This gives me about 60% reduction in size for 1280x720 frames, and my test system (Linux in VirtualBox on a single core) can do ~40 such compressions per second (without much optimization). Looks like a good choice for data sets of mostly small numerical

Applied Rice coding to the difference.

Than in the original data, in fact, most of the values are, not Variation (or what's the correct mathematical term) becomes much less Difference dataīecomes signed and the range increases to -1023~1023, but the data

Took binary difference between two consecutive frames.

Huffman etc.), that redundancy would be automatically taken into account, so no extra packing is necessary. However, I thought that if I use some entropy coding (e. Of frames and compress them all together, but I can look 1 frame backīased on the above, my first thought was to bit-pack the data, so that those 6 redundant bits are not wasted on every word.

The compression has to be done on the fly, so I can not gather a lot.

Very little, but still, even the stationary objects are not completely so in the image space. However, it looks like the camera also shakes a little.Because of high FPS, even the parts that change don't change much.Noise (denoising is not an option, as it is supposed to be lossless Nearly identical - but still not completely, due to the inevitable The camera does not move, so big parts of consecutive frames are.I want to implement some kind of lossless compression of the data before transmitting it over a network.

As there's no 10 bit data type on my platform, the original data is stored in memory using 16-bit words. The data I get is in Bayer format and uses 10 bit per pixel. Both the resolution and the FPS are quite high. I have a video coming from a stationary camera.

Uniplayer h264