OutofAi
OutofAi
I tried to keep the whole thing simple, in terms of implementation, as much as I could and I didn't run more experiments - You could run some ablation studies...
thanks for running that performance study, i am looking at writing a differntiable renderer with my co-founder, but still early stages, we worked on real-time pytorch visualisation as a first...
I investigated a bit more, and seems like some of the issues are how pytorch handles things, mainly the decision of whether the operation is done on the cpu or...
yeah, valid points, I ran more tests on it, majority of the bottleneck is either torch deciding to do an operation on cpu even though everything is on cuda or...
it's really difficult to pinpoint why! but if I had to guess it could either be related to pytorch version and they've recently optimised it and I was using an...
I think the blurriness is due to the naive initialisation in my method or the kernel size used, technically I should be able to get a super sharp image from...
also in terms of memory, I am currently using significant amount of extra backup points, which are not really needed considering the image converge with around 3000 points at the...
Thanks for the feedback, am not entirely sure what you mean, If I understood correctly and you want to visualise Stable Diffusion output during sampling of inference steps, you should...
With PIL and numpy you get the CPU transfer, which is not the idea behind this module, you can modify the callback to store the image for you in ,...
+1 the same issue