opengl c ++：如何将32位png文件转换为16位565RGB文件

Question

I am trying to convert 32bit png files to 16 bit file formats, I understand how to convert 16 bit file formats between eachother (eg RGB565 RGBA4444) However I'm not sure how to go about converting from a 32 bit to a 16 bit.

My main questions are: How do I find how the 32 bit pngs stored (are 8 bits each assigned to R,B,G, and A values)?

How do I lost precision but still maintain roughly the same value?

Thanks in advance

Answer 1

使用libpng会比手动实现更好。

Answer 2

I am not familiar with the exact layout of the 32bit png pixel, but assuming it is relatively consistent with other formats you probably want to do something similar to this:

// Get the pixel from the png:
unsigned int pngPixel = getPngPixel();

unsigned char r = (pngPixel & 0xFF000000) >> 24;
unsigned char g = (pngPixel & 0x00FF0000) >> 16;
unsigned char b = (pngPixel & 0x0000FF00) >> 8;
unsigned char a = (pngPixel & 0x000000FF);

// you can collapse this to one line, but for clarity...
// masking off the least significant bits.
unsigned short rgb565Pixel = (r & 0xF8) << 11; 
rgb565Pixel |= (g & 0xFC) << 5;
rgb565Pixel |= (b & 0xF8);

// Again you could collapse this down to one line, but for clarity...
// masking off the least significant bits.
unsigned short rgba4Pixel = (r & 0xF0) << 12;
rgba4Pixel |= (g & 0xF0) << 8;
rgba4Pixel |= (b & 0xF0) << 4;
rgba4Pixel |= (a & 0xF0);  

Consider this pseudocode.

One could argue that masking off the least significant bits, especially when converting from 8 bit to 4 bit, is not a very good way to convert between the two, and they would be right. You could instead use a conversion function:

unsigned int convertColor(unsigned char c, unsigned int oldMax, unsigned int newMax) {
   double oldColor = c;
   double percentOfMax = oldColor / oldMax;
   return ((unsigned int)(newMax * percentOfMax)) & newMax;
}

// now we can do this
unsigned short rgba4Pixel = convertColor(r, 0xFF, 0x0F) << 12;
rgba4Pixel |= convertColor(g, 0xFF, 0x0F) << 8;
rgba4Pixel |= convertColor(b, 0xFF, 0x0F) << 4;
rgba4Pixel |= convertColor(a, 0xFF, 0x0F);