I am creating a particular type of file compression schema for a particular computer graphic application and I need some help. I have already designed the schema that I would like to use, but now I need to be able to compress small blocks of pixels (8x8 pixels) without retaining all the JPG headers. Optimally, I need to get instead of the 192 bytes (8x8 block of pixels before compression) something in the order of 15-20 bytes (10:1 compression).
Even more ideally, I would like to get *always* 20 bytes out of an 8x8 block of pixels, no matter what they look like.
Any idea on how to get the simpler code that does achieve this kind of real ration (~20 bytes to represent 64 pixels)?