Comfy native support is now available

[Amazon90]

Comfy officially provides native support for VACE. Please provide a usable workflow.
Comfy-Org/ComfyUI#7711

[Amazon90]

Bytedance recently released a subject-consistent video generation model also based on WAN. However, due to the good results I’ve had with ACE Plus, my confidence in it, and the native support for VACE in ComfyUI, I’m more looking forward to seeing how VACE performs. Everyone can give it a try and make a comparison.

https://github.com/Phantom-video/Phantom

https://huggingface.co/bytedance-research/Phantom

[sobermh]

@Amazon90 Do you have any workflows based on WanVaceToVideo that I could use as a reference? thx!!!

[Amazon90]

@Amazon90 Do you have any workflows based on WanVaceToVideo that I could use as a reference? thx!!!

No, because I haven’t figured out what the control_video, control_masks, and reference_image inputs do in the WanVaceToVideo node, or what the trim_latent output is actually used for.

[sobermh]

@Amazon90 Do you have any workflows based on WanVaceToVideo that I could use as a reference? thx!!!

No, because I haven’t figured out what the control_video, control_masks, and reference_image inputs do in the WanVaceToVideo node, or what the trim_latent output is actually used for.

I tried to create a simple flow and run it using parameters from the source code, but got badresults.

hope the official will provide a standard workflow soon.

[Amazon90]

@Amazon90 Do you have any workflows based on WanVaceToVideo that I could use as a reference? thx!!!

No, because I haven’t figured out what the control_video, control_masks, and reference_image inputs do in the WanVaceToVideo node, or what the trim_latent output is actually used for.

I tried to create a simple flow and run it using parameters from the source code, but got badresults.

hope the official will provide a standard workflow soon.

https://docs.comfy.org/tutorials/video/wan/vace