bloc97

The wiki is pretty much outdated at this point, and I've been too busy to work on this... I've already put the installations link in the releases section though. Do...

I can understand both sides of the problem... While catering to everyone would be the best case scenario, it unfortunately takes a lot of time. For each iteration of Anime4K...

Interesting paper and the implementation looks correct, however if I understood it correctly, it constructs a generalized interpolation kernel that can _approximate_ existing interpolation methods (such as Mitchell-Netravali). However, it...

Small notes after further examination, these should be addressed before merge: - You might want to use the normalized sinc function (sin(pi x)/(pi x)), if that's what libplacebo expects. -...

This is an artifact of the pre-upscaling filter used in Waifu2x, if I'm not mistaken. Using a transposed convolution instead of a fixed bilinear/bicubic filter could make extremely-low resolution cases...

In this example you can see that the effective "blur" for the 4x4 pixel image upscaled to 8x8 occupies a large portion of the resulting image, thus making the NN...

As a workaround, it is possible to pass the original image through a xBRZ filter, downscaling it using bicubic/bilinear, and then applying Waifu2x. This is equivalent of doing a pixel-art...

The problem is not that training such a network is hard. There's new network structures that are extremely simple, fast to train and are better than what Waifu2x was originally...

Until there's a good amount of demand for what you need, or if you are willing to give your own time and develop a NN-based pixel art upscaler, I am...

I think the "k-diffusion" you are referring to is the k-lms scheduler? It is already used by default in this implementation. About learning about diffusion models, I guess it would...