> Real-time video is significantly different than still-frames. It's why video compression uses completely different algorithms than jpeg compression, for examp

> Real-time video is significantly different than still-frames. It's why video compression uses completely different algorithms than jpeg compression, for example. That's not quite true, mainstream video codecs are all based around discrete cosine transforms (or integer approximations thereof) + scalar quantization, followed by entropy coding. If you look at early codecs like MPEG1 or MPEG2, individual frames are compressed in a manner almost identical to JPEG. The main difference is that video allows you to do motion compression, but that still leaves you with a residual (or difference) frame, which is compressed much like a jpeg. Modern codecs like h264 and h265 are far more complex, but the basic algorithms at the heart of them are not fundamentally different from JPEG. In fact you can use h264 as a still image codec, at about 2x the compression efficiency! > If it's possible to clean a puppy, it might be possible to extrapolate how light should behave in the general case. That bot did not "clean" a puppy, it just converted it to black and white. Since the dye only changed the fur color (chroma) without darkening/lightening it, it was not visible in the newly added synthetic colors. If you look at the image in black and white, the dye is obviously not present: https://i.imgur.com/xIkCP2m.jpg https://i.imgur.com/xIkCP2m.jpg