Very interesting!

This seems somewhat similar to the recently published GIFnets[1]. However, I believe GIFnets is training a reusable network to a predict palettes, and pixel assignments, while this post is focusing on optimising the "weights" (i.e. pixel values) for a single image.

I wonder if the loss functions from GIFnets could be applied to this single-image approach to potentially solve the banding problem via something a little more "perceptual" than the variance term mentioned.

[1]: "GIFnets: Differentiable GIF Encoding Framework" https://arxiv.org/abs/2006.13434