Skip to content

Commit

Permalink
Added function: ImageDither()
Browse files Browse the repository at this point in the history 
Corrected some code details
  • Loading branch information
@raysan5
raysan5 committed Dec 30, 2015
1 parent 6cb0603 commit 5dbb93d
Show file tree
Hide file tree
Showing 2 changed files with 134 additions and 29 deletions.
2 changes: 2 additions & 0 deletions src/raylib.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -654,6 +654,7 @@ Color *GetImageData(Image image);
Image GetTextureData(Texture2D texture); // Get pixel data from GPU texture and return an Image
void ImageToPOT(Image *image, Color fillColor); // Convert image to POT (power-of-two)
void ImageFormat(Image *image, int newFormat); // Convert image data to desired format
void ImageDither(Image *image, int rBpp, int gBpp, int bBpp, int aBpp); // Dither image data to 16bpp or lower (Floyd-Steinberg dithering)
Image ImageCopy(Image image); // Create an image duplicate (useful for transformations)
void ImageCrop(Image *image, Rectangle crop); // Crop an image to a defined rectangle
void ImageResize(Image *image, int newWidth, int newHeight); // Resize and image (bilinear filtering)
Expand Down Expand Up @@ -692,6 +693,7 @@ Vector2 MeasureTextEx(SpriteFont spriteFont, const char *text, int fontSize, int

void DrawFPS(int posX, int posY); // Shows current FPS on top-left corner
const char *FormatText(const char *text, ...); // Formatting of text with variables to 'embed'
const char *SubText(const char *text, int position, int length); // Get a piece of a text string

//------------------------------------------------------------------------------------
// Basic 3d Shapes Drawing Functions (Module: models)
Expand Down
161 changes: 132 additions & 29 deletions src/textures.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -549,8 +549,7 @@ void ImageFormat(Image *image, int newFormat)

for (int i = 0; i < image->width*image->height; i++)
{
((unsigned char *)image->data)[i] = (unsigned char)((float)pixels[k].r*0.299f + (float)pixels[k].g*0.587f + (float)pixels[k].b*0.114f);
k++;
((unsigned char *)image->data)[i] = (unsigned char)((float)pixels[i].r*0.299f + (float)pixels[i].g*0.587f + (float)pixels[i].b*0.114f);
}

} break;
Expand All @@ -570,9 +569,9 @@ void ImageFormat(Image *image, int newFormat)
{
image->data = (unsigned short *)malloc(image->width*image->height*sizeof(unsigned short));

unsigned char r;
unsigned char g;
unsigned char b;
unsigned char r = 0;
unsigned char g = 0;
unsigned char b = 0;

for (int i = 0; i < image->width*image->height; i++)
{
Expand All @@ -581,8 +580,6 @@ void ImageFormat(Image *image, int newFormat)
b = (unsigned char)(round((float)pixels[k].b*31/255));

((unsigned short *)image->data)[i] = (unsigned short)r << 11 | (unsigned short)g << 5 | (unsigned short)b;

k++;
}

} break;
Expand All @@ -600,45 +597,43 @@ void ImageFormat(Image *image, int newFormat)
} break;
case UNCOMPRESSED_R5G5B5A1:
{
#define ALPHA_THRESHOLD 50

image->data = (unsigned short *)malloc(image->width*image->height*sizeof(unsigned short));

unsigned char r;
unsigned char g;
unsigned char b;
unsigned char a = 1;
unsigned char r = 0;
unsigned char g = 0;
unsigned char b = 0;
unsigned char a = 0;

for (int i = 0; i < image->width*image->height; i++)
{
r = (unsigned char)(round((float)pixels[k].r*31/255));
g = (unsigned char)(round((float)pixels[k].g*31/255));
b = (unsigned char)(round((float)pixels[k].b*31/255));
a = (pixels[k].a > 50) ? 1 : 0;
r = (unsigned char)(round((float)pixels[i].r*31/255));
g = (unsigned char)(round((float)pixels[i].g*31/255));
b = (unsigned char)(round((float)pixels[i].b*31/255));
a = (pixels[i].a > ALPHA_THRESHOLD) ? 1 : 0;

