Fri Sep 23, 2022 7:11 pm
I can expand on the behavior a bit. When the mask is in default mode, its size is usually inherited from whatever it's plugged into. If that node doesn't calculate its own dimensions for some reason (you've plugged the mask into the main/yellow input of a blur or a transform, for instance), it will fall back to the comp's settings. Assuming that whatever you use the mask for matches the comp's preferences, all will be well.
If, however, the masked image is not the same size as the comp's prefs, strange things can happen. The mask may adapt its size correctly when you first plug it in, and as long as you're looking at the comp on the Fusion page. When you leave Fusion, or submit a render to a render farm, or otherwise get it to request a frame in a non-interactive mode, the mask may actually process before the downstream node it's supposed to query for its size. So it will fall back to the default in that situation even though it doesn't when you're actually looking at the comp to debug it. This creates much confusion.
So the solution when you have masks that don't behave like they should, or change their behavior when a comp is closed and then reopened, is usually to set their resolution manually.
There are, however, some other less likely possibilities, which is why I asked for more information.
Bryan Ray
http://www.bryanray.name
http://www.sidefx.com