NSQY

Still nothing on this? It is, in my opinion, the biggest detractor.

Have you seen this implementation? https://github.com/AmusementClub/vs-dfttest2 On my 3900X, with AVX2 enabled, it is more than twice as fast as https://github.com/HomeOfVapourSynthEvolution/VapourSynth-DFTTest in a synthetic VS script. ``` Output 1000 frames...

A lot of functions ported from Avisynth have been outdated for years, if you have added anything to lvsfunc which was originally ported from AVS there's a good chance that...

Nothing currently strikes me as being a problem, my comment was mostly a preemptive warning for future rewrites. If I were to guess, the main area of trouble would be...

Seems to be caused by this commit: https://github.com/AUTOMATIC1111/stable-diffusion-webui/commit/adf723a9b2a8e83c71219a33a914f5dcd01d88f8 simply ``git checkout ddc503d1`` as a workaround.

fvsfunc is also slightly faster: ``Output 240 frames in 26.83 seconds (8.94 fps)`` Here is my input file, but any YUV420P16 clip should do I guess. 

Because the AVS script is single threaded. It would not be a fair comparison otherwise. What I am trying to say is that both Vapoursynth versions of GradFun3 use more...

One thing I noticed when looking at --filter-time output is that this function uses a lot of BoxBlur calls. std.Convolution is much faster when boxblur is using a radius if...

>As for the single-threaded performance, I don't think VapourSynth is designed to optimize for latency rather than throughput. You make a salient point. I had forgotten about PreFetch(). However, as...

Relevant https://github.com/vapoursynth/vapoursynth/issues/787