Maybe this is part of the reason you were having trouble with your display mapping properly with the existing subpixel layouts?
It should be possible to retrace my steps with what I did to get the RRBBGGX available in Sony Megatron Colour Video Monitor to find a compatible mask layout for your particular display. All it would take is a little trial and error, to test the various combinations of subpixel colour order until you get one which lines up with your display.
Congratulations and thanks @guest.r, @HyperspaceMadness and the rest of the Mega Bezel team! It has definitely been an honour!
Hey there! I’ve been using this shader for a while now and I absolutely love it.
I pair it with ScaleFx because I want to slightly scale the image while keeping that pixel art feel, just less aggressively.
After a lot of tweaking, I haven’t been able to achieve a uniform 2x horizontal and vertical scale. The closest I’ve gotten to what I’m looking for is the image I’ve attached. In that image, if I’m not mistaken, it generates a 1x horizontal (cyan circle) and a 3x vertical (orange circle) scale.
I’ve tried other scalers with GuestFx as well, but no luck. Does anyone know if what I’m trying to do is even possible?
Hey! Nice for stopping by.
Regarding ScaleFX, it’s natively scaling to 3x in both directions, so it’s best to honor the original scaling or some kind of interpolation artifacts are happening.
You can simply achieve this by pre-pending the original ScaleFX preset from
shaders_slang/edge-smoothing/scalefx folder.
If you are still after 2x scaling, then the process can be a bit different. It’s best to replace one of the stock passes with a shader file (not preset). Good candidates are xbr and xbrz-freescale.
When the stock pass is updated, you can set the involved pass scaling to 2x, or you simply save the preset and use a text editor to set the pass scaling.
It sort off looks like this:
Nevertheless, i recommend 3x scaling even here, more scaling gives crisper image.
If you want 2x scanline resolution, then it’s best to set Internal Resolution Y: 0.5...y-dowsample divider to 0.5, works even with 3x scaling:
Wow! This looks lovely! Maybe it’s time to go full circle again and try to get some better enhanced looks. I’ve been challenging myself for a while to get great balanced looking images just by playing around with the NTSC Resolution Scale and stuff like that and I’ve gotten pretty decent results.
The 2X is definitely more my style.
Great shaders for sure, with lots of features. Any chance of an updated readme or documentation to learn of all the parameters’ options?
Will be in the works, but no ETA atm.
Is was going to say… but thought it’s better to wait for a better time 
.The old Guest Venom had some very nice documentation that was maybe even a bit to extensive, but since then there’s a lot of parameters where it’s just guessing, but trying out without any background info and how things interact is probably not very fruitful for realizing the full potential.
New Release Version (2025-06-09-r1):
Notable changes:
Download link:
https://mega.nz/file/Y9xhmRAD#FFUd-YkQY-PWRDf2KwROFe1XOkG_f0JeLhjLj31XB24
Which of the scaling shaders is used in this example @guest.r? ScaleFX, XBR or XBXZ-Freescale?
For this 2x upscaled image xbr-lv2-standalone.slang was used. It also works with no-blend mode, which is nice for crt usage.
Late for this update but just grabbed it and did a quick spin on it. Here’s with deblur off:
Here’s with deblur at 3.00:
Still does that “pruning” thing. Maybe that’s just how that setting is supposed to behave, not sure.
Also here’s RGB for the hell of it:
Hey, whats about this Sparkster pic??
I made a shader “HD TV”
for Retroarch with combination of default slang shaders please check these screen shots. If anybody is interested in it than I can share it with you for more improvements. This shader is With scanlines and geomatry. But I don’t Like Without scanlines and Geomatry.Something I’ve wondered for a while, are there any advantageous scale factors when doing noninteger scaling, or is it just “the more resolution, the better”? So for example scaling 3.5 isn’t going to be better then 3.8 if the resolution is available (mostly thinking vertical res because of the scanlines)?
More resolution should be better. From my experience scaling factors below 4.0 aren’t this good, but like 240*4 = 960, so 1080p users should be fine with most content.
800p users could use the no-scanline mode or very mild scanlines (800/224 = 3.57 for example).
One more thing is quite important with integer scaling, namely if the vertical scaling factor is odd or even. Odd scaling factors can have better shaped scanlines.
Yes, I was doing some testing and it seems 3x is the factor where the most significant artifacts disappear, but it’s still not quite enough, at least with low resolutions and notable scanlines.
I’m surprised that odd scaling is beneficial, I got the impression that masks work better mostly with even.
Odd or even integer vertical scaling factors are mostly notable with 4x vs 5x. If i summarize it a bit:
4x scaling:
1
2
2
1
5x scaling:
1
2
3
2
1
6x scaling:
1
2
3
3
2
1
7x scaling:
1
2
3
4
3
2
1
etc.
With even scaling factors there are two “same scanline strength” pixel rows in the middle of the scanline, making them look more “flat”.
Ofc. more resolution is better in general, but it’s an explanation why a possibly desired scanline shape eludes with even scaling. With 4k this shouldn’t be an important issue, at least not with most low resolution content.
Maybe this is a case for slot and shadow masks, which indeed like flatter, milder scanlines better. Aperture masks should definitelly prefer odd integer scaling, because it’s easier to generate better mask falloff effects (a.k.a. scanline saturation with guest-advanced).