Wan 14B Self Forcing T2V Lora by Kijai

Kijai extracted 14B self forcing lightx2v model as a lora:
https://huggingface.co/Kijai/WanVideo\_comfy/blob/main/Wan21\_T2V\_14B\_lightx2v\_cfg\_step\_distill\_lora\_rank32.safetensors
The quality and speed are simply amazing (720x480 97 frames video in \~100 second on my 4070ti super 16 vram, using 4 steps, lcm, 1 cfg, 8 shift, I believe it can be even faster)


also the link to the workflow I saw:
https://civitai.com/models/1585622/causvid-accvid-lora-massive-speed-up-for-wan21-made-by-kijai?modelVersionId=1909719

TLDR; just use the standard Kijai's T2V workflow and add the lora,
also works great with other motion loras

https://redd.it/1lcz7ij
@rStableDiffusion
Is SUPIR still the best upscaler if so, what is the last updates they have made?

Hello,
I’ve been wondering about SUIPIR it’s been around for a while and remains an impressive upscaler. However, I’m curious if there have been any recent updates to it, or if newer, potentially better alternatives have emerged since its release.

https://redd.it/1ld4rg0
@rStableDiffusion
Is CivitAI still the place to download loras for WAN?

I know of tensor art and huggingface, but CivitAI was a goldmine for WAN video loras. The first month or two of its release I could find a new lora every day that I wanted to try. Now there is nothing.

Is there a site that I haven't listed yet that is maybe not well known?

https://redd.it/1ld7agc
@rStableDiffusion
This media is not supported in your browser
VIEW IN TELEGRAM
Control the motion of anything without extra prompting! Free tool to create controls

https://redd.it/1lddjkx
@rStableDiffusion