@mas The screenshot collection looks great!
When posting Sony Megatron Color Video Monitor screenshot it might be a good idea to include some viewing tips, for example, brighten screen or zoom in until it looks good for users who might not have calibrated displays or they might just see them and think too dark and move on.
These look like they’re from my Megatron NX preset pack, am I right? Which presets if you don’t mind me asking?
I will reupload them, when I get home today.
I think, it is a PCE Bad Composite preset with some little changes.
Raw
Guest Advanced Fast, with scanlines only: With mask only:
Mask + scanlines on:
No composite, example of how patterns transform also with RGB.
New CyberLab Megatron miniLED Death To Pixels Shader Preset Pack reference video!
Still tweaking Presets-ng/Monitor-Screen_Hmask-Screen_SlotMask_Taller_Brighter.slangp
The intent is to preserve as much slotmask as possible at higher brightness:
Seems pretty good already.
What are the minimum requirements for this preset to look as intended? Would a standard 1080p screen be sufficient?
Yes definitely; 1080p is my target since that’s what I use 99.9% of the times.
Edit:
I tested it deeper, and it seems to be a lucky one with regard to moire.…
After a few weeks with EGA games, I went back to SVGA territory.
I think scaling 480p to 1080p is greatly improved by your shaders. They are amazing tools really, thank you so much! For Warcraft 2 in example here with a preset for CRT-Guest-advanced, I would say it looks even better than the remaster.
The goal was to even out the upscaling and smoothen the uneven chekerboard of the dithered fog of war and various shadows. Also get some color correction and very thin scanline and mask for texture.
Here are the raw pixels :
Here it is with the shader :
You can zoom in on this section for the fine details. There’s still a barely noticeable moiré effect that will completely disappear with integer scaling at 1440p or 4K, but 1080p is looking good IMO.
I also played through the cancelled win98 Warcraft Adventures point and click with the same shader, you can watch it in motion here :
@RetroGames4K, your shots are always an inspiration and a challenge 
Also:
Screw the mask! What if we could just paint it regardless of any screen alignment rule, let’s just draw it and see what happen. curve everything!
Results expectedly look like s**** if you look too near, but surprisingly they give a nice effect from a proper distance.
…of course it happens that there is moire everywhere, just like it happen in photos, so here is a less curved version with some mitigations to the mask; also no green casting and properly more saturated:
-EDIT- Maybe this is ok too, letting just the “rgb” mask to bleed more:
Thank you, I’m always learning and trying to improve stuff…
An experimental shader i did some time ago (never uploaded), wanted something soft but not blurry, more how a 1084S would look like i suppose.
Well, everytime I look at a good shot of a screen, there’s something into it that suggests me that it is real and accurate, makes me wanna play with it.
So I tried to replicate the way it looks at 1080p… do you remember [this]?(Please show off what crt shaders can do!):?
More or less the same, but this time I tried to replicate a mask that is even more “compromised” by the photo and by its downscaling to 1080p.
Still my brain feels it to be right.
Usually shaders, to avoid artifacts or exploit its interaction with real lcd subpixels, draw a straight steep mask aligned to them, which is nice; but then they either take the Megatron path, requiring your display to mitigate it, or assume your display is too dull and take mitigation measures by themselves (draw a proper mask then mitigate it).
This time I just tried to draw a mitigated mask from the beginning, this means it is spreaded and badly defined as you see in photos of real screens (lately added deartifacting code helps with the slotmask); it turns out the effect is convincing.
Also, is curved, (and I think it contributes alot to trick the eye) and indeed there is moire that needs to be mitigated; exactly like in real crt photos; fortunately there are countermeasures for that.
Those screens are a proof of concept ( the slotmask dividers don’t even align properly to the mask width); I want to implement an addition/change to a parameter so that the user can target a specific TVL to draw a fully curved mask and then change how much the mask should be “in focus” .
Well this is basically similar to the premise behind this video that I made. I didn’t mind that I was using a 1080p camcorder to record an almost 4K CRT-Shader preset. When recorded its supposed to look similar a recording of a CRT. You don’t need HDR to demonstrate how a CRT looks in a recording either. The whole HDR and extra brightness thing is about mitigation of brightness loss when trying to simulate a CRT using very dark colours on the subpixels to make up the mask plus lots of black to make up the mask wires or slots and the scanlines gaps. The end result of this after we apply such a shader to a 1,000+ nit HDR LCD display is something that resembles a 150nit CRT. We don’t need HDR or 4K to capture and show that.
Look at the description and you’ll see.
So from a practical perspective this makes more sense than posting super dark, raw screenshots then expecting the average user to brighten their screen or view in HDR to apply the same darkness/dimmness mitigation in order to view them properly.
At some point, I can try to use a 4K camera to improve the fidelity and detail of my capture but it really isn’t necessary to give a good idea of what the shader can do while bringing a nice recorded CRT feel to the equation as well.
If you look at my more recent slot mask presets compared to my presets from more than 4 to 6 months ago, you can see that many of the older ones have some weird intersections and misalignment between the slots of the Slot-Mask and the scanlines.
It looks horrible but I didn’t know how to resolve at least one of the issues until I played with the Vertical offset of the Mask grid and was able to get everything more centered on the scanlines
This still left the weird looking situations where there was a horizontal slot that was closely above or below the scanline gap. I solved that issue by increasing the height of the scanline gaps and inversely decreasing the height of the scanline so that the aligned slot mask wires and can seamlessly blend into the scanline gaps.
I’m only left with a slight gridlike bright vertical line pattern in some of my slot mask presets that I don’t remember being a characteristic of a real CRT.
Even this is something I had to do for my recording as when I initially set things to be on focus the moiré patterns were very heavy so I dialed back the focus slightly until they disappeared.
Of course this resulted in a softer video which was not fully representative of what is seen in the preset in real-time so I added a little sharpening in post to help make it a little closer to how it might have looked.
Aesthetically both the more sharpened and less sharpened versions have their place but I would really like to record a video in focus just showing how the camera sees it and also take the opportunity to do some close up shots which will mitigate the lower resolution of the camera and be able to show the subpixel generated “phosphor” detail.
That is more or less what crt-royale does. Its mask method isn’t exactly aligned to the content. It uses tiles, which get misaligned between passes. On top of that, when I was adjusting the slotmask for crt-royale-fast, I figured out that a misaligned mask was more pleasant to the eyes, so I turned it the main slot in royale-fast.
Here’s how it looks (presets/crt-royale/crt-royale-fast-rgb-slot.slangp) at 1080p:
Yeah, triads can be misaligned in Royale too, but even if I’m not expert as you with Royale tweaking, from a quick look I think there are some key differences.
The first is that the the triads are not curved:
And the second is that apparent phosphor size width is fixed, probably because it samples them from a lut:
So it still uses the way: “draw a proper mask, then mitigate it.”, while my approach/idea is “draw a mitigated mask”, directly, where the phosphors glow is represented by spreading in width themselves.
See, this is with a ridicously low TVL to better show the effect (and unhandled defects due to performance reasons ofc):
Ok, now I understand your approach. In fact, a new way to solve this. 
That looks pretty damn accurate