Just for the record - in my case using VL.IO.NDI and a NDI Webcam works waaay better than SpoutCam.
The ML models used are working on pretty low resolutions anyway, so there is not much to be gained hereā¦
e.g.: Leitfaden zur Bildsegmentierung | Google AI Edge | Google AI for Developers
The intention was not to gain something on the tracking side, but to use a highres texture from the webcam in vvvv without the NDI/Spout bridge.
This technique works perfectly when setting the texture size and then restarting the patch, found out thanks to @levoxtrip
https://forum.vvvv.org/t/how-to-scale-mediapipe-image-to-full-renderwindow/
1 Like