Orin Nanos are able to see IR cameras on webcams such as the Logitech Brio. Because of this, they will actually grab the IR camera instead of the color camera, which produces not-so-useful images. Interestingly, Xaviers do not see the IR camera, so it's not an issue there. Sunil also reported this issue on a Raspberry Pi in https://github.com/groundlight/framegrab/issues/26.
On Orin Nano:
$ ls /dev/video*
/dev/video0 /dev/video1 /dev/video2 /dev/video3
On Xavier:
$ ls /dev/video*
/dev/video1 /dev/video2
To fix this, we will skip an devices that have a name of "Logitech Brio" and produce grayscale frames. This is kind of a special case fix, but it seems like the best thing to do for now. If we discover any other cameras with this same behavior, we can update the list of devices that are known to have IR cameras.
Orin Nanos are able to see IR cameras on webcams such as the Logitech Brio. Because of this, they will actually grab the IR camera instead of the color camera, which produces not-so-useful images. Interestingly, Xaviers do not see the IR camera, so it's not an issue there. Sunil also reported this issue on a Raspberry Pi in https://github.com/groundlight/framegrab/issues/26.
On Orin Nano:
On Xavier:
To fix this, we will skip an devices that have a name of "Logitech Brio" and produce grayscale frames. This is kind of a special case fix, but it seems like the best thing to do for now. If we discover any other cameras with this same behavior, we can update the list of devices that are known to have IR cameras.
Once this PR is merged in and released to pypi, we can update the RPC server here: https://github.com/positronix-ai/rpc-server/pull/17