((unsigned short *)image->data)[i] = (unsigned short)r << 11 | (unsigned short)g << 6 | (unsigned short)b << 1| (unsigned short)a;

k++;
((unsigned short *)image->data)[i] = (unsigned short)r << 11 | (unsigned short)g << 6 | (unsigned short)b << 1 | (unsigned short)a;
}

} break;
case UNCOMPRESSED_R4G4B4A4:
{
image->data = (unsigned short *)malloc(image->width*image->height*sizeof(unsigned short));

unsigned char r;
unsigned char g;
unsigned char b;
unsigned char a;
unsigned char r = 0;
unsigned char g = 0;
unsigned char b = 0;
unsigned char a = 0;

for (int i = 0; i < image->width*image->height; i++)
{
r = (unsigned char)(round((float)pixels[k].r*15/255));
g = (unsigned char)(round((float)pixels[k].g*15/255));
b = (unsigned char)(round((float)pixels[k].b*15/255));
a = (unsigned char)(round((float)pixels[k].a*15/255));
r = (unsigned char)(round((float)pixels[i].r*15/255));
g = (unsigned char)(round((float)pixels[i].g*15/255));
b = (unsigned char)(round((float)pixels[i].b*15/255));
a = (unsigned char)(round((float)pixels[i].a*15/255));

((unsigned short *)image->data)[i] = (unsigned short)r << 12 | (unsigned short)g << 8| (unsigned short)b << 4| (unsigned short)a;

k++;
}

} break;
Expand All @@ -664,6 +659,114 @@ void ImageFormat(Image *image, int newFormat)
}
}

// Dither image data to 16bpp or lower (Floyd-Steinberg dithering)
// NOTE: In case selected bpp do not represent an known 16bit format,
// dithered data is stored in the LSB part of the unsigned short
void ImageDither(Image *image, int rBpp, int gBpp, int bBpp, int aBpp)
{
if (image->format >= 8)
{
TraceLog(WARNING, "Compressed data formats can not be dithered");
return;
}

if ((rBpp+gBpp+bBpp+aBpp) > 16)
{
TraceLog(WARNING, "Unsupported dithering bpps (%ibpp), only 16bpp or lower modes supported", (rBpp+gBpp+bBpp+aBpp));
}
else
{
Color *pixels = GetImageData(*image);

free(image->data); // free old image data

if ((image->format != UNCOMPRESSED_R8G8B8) && (image->format != UNCOMPRESSED_R8G8B8A8))
{
TraceLog(WARNING, "Image format is already 16bpp or lower, dithering could have no effect");
}

// Define new image format, check if desired bpp match internal known format
if ((rBpp == 5) && (gBpp == 6) && (bBpp == 5) && (aBpp == 0)) image->format = UNCOMPRESSED_R5G6B5;
else if ((rBpp == 5) && (gBpp == 5) && (bBpp == 5) && (aBpp == 1)) image->format = UNCOMPRESSED_R5G5B5A1;
else if ((rBpp == 4) && (gBpp == 4) && (bBpp == 4) && (aBpp == 4)) image->format = UNCOMPRESSED_R4G4B4A4;
else
{
image->format = 0;
TraceLog(WARNING, "Unsupported dithered OpenGL internal format: %ibpp (R%iG%iB%iA%i)", (rBpp+gBpp+bBpp+aBpp), rBpp, gBpp, bBpp, aBpp);
}

// NOTE: We will store the dithered data as unsigned short (16bpp)
image->data = (unsigned short *)malloc(image->width*image->height*sizeof(unsigned short));

Color oldpixel = WHITE;
Color newpixel = WHITE;

int error_r, error_g, error_b;
unsigned short pixel_r, pixel_g, pixel_b, pixel_a; // Used for 16bit pixel composition

