Home Creators Posts Import Register Favorites Logout
haven't archived this post yet. have a subscription? use the importer!

Downloads

  • AnimateDiff Lightning workflow.json

Missing 1 full-res photo, 1 file, as this post has not yet been imported.

Content

Here is my Tutorial and the ComfyUI Workflow below. Make sure you UPDATE your ComfyUI + Extensions first!

Model: https://huggingface.co/ByteDance/AnimateDiff-Lightning/tree/main
ComfyUI Manager (in case you don't have it):

1) go into custom nodes Folder and write CMD in the address bar and hit enter.

2) git clone https://github.com/ltdrdata/ComfyUI-Manager

3) hit enter, wait for it to finish, close CMD and start Comfyui. Now you have the manager.

4) open my workflow and click on "Install Mission Nodes" and install all missing nodes

Files

Previews only

Stable Video Lightning - Local Install Guide

#### Links from my Video #### #### Join and Support me #### Buy me a Coffee: https://www.buymeacoffee.com/oliviotutorials Join my Facebook Group: https://www.facebook.com/groups/theairevolution Joint my Discord Group: https://discord.gg/XKAk7GUzAW AI Newsletter: https://oliviotutorials.podia.com/newsletter Support me on Patreon: https://www.patreon.com/sarikas

Comments

Thomas Coispel

Thanks for sharing ! How can we add a Img2Vid imput with this workflow ? (in ComfyUI)

Olivio Sarikas

Hi, you can use a "Video Load (Path)" Node from the Video Helper Suite Nodes Pack and put that into a VAE Encoder and then into the Latent Input of the Ksmapler. Make sure you use a mp4 with 512 and no longer than 16 frame. Then set the ksampler to euler_acestral and play with the denoise setting. I didn't find a img2video way, since it needs at least 2 frames. You might be able to do img2video with a IPAdapter for stlye transfere though. But i was told IPAdapter does not play well with this model. So a lot of experimentation might be needed

EuropesHeart

I have the following problem. Error occurred when executing GMFSS Fortuna VFI: No module named 'cupy'

Thomas Coispel

Thanks for that solution ! I will test it in the future. Meanwhile I've managed to link the following nodes : - Load image - image scale to side - RepeatImageBatch (x16) -VAE Encode - KSampler By setting steps to 5 (with a model at 4 steps), I success to make some interesting IMG2VIDEO outputs