A workflow uses the latest Wan2.2 All-In-One-Mage model and Next Scene Lora and ComfyUI subgraph to create an unlimited video generation process with low VRAM. Perfect smooth transaction between different scenes. With consistency improvements in ID and Color. No more weird pause and direction change.
Please check the video for detailed instructions before you start:
If you think this is helpful, please like the video and subscribe for more.
Description
FAQ
Comments (8)
Excellent workflow and video explanation, it is super complete. And I didn't know the AIO models.
The arrangement with multiple scene nodes is much better than the 'loop' approach IMHO, because this way you can tweak as much as you want before proceeding to the next step.
is there any explanation available?
Great workflow! Many thanks!
it's a cool workflow, but unfortunately the next scene design group drastically changes faces. other than that it's really good
thanks . you could give the V2 a try. https://huggingface.co/lovis93/next-scene-qwen-image-lora-2509
I'll need to examine this but a condensed approach would be use one Next Scene node in a loop with node that iterates through a prompt array. Then you wouldn't need so many subgraphs. At most two, the initial and one next scene node. In theory.
Anybody Wanna Explain this one?