Face/body detection filter
I am currently trying to use SnapCam to detect which camera I am in front of so that my OBS stream can automatically detect where I am and switch what it shows based on my position. I intend to use this for recording lectures, as I regularly switch between my computer (showing slides, examples, etc.) and a whiteboard, which I need to stream through a camera.
I know essentially nothing about SnapChat lenses, but I am currently using Green Smileys and detecting if there is any green on the camera.
However, I don't necessarily face the camera for long enough for SnapCam to apply its filter, so I don't know how reliable this would be. Also, I appear to be forced to use rather low resolution images (I can only have one instance of SnapCam open at a time and only one SnapCam driver, so I am putting the two images side by side), so its range is not satisfactory.
Is there a filter that can detect if a person is present and cause bright green to appear on their body if they either facing the camera or facing away (or any other reasonable orientation), or apply the shade of green to movement? If it doesn't exist, how hard would it be to make?
I don't believe there is an existing filter which does what you are looking for. You could create your own using the 3d body tracking and some scripting to make the green appear based on which direction the person is facing.
I ended up creating a filter that does what I want. It uses face detection to detect a face and play a gif of a moving green square over the face. If it doesn't see a face, it will try upper body detection.