In my examples above if I leave Bezel Independent Scale Off and leave Integer Scale Mode Offset on 0, I still get the anomalies. It’s only when I increase it to 1 (or alternatively step up to the next Max Integer Scale Height % stop) that everything seems to render cleanly in all SNES games I’ve tested so far.
If I try to do this while leaving Bezel Independent Scale Off, too much of the Bezel/Frame ends up off screen and it doesn’t look nice in my opinion. In the case of my presets which include graphics, I at least wanted the top of Bezel/Frame to be aligned with the top of the graphics and for the Bezel/Frame not to overlap with the graphics which are already at the extreme horizontal edges of the screen so there’s not much room for the Bezel/Frame to grow without some overlap there.
As mentioned, this power bar rendering anomaly doesn’t seem to occur when using all mask type and size combinations.
Some of them look completely fine without increasing the Integer Scale (and possibly even using Non-integer scale).
In my examples above the issue is present even with Curvature turned off.
In my further testing, I’ve observed a strip where there’s overlapping going on between different layers behind the bezel which is increasing the brightness. It’s not just reflection but also the non-mirrored edges of the viewport being filtered/blurred and “shining through”. This seems to be mixing with the reflections which are stemming from the reflection plane, which starts at the edge of the content, even if the edge is supposed to be obscured by the bezel, (which is possibly incorrect behaviour) but that plane is not lining up with the inner edge of the tube/bezel or otherwise scaling properly.
I’ve observed some further strange behaviour where when using Bezel Independent Scale and some of the image is cut off, if I reduce the Integer Scale Offset or even use Non-integer Scale the reduced image, part of the image that was off screen/beneath the Bezel is still missing.
When I tried the new Show Pass debugging tool, I think I was able to see the strip of overlap when showing either Pass 2 or 3.
Adjusting the Inner Black Edge Thickness seemed to have an effect on the strip of overlap.
If Bezel Independent Scale is not used, everything “seems” fine including where the reflections stem from.
I say “seems” because I haven’t really gone back and scrutinized the reflection with Bezel Independent Scale Turned Off to see if there might be other less obvious anomalies which I might have previously missed.