Thanks for this custom node – it works perfectly in my tests and integrates really well into ComfyUI.
The mask quality I get from VideoMaMa is noticeably better than what I usually get from SAM3.
One thing that would make it even more powerful is having some control over the mask via points or an iterative refinement mechanism. For example:
_Allowing foreground/background point hints on specific frames to guide the matting.
_Being able to refine the mask over several iterations, propagating corrections across the sequence.
If this is already possible in a way I missed, I’d be happy to adjust my workflow, but if not, please consider this a feature request. Thanks again for sharing this node!
Thanks for this custom node – it works perfectly in my tests and integrates really well into ComfyUI.
The mask quality I get from VideoMaMa is noticeably better than what I usually get from SAM3.
One thing that would make it even more powerful is having some control over the mask via points or an iterative refinement mechanism. For example:
_Allowing foreground/background point hints on specific frames to guide the matting.
_Being able to refine the mask over several iterations, propagating corrections across the sequence.
If this is already possible in a way I missed, I’d be happy to adjust my workflow, but if not, please consider this a feature request. Thanks again for sharing this node!