Open eirikhollis opened 4 years ago
Is this Desktop or UWP?
In Unity, so desktop?
Yes, Desktop x64 then. I ask because the codepaths are quite different internally.
On Desktop, the video capture is handled by the Google implementation directly. And indeed there a bug in it, already identified by their devs it seems: https://github.com/webrtc-uwp/webrtc/blob/baa89682df1ab891ae274a1cb2d9be3b1ee7e20b/media/engine/webrtcvideocapturer.cc#L138-L150
This is annoying because there's no easy way currently for us to fix that. We will have to either backport a fix or upgrade to a later Google release.
Note this seems fixed on Google's HEAD, if I am not mistaken in thinking that VcmCapturer
is the replacement for the current WebRtcVideoCapturer
on M71.
https://webrtc.googlesource.com/src/+/refs/heads/master/test/vcm_capturer.cc#42
So kinda clunky title, but I'll explain.
I'm trying the new "multiple videotracks per peerconnection" feature, and I have two webcameras that are of the same model,
Logitech BRIO
. When runningPeerConnection.GetVideoCaptureDevicesAsync()
both cameras appear and are recognized as seperate cameras. However, when settingPeerConnection.LocalVideoTrackSettings.VideoDevice
to either one of them, only one of them is used. If I try to enable the first, then the second, in the same peerconnection, the app crashes, which I think is because it tries to access the first camera two times. I have a third webcamera which has a different name,Microsoft® LifeCam Studio(TM)
, and this is recognized and utilized correctly in conjunction with either of the two Brios. I'm assuming then that the WebRTC code checks the name of the camera instead of the ID? That would at least explain my findings.