I'm thinking of extending either cam or qcam applications to be able to send frames in inferencing servers that run vision models like ramalama

[ericcurtin]

I wanna re-create this, but use libcamera more directly (this is actually macOS Chrome processing the camera frames, RamaLama running on Linux):

https://www.linkedin.com/posts/curtine_redhatsummit-ramalama-llamacpp-activity-7329930381510520832-IkDW?utm_source=share&utm_medium=member_desktop&rcm=ACoAAAZRsnUBW5iL5v_Xf22wBZVUAoKR30FOY_w

I think it would be a good showcase of how RamaLama and libcamera can be used together for AI/vision models.

As an aside, I know it's a completely different OS but, did anyone ever consider trying to get libcamera working on macOS (as drastically different it may be)?

[ericcurtin]

I wonder what we should use for a http client, libcurl or try and find some http client in qt or sdl since they are dependancies already