Sony Megatron Colour Video Monitor

[Cyber]

@Nesguy

That’s been working best in my case, as well. It was the brightness setting causing the clipping.

Excellent, I’m glad you sorted that out.

@Nesguy

Would this be the “CRT” gamma?

Yes, most times I’m not in front of the menu when posting so the terms might not be perfectly accurate plus I think I read @MajorPainTheCactus refer to it as “Gamma In” once.

@Nesguy

a bit of “digital vibrancy” or something that’s missing from the image, even after setting everything up correctly.

Hmmm…have you tried prepending the decoupled CRT-Guest-Advanced-NTSC or even using my pre-integrated version from my preset pack?

I have presets which use 3 variations, Stock Megatron, Stock Megatron with CRT-Guest-Advanced-NTSC integrated and Stock Megatron + CRT-Guest-Advanced-NTSC + Super-XBR integrated.

You can do a lot to the image using a combination of the NTSC effects as well as the Chroma/Blending and various Sharpness settings.

I only used the Original decoupled CRT-Guest-Advanced-NTSC which doesn’t include the Horizontal Filtering Section but both should be available if you look back in the thread.

I also did some tweaking to the Shader/Scaling settings in order to have the NTSC Component work even better.

Perhaps after you try these out, you can balance higher Saturation Levels with Gamma. You might see nice results if you start high with the Gamma (Darker, then gradually work your way towards brighter settings).

Don’t be afraid to push the Saturation. You can use the 240p Test Suite to confirm if it’s causing clipping.

If you change TVL you have to readjust everything by the way.

Feel free to take my presets for a spin.

I’m really looking forward to seeing things continue to improve with this shader, especially what I would consider the fundamentals but we would need more people like yourself to stick with it and identify the little outstanding bugs here and there and point out where might need some improvement.

[Cyber]

Slot Mask - Deconvergence Off, Zoomed In, Cropped

@MajorPainTheCactus, @Azurfel, @Nesguy I’ve been analyzing the above image and what appears to be causing the uneven scanlines issue in this case is the differring interaction between the Odd and Even Scanlines with the horizontal slots of the Slot Mask.

When the Scanlines align perfectly with the Slots, the scanlines appear dark and sharp.

When the Scanlines align with the middle of the Phosphor triad and the Slots are close but above or below the Scanlines, the scanline appears less intense and less sharp.

What might be needed is some sort of mitigation baked into the Shader to ensure that the conditions which produce the undesirable artifacts do not occur.

So probably a vertical Offset to the position of the entire phosphor grid (or all of the horizontal slots themselves.

Or an adjustment to the height of the phosphors at the various TVLs to values which might minimize the chance of this.

In the meantime, I can try to play around with the existing options to see if I can find a different TVL/Display’s Resolution combination or maybe build upon a preset which uses a slightly different phosphor height.

This phenomenon is also pretty glaring when the Mask Type is set to Shadow Mask at 300TVL.

I would really like if we could customize the Shadow Mask by being able to rotate/stagger the dot pattern to allow us to fine tune the look and better match our individual displays’ varying subpixel layouts.

Am I correct in my thinking that the CRT-Guest-Advance family of Shaders already has some sort of consideration for these types of scenarios @guest.r?

I could remember observing the SlotMask and scanline interaction a while back and wondered how the Horizontal Slots never seemed to be symmetrically aligned or evenly spaced in relation to the Scanlines.

I felt it was a bit strange but now I can understand the benefit.

@MajorPainTheCactus I’m thinking that if I lower the scanline intensity, to make them very light or almost invisible that I might not see this particular undesirable interaction. I’m now getting familiar with the controls so I’ll have to play around with them so see how I can go about accomplishing this.

[Nesguy]

@Cyber

So probably a vertical Offset to the position of the entire phosphor grid (or all of the horizontal slots themselves.

