This is a slapping together of bits that work to make decent Img-2-Vid, none of it is my work outside of arranging things. I strongly recommend using my newer WF over this one as it can do moar!
(v1.2 -- Double Blocks Loras 'model' output can be linked to the 'model' input on the Flexi Lora Loader node, also added a TeaCache Sampler (with 1.6x and 2.1x speeds) and a VRAM Cleaner.
Loras used:
Img2Vid info and links to checkpoint.
Info on Flexi Lora Loader. (All I know is varied weights as a vid is sequenced = more coherence)
Description
Slight alteration to WF a removal of a note and linked the Difffusion model to the Double Blocks Lora loader as it should have been in the first place.
FAQ
Comments (7)
nice. also works with skyreels models. thanks for posting.
generates some very nice videos... finally bring flux to life :)
is there something capping video length to 3 seconds? increasing frame count doesn't increase video length past 3.
WOW just wow... i2v brought one of my charcoal drawing line-art images to life. frickin' awesome.
Will this workflow work with WAN also?
FYI. There is a new FIXED version of the model that came out a few days after the initial I2V model, you guys may want to give it a try. I found the intial version changed the persons face quite a bit. When I used the newer FIXED version I change guidance_type in the HunyuanImageToVideo node to V2. If you don't, it looks like it ignores the input photo. Good luck.
I'm encountering an issue with the workflow. I keep getting the following error:
CLIPVisionLoader 'NoneType' object is not callable
even though the file llava_llama3_vision.safetensors is correctly placed in the clip_vision folder
Looks like we don't have an active mirror for this file right now.
CivArchive is a community-maintained index — we catalog mirrors that volunteers upload to HuggingFace, torrents, and other public hosts. Looks like no one has uploaded a copy of this file yet.
Some files do get recovered over time through contributions. If you're looking for this one, feel free to ask in Discord, or help preserve it if you have a copy.