEnable(GL_CULL_FACE); glDisable(GL_DEPTH_TEST); - nvgBeginFrame(vg, winWidth, winHeight, pxRatio); + nvgBeginFrame(vg, winWidth, winHeight, pxRatio, premult ? NVG_PREMULTIPLIED_ALPHA : NVG_STRAIGHT_ALPHA); renderDemo(vg, mx,my, winWidth,winHeight, t, blowup, &data); @@ -171,6 +180,11 @@ int main() for (i = 0; i < n; i++) updateGraph(&gpuGraph, gpuTimes[i]); + if (screenshot) { + screenshot = 0; + saveScreenShot(fbWidth, fbHeight, premult, "dump.png"); + } + glfwSwapBuffers(window); glfwPollEvents(); } diff --git a/example/example_gles2.c b/example/example_gles2.c index ebb4ccd..342779f 100644 --- a/example/example_gles2.c +++ b/example/example_gles2.c @@ -32,6 +32,8 @@ void errorcb(int error, const char* desc) } int blowup = 0; +int screenshot = 0; +int premult = 0; static void key(GLFWwindow* window, int key, int scancode, int action, int mods) { @@ -41,6 +43,10 @@ static void key(GLFWwindow* window, int key, int scancode, int action, int mods) glfwSetWindowShouldClose(window, GL_TRUE); if (key == GLFW_KEY_SPACE && action == GLFW_PRESS) blowup = !blowup; + if (key == GLFW_KEY_S && action == GLFW_PRESS) + screenshot = 1; + if (key == GLFW_KEY_P && action == GLFW_PRESS) + premult = !premult; } int main() @@ -109,7 +115,10 @@ int main() // Update and render glViewport(0, 0, fbWidth, fbHeight); - glClearColor(0.3f, 0.3f, 0.32f, 1.0f); + if (premult) + glClearColor(0,0,0,0); + else + glClearColor(0.3f, 0.3f, 0.32f, 1.0f); glClear(GL_COLOR_BUFFER_BIT|GL_DEPTH_BUFFER_BIT|GL_STENCIL_BUFFER_BIT); glEnable(GL_BLEND); @@ -117,13 +126,18 @@ int main() glEnable(GL_CULL_FACE); glDisable(GL_DEPTH_TEST); - nvgBeginFrame(vg, winWidth, winHeight, pxRatio); + nvgBeginFrame(vg, winWidth, winHeight, pxRatio, premult ? NVG_PREMULTIPLIED_ALPHA : NVG_STRAIGHT_ALPHA); renderDemo(vg, mx,my, winWidth,winHeight, t, blowup, &data); renderGraph(vg, 5,5, &fps); nvgEndFrame(vg); + if (screenshot) { + screenshot = 0; + saveScreenShot(fbWidth, fbHeight, premult, "dump.png"); + } + glEnable(GL_DEPTH_TEST); glfwSwapBuffers(window); diff --git a/example/example_gles3.c b/example/example_gles3.c index c0c06ef..d2713e6 100644 --- a/example/example_gles3.c +++ b/example/example_gles3.c @@ -32,6 +32,8 @@ void errorcb(int error, const char* desc) } int blowup = 0; +int screenshot = 0; +int premult = 0; static void key(GLFWwindow* window, int key, int scancode, int action, int mods) { @@ -41,6 +43,10 @@ static void key(GLFWwindow* window, int key, int scancode, int action, int mods) glfwSetWindowShouldClose(window, GL_TRUE); if (key == GLFW_KEY_SPACE && action == GLFW_PRESS) blowup = !blowup; + if (key == GLFW_KEY_S && action == GLFW_PRESS) + screenshot = 1; + if (key == GLFW_KEY_P && action == GLFW_PRESS) + premult = !premult; } int main() @@ -109,7 +115,10 @@ int main() // Update and render glViewport(0, 0, fbWidth, fbHeight); - glClearColor(0.3f, 0.3f, 0.32f, 1.0f); + if (premult) + glClearColor(0,0,0,0); + else + glClearColor(0.3f, 0.3f, 0.32f, 1.0f); glClear(GL_COLOR_BUFFER_BIT|GL_DEPTH_BUFFER_BIT|GL_STENCIL_BUFFER_BIT); glEnable(GL_BLEND); @@ -117,7 +126,7 @@ int main() glEnable(GL_CULL_FACE); glDisable(GL_DEPTH_TEST); - nvgBeginFrame(vg, winWidth, winHeight, pxRatio); + nvgBeginFrame(vg, winWidth, winHeight, pxRatio, premult ? NVG_PREMULTIPLIED_ALPHA : NVG_STRAIGHT_ALPHA); renderDemo(vg, mx,my, winWidth,winHeight, t, blowup, &data); renderGraph(vg, 5,5, &fps); @@ -126,6 +135,11 @@ int main() glEnable(GL_DEPTH_TEST); + if (screenshot) { + screenshot = 0; + saveScreenShot(fbWidth, fbHeight, premult, "dump.png"); + } + glfwSwapBuffers(window); glfwPollEvents(); } diff --git a/example/stb_image_write.h b/example/stb_image_write.h new file mode 100644 index 0000000..5de3159 --- /dev/null +++ b/example/stb_image_write.h @@ -0,0 +1,511 @@ +/* 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 +#include +#include +#include +#include + +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 +*/ diff --git a/src/nanovg.c b/src/nanovg.c index 8a1efad..96d071f 100644 --- a/src/nanovg.c +++ b/src/nanovg.c @@ -107,6 +107,7 @@ struct NVGcontext { float devicePxRatio; struct FONScontext* fs; int fontImage; + int alphaBlend; int drawCallCount; int fillTriCount; int strokeTriCount; @@ -200,6 +201,8 @@ struct NVGcontext* nvgCreateInternal(struct NVGparams* params) ctx->ncommands = 0; ctx->ccommands = NVG_INIT_PATH_SIZE; + ctx->alphaBlend = NVG_STRAIGHT_ALPHA; + ctx->cache = nvg__allocPathCache(); if (ctx->cache == NULL) goto error; @@ -249,7 +252,7 @@ void nvgDeleteInternal(struct NVGcontext* ctx) free(ctx); } -void nvgBeginFrame(struct NVGcontext* ctx, int windowWidth, int windowHeight, float devicePixelRatio) +void nvgBeginFrame(struct NVGcontext* ctx, int windowWidth, int windowHeight, float devicePixelRatio, int alphaBlend) { /* printf("Tris: draws:%d fill:%d stroke:%d text:%d TOT:%d\n", ctx->drawCallCount, ctx->fillTriCount, ctx->strokeTriCount, ctx->textTriCount, @@ -260,8 +263,9 @@ void nvgBeginFrame(struct NVGcontext* ctx, int windowWidth, int windowHeight, fl nvgReset(ctx); nvg__setDevicePixelRatio(ctx, devicePixelRatio); + ctx->alphaBlend = alphaBlend; - ctx->params.renderViewport(ctx->params.userPtr, windowWidth, windowHeight); + ctx->params.renderViewport(ctx->params.userPtr, windowWidth, windowHeight, ctx->alphaBlend); ctx->drawCallCount = 0; ctx->fillTriCount = 0; @@ -271,7 +275,7 @@ void nvgBeginFrame(struct NVGcontext* ctx, int windowWidth, int windowHeight, fl void nvgEndFrame(struct NVGcontext* ctx) { - ctx->params.renderFlush(ctx->params.userPtr); + ctx->params.renderFlush(ctx->params.userPtr, ctx->alphaBlend); } unsigned int nvgRGB(unsigned char r, unsigned char g, unsigned char b) diff --git a/src/nanovg.h b/src/nanovg.h index 9c4c166..15be9eb 100644 --- a/src/nanovg.h +++ b/src/nanovg.h @@ -62,7 +62,7 @@ enum NVGpatternRepeat { NVG_REPEATY = 0x02, // Repeat image pattern in Y direction }; -enum NVGaling { +enum NVGalign { // Horizontal align NVG_ALIGN_LEFT = 1<<0, // Default, align text horizontally to left. NVG_ALIGN_CENTER = 1<<1, // Align text horizontally to center. @@ -74,6 +74,10 @@ enum NVGaling { NVG_ALIGN_BASELINE = 1<<6, // Default, align text vertically to baseline. }; +enum NVGalpha { + NVG_STRAIGHT_ALPHA, + NVG_PREMULTIPLIED_ALPHA, +}; // Begin drawing a new frame // Calls to nanovg drawing API should be wrapped in nvgBeginFrame() & nvgEndFrame() @@ -83,7 +87,11 @@ enum NVGaling { // For example, GLFW returns two dimension for an opened window: window size and // frame buffer size. In that case you would set windowWidth/Height to the window size // devicePixelRatio to: frameBufferWidth / windowWidth. -void nvgBeginFrame(struct NVGcontext* ctx, int windowWidth, int windowHeight, float devicePixelRatio); +// AlphaBlend controls if drawing the shapes to the render target should be done using straight or +// premultiplied alpha. If rendering directly to framebuffer you probably want to use NVG_STRAIGHT_ALPHA, +// if rendering to texture which should contain transparent regions NVG_PREMULTIPLIED_ALPHA is the +// right choice. +void nvgBeginFrame(struct NVGcontext* ctx, int windowWidth, int windowHeight, float devicePixelRatio, int alphaBlend); // Ends drawing flushing remaining render state. void nvgEndFrame(struct NVGcontext* ctx); @@ -430,8 +438,8 @@ struct NVGparams { int (*renderDeleteTexture)(void* uptr, int image); int (*renderUpdateTexture)(void* uptr, int image, int x, int y, int w, int h, const unsigned char* data); int (*renderGetTextureSize)(void* uptr, int image, int* w, int* h); - void (*renderViewport)(void* uptr, int width, int height); - void (*renderFlush)(void* uptr); + void (*renderViewport)(void* uptr, int width, int height, int alphaBlend); + void (*renderFlush)(void* uptr, int alphaBlend); void (*renderFill)(void* uptr, struct NVGpaint* paint, struct NVGscissor* scissor, const float* bounds, const struct NVGpath* paths, int npaths); void (*renderStroke)(void* uptr, struct NVGpaint* paint, struct NVGscissor* scissor, float strokeWidth, const struct NVGpath* paths, int npaths); void (*renderTriangles)(void* uptr, struct NVGpaint* paint, struct NVGscissor* scissor, const struct NVGvertex* verts, int nverts); diff --git a/src/nanovg_gl2.h b/src/nanovg_gl2.h index cbb5acb..14e2db8 100644 --- a/src/nanovg_gl2.h +++ b/src/nanovg_gl2.h @@ -626,17 +626,24 @@ static int glnvg__setupPaint(struct GLNVGcontext* gl, struct NVGpaint* paint, st return 1; } -static void glnvg__renderViewport(void* uptr, int width, int height) +static void glnvg__renderViewport(void* uptr, int width, int height, int alphaBlend) { struct GLNVGcontext* gl = (struct GLNVGcontext*)uptr; gl->viewWidth = (float)width; gl->viewHeight = (float)height; + + if (alphaBlend == NVG_PREMULTIPLIED_ALPHA) + glBlendFuncSeparate(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA, GL_ONE, GL_ONE_MINUS_SRC_ALPHA); + else + glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA); } -static void glnvg__renderFlush(void* uptr) + +static void glnvg__renderFlush(void* uptr, int alphaBlend) { // struct GLNVGcontext* gl = (struct GLNVGcontext*)uptr; NVG_NOTUSED(uptr); + NVG_NOTUSED(alphaBlend); } static int glnvg__maxVertCount(const struct NVGpath* paths, int npaths) diff --git a/src/nanovg_gl3.h b/src/nanovg_gl3.h index ca2abe8..233aad6 100644 --- a/src/nanovg_gl3.h +++ b/src/nanovg_gl3.h @@ -632,14 +632,19 @@ static int glnvg__setupPaint(struct GLNVGcontext* gl, struct NVGpaint* paint, st return 1; } -static void glnvg__renderViewport(void* uptr, int width, int height) +static void glnvg__renderViewport(void* uptr, int width, int height, int alphaBlend) { struct GLNVGcontext* gl = (struct GLNVGcontext*)uptr; gl->viewWidth = (float)width; gl->viewHeight = (float)height; + + if (alphaBlend == NVG_PREMULTIPLIED_ALPHA) + glBlendFuncSeparate(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA, GL_ONE, GL_ONE_MINUS_SRC_ALPHA); + else + glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA); } -static void glnvg__renderFlush(void* uptr) +static void glnvg__renderFlush(void* uptr, int alphaBlend) { // struct GLNVGcontext* gl = (struct GLNVGcontext*)uptr; NVG_NOTUSED(uptr); diff --git a/src/nanovg_gl3buf.h b/src/nanovg_gl3buf.h index 4a4984a..7f741a5 100644 --- a/src/nanovg_gl3buf.h +++ b/src/nanovg_gl3buf.h @@ -636,9 +636,10 @@ static void glnvg__setUniforms(struct GLNVGcontext* gl, int uniformOffset, int i } } -static void glnvg__renderViewport(void* uptr, int width, int height) +static void glnvg__renderViewport(void* uptr, int width, int height, int alphaBlend) { struct GLNVGcontext* gl = (struct GLNVGcontext*)uptr; + NVG_NOTUSED(alphaBlend); gl->view[0] = (float)width; gl->view[1] = (float)height; } @@ -727,7 +728,7 @@ static void glnvg__triangles(struct GLNVGcontext* gl, struct GLNVGcall* call) glDrawArrays(GL_TRIANGLES, call->triangleOffset, call->triangleCount); } -static void glnvg__renderFlush(void* uptr) +static void glnvg__renderFlush(void* uptr, int alphaBlend) { struct GLNVGcontext* gl = (struct GLNVGcontext*)uptr; int i; @@ -736,6 +737,10 @@ static void glnvg__renderFlush(void* uptr) glUseProgram(gl->shader.prog); glEnable(GL_CULL_FACE); + if (alphaBlend == NVG_PREMULTIPLIED_ALPHA) + glBlendFuncSeparate(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA, GL_ONE, GL_ONE_MINUS_SRC_ALPHA); + else + glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA); // Upload ubo for frag shaders glBindBuffer(GL_UNIFORM_BUFFER, gl->fragBuf);