Innovate Futures @ Benji
patreon
(ComfyUI Workflow) My Mod Skyreels V2 Diffusion Forcing Video Extend
🕑 Added 2025-04-22 17:02:12 +0000 UTC
Mentioned in this video tutorial : https://youtu.be/PBwDq_MDjKM
It's very easy to use, shouldn't have any problem to run this.
Other links and resources: https://www.patreon.com/posts/127221329
Comments
Andrewb
Ah, I'm assuming the shorter length addition would be the 53 left in the start index/Get Image or Mask Range From Batch. Guess I'll find out when I try another run. Might conceivably help with quality if it's not getting enough context or whatever maybe.
Andrewb
Heya, one question here, using I2V, the latter extensions seem to degrade very quickly and not extend particularly long, so the first generation lasts 6-7 seconds, then each subsequent extension adds ~2 with a drop in quality that is roughly the same as a last-image-for-first-frame-of-next-video workflow.
Benjamin Law
Have you check the input image size as well?
Thap Huy
Hi I got this error. I dont know how to fix. can you help me please ? thanks a lot WanVideoDiffusionForcingSampler Trying to create tensor with negative dimension -13: [-13, 4096] My video is 480 width and 832 height. I set tile_x = tile_y = 272. tile_stride_x = 128, tile_stride_y = 160
