Masking Objects with SAM 2
More Infor Here:
https://github.com/kijai/ComfyUI-segment-anything-2
Download Models:
https://huggingface.co/Kijai/sam2-safetensors/tree/main
Save to:
\ComfyUI\models\sam2
Step 1: Import Your Video
Import your video clip into the software.
Step 2: Save a Single Frame
Save a single frame from your video clip as an image.
Step 3: Import the Image to the Points Editor
Right-click on the Points Editor and select "Load Image" to load the saved frame into the editor.
Step 4: Select the Object
In the Points Editor, Point to the object you want to isolate or manipulate to select it.
Description
FAQ
Comments (10)
Excuse me. Can I use your model in webui? I tried to use it in webui, but I got KeyError: 'sam2_hiera'...
can not you run on Comfyui ?
ah so cool..all the masking I have seen is whole body besides samv2 from meta..can I mask individual object in video like hair but not face,,face but not outfit etc? Also can this mask the background?
Awesome, what diffusion models is it compatible with? ..sd1.5 ?... is it also compatible with Flux Dev GGUF models by any chance?...I noticed this is a workflow but every tutorial shows to put the samv2 download in sam2 folder..if that is the case then how do i load the workflow?
@ThricePi it works with everything :)
@Crimson_Intellect oohh k sounds good. il try it out with flux..thanks for the replies.
When I try to change the resolution size to 1024x1024 it will only let me change it to xyxy not a number. Any ideas?
That is not resolution XY are vertical and horizontal coordinate, try to use upscaler , Thx
Thank you so much do you have one for images instead