It’s definitelly because of more tolerant horizontal mask distribution. If mask zoom would allow fractional value setups, then this could sync even better with sub-pixel patterns on non-native resolutions.
Non-integer mask widths are indeed a thing if a mask is big enough, like above 5.0. This is more or less a 4k thingie regardless…
Is there a way to get 100% strength just on the black pixel, using RRGGBBX? Or independent adjustment of the black pixel? Could allow for a much more realistic slot mask effect, I think.
edit: also, I have to manually set the slot mask width when using RRGGBBX to 7, not sure if that’s correct behavior.
New Release Version (2025-11-30-r1):
Notable changes:
Download link:
https://mega.nz/file/QoI0wAoS#z53_-BJN5hw9QieusZzhjDAbeuYJqL4gCvBfsfnpFv4
Quick demo using an overlay to illustrate what I’m talking about, but I can’t get it to line up with the triads correctly- it’s like there’s a single pixel border at the edge of the screen, so the overlay gets shifted by a single pixel (is this something the shader is doing, or do I need I need to check my video output settings?)
No overlay
Add overlay, darken every 7th pixel. Mask low/high 75/100, Overlay 75
The overlay helps to preserve the slot shape and just looks nicer in general (especially closer to the screen). Unfortunately, it’s not lined up correctly with the triads (shifted by 1 pixel horizontally). Overlays are annoying, it’s better to do this with a shader 
Currently, the vertical black lines and horizontal black lines that comprise the slot triad get altered by different amounts when you adjust brightnesses, blooms, glows, mask strengths, etc. For better looking and more accurate results, the horizontal black lines and vertical black lines comprising the slot should be of equal strength. On a real CRT, they’d always be 100% black no matter how bright the phosphors are shining, but we have to mitigate this on an LCD. We can get better results if the black lines defining the slot triad are all of a roughly equal and consistent strength.
This might require an additional “mask” to be added, and would work with those masks containing a black pixel (MGX, RGBX, RYCBX, RRGGBBX). The code would simply be: darken every 3rd, 4th, 5th, or 7th pixel by 0-100%.
I think I got a solution, which would only complicate slotmask and lottes mask 2 combos.
With this solution slotmask strength controls are managed by mask strength controls instead. It would also save one parameter. Existing presets should not be affected at an disturbing level.
To summarize, instead of slotmask strength controlls there would be a slotmask ON/OFF switch, slotmask strength = mask strength.
I hope this would also improve the issues you are mentioning.
It’s worth a shot!
I found something else that might be improved:
What would be nice is something closer to this:
EDIT: Ok, so mask 12 is actually XRRGGBB (Not RRGGBBX as I had assumed)
No biggie, but that’s what was throwing me off. I’ve got the overlay working now. Here’s what I’d like to accomplish (without the use of a clunky overlay):
Why is the red phosphor wider than the green and blue phosphors in this example?
I get that you would like to get this same result via the shader but without the “clunky” overlay but do you realize what you’ve done here?
You’ve basically perfectly abstracted the mask and slot wires for CRT Emulation. As a resut of this all of the vertical and horizontal slots are perfectly aligned and there are no compromises or ambiguity when it comes to at least the Mask. what’s probably missing is if you could have added the vertical lines between the phosphor colours themselves but I’m not sure if you have enough resolution for that at 4K since I’m assuming that you’re already at 1 subpixel wide for the black mask outline and horizontal slot.
The preset looks amazing at least when not zoomed in. We just have to do something to improve the realism of the scanlines and the phosphors.
Is that overlay your custom creation? Am I right to assume that it only works at a specific scale factor?
Great job on the preset so far though!
https://mega.nz/file/hYwyUByR#iEZoE1tcPPm8nFEqgc6S1lWGRZ4ZDvynKN0QPKw2gX8
Please excuse the jpeg compression artifacts in all of the screenshots.
Also, if we could have gotten rid of one of the vertical black mask lines then we wouldn’t have the double line when tiled and that should also help in the brightness department.
The phosphor itself isn’t wider, it’s just bleeding into the black spaces of the slot mask which makes it appear that way. I think it’s a decent way to compensate for the brightness loss. You can also see this happening in this example:
It’s an easy pattern, but it will only work with XRRGGBB at 4k This could be scale-dependent, I don’t know. I’ve only tested it with Nestopia, Core Provided, Overscale
That’s an interesting pattern, right off the bat the vertical lines separating the slots are a bit thick, though. The noise(?) is an interesting way to increase brightness.
I’ve been compromising a bit on that lately with my local dimming setups. Once you get used to local dimming, it’s really hard to go back.
That’s just a cropped section of larger image; the pattern itself doesn’t have a double black line (as you can see) 
This is 308 TVL already with a mask width of 7 pixels, to emulate the black space between phosphors would take at a minimum 14 pixels per triad if you want the proportions to be accurate (RRRXGGGXBBBXXX). So you’d need 8K resolution for the same TVL.
Yes, they are and I only checked and have now noticed this because of your example. So in my case, this mask could probably benefit from a little trim if possible.
There was a particular slot mask preset combination that I had really liked, except for one thing, I used to see some light vertical lines that looked like a kind of moire pattern or artifact. It’s possible that it could be related to this double black vertical line.
It could be noise that you’re seeing but it could also be the jpeg compression artifacts that you’re seeing in addition to the noise. In general the level of noise that I typically use in my presets is barely noticeable but I did struggle a bit with the parameter range/sensitivity when I initially added the grain shader from the Image Mod shader to my shader stack and combined it with XBR-LEVEL-2.
