I’d like to show my support for independent mask lines. If we had this feature we could implement proper band-limited scanlines (via lookup). Not band-limiting them is why scanlines look bad at 4.5x scale. If you use a 9-entry table in a 4.5x mode scanlines can look good at couch distances.
Here are the weights I use in the shader I wrote for emulators. 51.5% 79.3% 100.0% 79.3% 51.5% 62.5% 94.2% 94.2% 62.5%
The order has to be flipped if the first line is 4x and the second line is 5x.
The formula is cos((x-2)*pi/4.5)^2 * .5 + .5
The -2 is to adjust the phase so that the 100% coefficient is in the middle of the group of 5 lines not the group of 4.
The 4.5 corresponds to the ‘X’ of the image scale.
The two .5 on the end make 50% scanline intensity.
(Yes, I know that cos^2 is only an approximation of the “real” scanline formula.)