You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I've been using the snap release version with success. The GUI is simple and easy to use. The algorithm gives my video wobbly around the person and several artifacts are common outside of the persons shape. I did try the --beta install and did not have any luck out of the box getting it to run, might look into it a little more.
While I was looking at what other packages are available I found https://github.com/fangfufu/Linux-Fake-Background-Webcam. This repo has some nice enhancements to the issues with edge detection I've see on the release version here. They also added a foreground mask feature to allow you to stand both in front of an image and behind part of that same image (such as a podium).
Anyways, would be a potential effort savings to have smoother edge detection, frame rates, and foreground feature by using the snap package and GUI wrappers created here and wrap around the code developed on the Linux-Fake-Background-Webcam repo.
Any extra information?
No response
The text was updated successfully, but these errors were encountered:
I would love to see better edge detection with Fakecam. The camera connection seems spotty, too, with long delays in starting the video or even sometimes not starting at all. With an AMD GPU, I can't use any of the CUDA-based solutions, and Fakecam is the only solution I've found, so I'd love to see improvements and ideally, a non-snap build as well.
What changes would you like?
I've been using the snap release version with success. The GUI is simple and easy to use. The algorithm gives my video wobbly around the person and several artifacts are common outside of the persons shape. I did try the --beta install and did not have any luck out of the box getting it to run, might look into it a little more.
While I was looking at what other packages are available I found https://github.com/fangfufu/Linux-Fake-Background-Webcam. This repo has some nice enhancements to the issues with edge detection I've see on the release version here. They also added a foreground mask feature to allow you to stand both in front of an image and behind part of that same image (such as a podium).
Anyways, would be a potential effort savings to have smoother edge detection, frame rates, and foreground feature by using the snap package and GUI wrappers created here and wrap around the code developed on the Linux-Fake-Background-Webcam repo.
Any extra information?
No response
The text was updated successfully, but these errors were encountered: