leapfusion-image2vid-test

/

image2vid-960x544

like 17

Thank you for this release, it's much improved from the first release.

Can anyone share some workflows with this please ?

Can anyone share some workflows with this please ?

in kijai's hunyuan video wrapper, there is an example workflow: https://github.com/kijai/ComfyUI-HunyuanVideoWrapper/blob/main/example_workflows/hyvideo_leapfusion_img2vid_example_01.json

Thanks !

There's also a workflow that works on native comfy now
https://github.com/kijai/ComfyUI-KJNodes/blob/main/example_workflows/leapfusion_hunyuuanvideo_i2v_native_testing.json

If I wanted to merge this and base a new lora off of it, do img_mod linear layers such as double_blocks.10.img_mod.linear need to be included (or any blocks/layers need include/exclude aside from default musubi-tuner) ? The lora is a fair bit difficult for base hunyuan to learn and I'd like to also have some i2v functionality once a good one comes out.