Feature Request: More Options for SAM2 / Auto Segmenation (choose model option)

[a3nima]

Ok, I think Swarm could have at least more options for segementation because it is a great and handy feature to have, just the model often makes whole areas black Let me post an example, I wanted to select the red area behind the person, but the default model does not detect it:

Maybe an option for different models in the settings or model settings that you can adjust to get the model to differentiate more /less between different areas.