I’m looking for a light weight way to implement chroma/color keying. The included component is wayyy overkill for my needs (I just need source 1 + 2) and grinds my project to single digit framerates.
I followed a youtube tutorial that used a clamp to grab a x,y coordinate of an image to find the color key, as well as generate a matte with blur to soften the key edge. The tut’ was using a still image and I’m using video. My solution was to import a ‘color picker’ gradient with brightest values at the top and darkest at the bottom to be able to pick a color in similar manner and not have it jumping all over the place. This solution gets me close, but the control parameters within the key don’t seem to be as flexible or forgiving as I need. Even running two or three of these keying systems isn’t burning through the CPU like the ChromaKey component, but I’m struggling over how to implement better cleanup and control.
I thought of stripping down the included ChromaKey, but my head spins just looking at it, and I’m much more likely to break it than fix my performance issues.
I had another near-solution, which was running two Syphon outputs to OBS, and doing the keying there, but the layers weren’t consistently updating/syncing. I’m trying to process someone’s 4k footage in realtime, doing effects in TD, and trying to retain picture quality by recording source size and bitrate in OBS.
Sorry for the ramble, but perhaps I’m missing something entirely. I basically need to do feedback effects in place of skin-tones. Easy-peasy, right?