Copy link to clipboard
Copied
This is a feature I've been wanting for years, but as far as I know, it still doesn't exist. The idea is to sort of "green screen" without a green screen by detecting a steady backdrop and then detecting any change to that backdrop, effectively masking anything that changes or moves in front of the backdrop.
I think this might be how programs like Zoom handle background replacements, but I don't know for sure. Does anyone know a way to make a matte automatically like this in AE?
Copy link to clipboard
Copied
This is a feature I've been wanting for years, but as far as I know, it still doesn't exist. The idea is to sort of "green screen" without a green screen by detecting a steady backdrop and then detecting any change to that backdrop, effectively masking anything that changes or moves in front of the backdrop.
I think this might be how programs like Zoom handle background replacements, but I don't know for sure. Does anyone know a way to make a matte automatically like this in AE?
Copy link to clipboard
Copied
There are dozens of ways to create a procedural matte in After Effects. The most efficient workflow depends entirely on the shot. The automatic replacement of backgrounds uses AI to identify the background because it knows about how far the person is from the camera, the camera isn't moving or if it is, there is accelerometer data available, and the results are not nearly good enough for post-production compositing work.
Show me a shot and I can probably point you to an efficient solution. A magical background replacement plugin is probably going to take a while and a fairly large increase in AI technology.
Copy link to clipboard
Copied
You can easily build your own motion difference keyer based on Revision FX' RSMB Pro using the vector generation effect. That being said, unless you are working with locked shots, it's far from trivial to discern foreground and background entities. You may have motion, but the effects have no concept of scene depth. In contrast to Zoom you also can't get by using a realtime deconvolution algorithm based on image sharpness, too. nobody would accept that quality in post-production. So for what it's worth, it looks simple on paper and it sure is a pipedream I also had for years, but once you research the details a truly intelligent motion keyer is a whole different story.
Mylenium