#define MIN(a,b) (((a)<(b))?(a):(b))

for (int y = 0; y < image->height; y++)
{
for (int x = 0; x < image->width; x++)
{
oldpixel = pixels[y*image->width + x];

// TODO: New pixel obtained by bits truncate, it would be better to round values (check ImageFormat())
newpixel.r = oldpixel.r>>(8 - rBpp); // R bits
newpixel.g = oldpixel.g>>(8 - gBpp); // G bits
newpixel.b = oldpixel.b>>(8 - bBpp); // B bits
newpixel.a = oldpixel.a>>(8 - aBpp); // A bits (not used on dithering)

// NOTE: Error must be computed between new and old pixel but using same number of bits!
// We want to know how much color precision we have lost...
error_r = (int)oldpixel.r - (int)(newpixel.r<<(8 - rBpp));
error_g = (int)oldpixel.g - (int)(newpixel.g<<(8 - gBpp));
error_b = (int)oldpixel.b - (int)(newpixel.b<<(8 - bBpp));

pixels[y*image->width + x] = newpixel;

// NOTE: Some cases are out of the array and should be ignored
if (x < (image->width - 1))
{
pixels[y*image->width + x+1].r = MIN((int)pixels[y*image->width + x+1].r + (int)((float)error_r*7.0f/16), 0xff);
pixels[y*image->width + x+1].g = MIN((int)pixels[y*image->width + x+1].g + (int)((float)error_g*7.0f/16), 0xff);
pixels[y*image->width + x+1].b = MIN((int)pixels[y*image->width + x+1].b + (int)((float)error_b*7.0f/16), 0xff);
}

if ((x > 0) && (y < (image->height - 1)))
{
pixels[(y+1)*image->width + x-1].r = MIN((int)pixels[(y+1)*image->width + x-1].r + (int)((float)error_r*3.0f/16), 0xff);
pixels[(y+1)*image->width + x-1].g = MIN((int)pixels[(y+1)*image->width + x-1].g + (int)((float)error_g*3.0f/16), 0xff);
pixels[(y+1)*image->width + x-1].b = MIN((int)pixels[(y+1)*image->width + x-1].b + (int)((float)error_b*3.0f/16), 0xff);
}

if (y < (image->height - 1))
{
pixels[(y+1)*image->width + x].r = MIN((int)pixels[(y+1)*image->width + x].r + (int)((float)error_r*5.0f/16), 0xff);
pixels[(y+1)*image->width + x].g = MIN((int)pixels[(y+1)*image->width + x].g + (int)((float)error_g*5.0f/16), 0xff);
pixels[(y+1)*image->width + x].b = MIN((int)pixels[(y+1)*image->width + x].b + (int)((float)error_b*5.0f/16), 0xff);
}

if ((x < (image->width - 1)) && (y < (image->height - 1)))
{
pixels[(y+1)*image->width + x+1].r = MIN((int)pixels[(y+1)*image->width + x+1].r + (int)((float)error_r*1.0f/16), 0xff);
pixels[(y+1)*image->width + x+1].g = MIN((int)pixels[(y+1)*image->width + x+1].g + (int)((float)error_g*1.0f/16), 0xff);
pixels[(y+1)*image->width + x+1].b = MIN((int)pixels[(y+1)*image->width + x+1].b + (int)((float)error_b*1.0f/16), 0xff);
}

pixel_r = (unsigned short)newpixel.r;
pixel_g = (unsigned short)newpixel.g;
pixel_b = (unsigned short)newpixel.b;
pixel_a = (unsigned short)newpixel.a;

((unsigned short *)image->data)[y*image->width + x] = (pixel_r<<(gBpp + bBpp + aBpp)) | (pixel_g<<(bBpp + aBpp)) | (pixel_b<<aBpp) | pixel_a;
}
}

free(pixels);
}
}

// Convert image to POT (power-of-two)
// NOTE: Requirement on OpenGL ES 2.0 (RPI, HTML5)
Expand Down

0 comments on commit 5dbb93d

Please sign in to comment.