Hello,

is it possible to avoid simultaneous scaling of my custom alpha channel mask (Key/Fill) with my main input scaling. I've tried everything that I could think of. Using virtual sets, or multiview overlays, which does not work as the mask is applied under overlays. Virtual inputs copy original input scaling, thus scaling the mask as well. Only thing that works is outputting my scaled input on one of 4 outputs as NDI and then adding that output as NDI input, but this introduces video latency, which is unacceptable.

If the virtual input could be precomposed, this would solve the issue. As precomposed, I mean that the original scaling or overlays are prebaked (can't think of a better term) in the virtual input. Simmilar to the NDI example i gave above.

To give some context as to what am I trying to achieve with this. Lets assume I have, a virtual set with a virtual screen on the set. On the screen i have a list of videos that play. The floor is visible and glossy and I would like to create a fake reflection of the screen on the floor. Positioning and overlaying the video reflection on the floor as a virtual input is not hard. The reflection should be partially masked (e.g. to mask out the table, or something non-reflective on the floor, or applying simple feather). The problem arises when my mask scales with the reflection scaling. I would like it to be applied to the scaled and rotated reflection without the mask following that same scaling.

Any ideas how would I achieve this?