For many of my presets, I manually tweaked them to where I liked the noise but some remained unchanged.
All of those pics are from lossy originals. JXR uses JPEG, then I convert JXR, which is relatively high bitrate to jpegs which fit within the 4096KB limit for sharing to the forum.
The PNG is a 3rd Gen of the original JXR.
Looking at the regularity of what you described as possible noise, I think at least some of it might be the scanlines gaps as well. So noise + scanline gaps + compression artifacts.
Slotmask and scanline alignments are tricky to do, since there are the usual modulo missalignments. From my experience with a slotmask based CRT TV the scanlines were quite mild, not comparable with a PVM etc.
There is also the usual issues with curvature, odd integer scaling (symmetric slotmasks are even). To cut it short it would make a lot of mess to cover all the original and viewport resolution combos, not even considering other possible circumstances.
As a mitigating fix to these situations there is an slotmask option to round things up a bit.
The only option i can think of right now is to allow slotmask vertical shifts. It also took much creativity to get a proper slotmask - mask alighment on various gpu drivers because of rounding discrepancies.
Here are some real-world comparisons of Composite and RGB showing the infamous “m” as well as showing how RGB messes up the nice shading of gradients and has a more aliased appearance for very little gain in my opinion. It also shows how well colour saturation can hold up in Composite vs RGB and how smoothly mixed the composite colour artifacts and fringing are spread across the Image.
In my opinion these hardly take away from the Image while adding a bit of character, charm and warmth.
I’ve also recently learnt that these composite colour artifacts as well as loss of sharpness and clarity of especially on things like fonts are less visible in real-time so pictures which attempt to capture them should be taken with a pinch of salt.
https://www.reddit.com/r/crtgaming/s/dlWY2QBsBv
I also find it interesting that 3-Phase with Merge Fields - On takes the Composite signal even closer to RGB.
Yes, at the disposal of temporal effects, you can use more pronounced artifacting values to compensate.
Not yet.
Wow! Really?
Not sure if things have improved but I wouldn’t be surprised based on @guest.r’s rep.
The real challenge isn’t isolated to getting that “m” to look right though. It’s getting it to look clear and legible with minimal noticeable flickering and strange artifacs while at the same time getting all checkerboard dither patterns to blend into new shades/colours without the flickering giving away the illusion of more native colours.
When you increase Artifacts and other settings which improve the clarity of the font, they either sharpen the image and reduce the blending effect or increase flickering/noise/movement in the blended areas.
Reduce those settings and the font get’s more blurry and blended. It’s because the “m” in the font itself is not much different from the vertical line dither pattern that’s used in the Sonic The Hedgehog waterfall but with fonts certain characteristics can be used to differentiate them from dither patterns used in sprites and backgrounds that we actually want to remain blended so that’s where @guest.r’s clever algorithm comes into play.
So I’m definitely going to try it out when I get a chance.
This could also be a factor:
#reference "shaders_slang/crt/crt-guest-advanced-ntsc.slangp"
cust_artifacting = "0.800000"
cust_fringing = "0.000000"
ntsc_phase = "5.000000"
ntsc_charp = "10.000000"
ntsc_charp3 = "10.000000"
ntsc_cscale = "3.299998"
ntsc_cscale1 = "1.699999"
ntsc_sharp = "10.000000"
ntsc_fonts = "1.000000"Just sharing these to provide a little more insight into some of the things we have discussed concerning Font Preservation.
Unfortunately they’re in the wrong colourspace/format or something. That happens sometimes when using shadow play to capture HDR images. Sometimes it’s not enabled and you have to either restart or disable then reenable it.
Anyway, corrected screenshots can come at a later date. These screenshots show that XBR-LVL-2 settings might have been interefering with the font preservation process and contributing to the artifacts that I was seeing.
All of these tests were conducted with ver 2025-11-30-r1.
Note: 2025-11-11-r1 didn’t have any issue with XBR-LVL-2.
I realized I can just use a vertical line overlay and get the same or better results. The overlay is still a fiddly and less than ideal solution, obviously, but it illustrates the concept.
Sometimes it just works, other times I have to turn integer scale off and adjust the custom aspect ratio width one pixel at a time until the overlay is aligned with the triads correctly.
overlay 70%
I have only been able to get this overlay trick to work with XRRGGBB, I’m not sure why.
XRGB, XRYCB don’t work, it’s like the overlay gets blended by the shader (even though this shouldn’t happen). I’m stumped.
Why does “Internal Resolution” on the HD shader max out at 8? Shouldn’t this match “Internal Resolution Y”, which caps at 10? For example, if I’m adjusting for 384x216 on a 3840x2160 monitor. I’m using the ReShade port. Is it the same in the Slang version?
Hey! Nice for stopping by. With 384x216 you are good with slightly increased internal resolution. You would only need tremendous ammounts of Internal Resolution if you exaggerate input resolution beyond reasonable means. Monitor resolution is not connected to the a nice looking setup regarding Internal Resolution in it’s basic way.
For example, you run a 256x224 Mario game at 8k. Your preferred target Internal Resolution would still be arround 1.0.
Large Internal resolution is basically needed in you use the shader on point-upscaled graphics, where one original pixel is a rectangle of 8x6 pixels. Then you would need “Internal Resolution”. But usually a more elegant way is to adjust the input x and y resolution.
But, it you have some special configuration going on, you can still edit the shader and increase the UI parameter values for Internal Resolution how large you like it.