Part of the problem might be the regularity of the pattern? On a CRT, the scanlines don’t always fall on the same spot on the phosphor. How does it look when you disable integer scaling? Edit: nevermind, it doesn’t really help

[Cyber]

The first thing I tried was setting the Scanlines Max and Min to 2.00, which effectively disables the scanline Gaps and the “problem” obviously went away.

2.00 is too high of a value for the min/max though, so I lowered it to 1.00 and it looks beautiful.

Reminds me of my old CyberLab Slot Mask CRT 1P2RTA preset.

On checking with a white screen, the problem is still there though.

I’ll probably try gradually decreasing the Scanlines Max and Min to see where the “breaking point” is where the artifacts reappear then try playing around with the Scanlines Min to create some dynamics.

[Nesguy]

Isn’t this just a classic moire pattern? I think the only thing that’s going to work is to reduce the strength of the scanline pattern and/or the mask pattern, so you’re probably on the right track.

On a CRT it isn’t a problem because the beam hits on a slightly different part of the triad each time.

[Cyber]

@Nesguy

Isn’t this just a classic moire pattern?

Yep, I guess so but it took some close analysis of clear zoomed in photos to finally acknowledge and identify it as such.

There might have also been confusion caused by the other pulse like artifacts seen on 480i content.

On another note, I set the Scanlines Min/Max to 1.30 and things look much more consistent! Just have to adjust the Min slightly to create some dynamics.

[Nesguy]

I can still see it in motion, though

I wish I had never read these posts, lol.

[Cyber]

@Nesguy

I wish I had never read these posts, lol.

Well we can all work together to understand the Shader and it’s characteristics even better and report and assist in fixing bugs along the way so that eventually you might not have to see anymore undesirable things.

In the meantime, here’s how things are looking now!

[Nesguy]

I think the solution will be more resolution. Looks fine at 12x, but we need 5k for that

[Nesguy]

For whatever reason, setting the display resolution to 8k seems to help with the slot mask moire, although it doesn’t completely eliminate it (using a 4k display).

A scale of 12x will completely eliminate the problem.

Can you please give these a try and let me know what you find, when you have a moment?

[Cyber]

@Nesguy

For whatever reason, setting the display resolution to 8k seems to help with the slot mask moire,

This is how you switch TVLs and Masks in Sony Megatron.

This plus trying out the different Presets which might default to different Mask characteristics.

So what it did is make the Slot Mask pattern shorter, which can be desirable in some cases.

Looks good, that’s why we really need people like yourself to see this shader achieve its full potential.

On another note, have you tried using HDR with other shaders now that you have an HDR display?

[Nesguy]

I’m not sure how to get other shaders working with HDR? I’ve enabled auto-HDR in Windows.

I did try Azurfel’s edits to Megatron and it provided a noticeable improvement, I think.

In guest, mask 12 + slot mask width 3 is working pretty well (this might be the same mask configuration as Megatron slot mask 300 TVL 8k?) I currently can’t take screenshots with Vulkan so it makes it a bit difficult to see what’s going on.

[Cyber]

@Nesguy

I currently can’t take screenshots with Vulkan so it makes it a bit difficult to see what’s going on.

You can temporarily switch to the D3D11/12 driver, make sure HDR is on in Windows and Enable Experimental Features is toggled in GeForce Experience.

You should then be able to take HDR+SDR screenshots in jxr format by pressing Alt+F1.

You should be able to capture video by pressing Alt+F9. Alt+Z brings up the Shadow Play Menu where you can do things like enable custom bitrates.

[Nesguy]

So taking and uploading a screenshot is kinda painful now, huh? Price of progress, I suppose. I probably also need to use a third party image hosting service to get the screenshot to display properly, as well?

[Wilch]

I haven’t used RA in a while, but I remember enabling ‘integer scale overscale’ fixed this issue with the slotmask pattern in Megatron. The cropping at the top and bottom of the image is probably no more than you’d lose on a real crt either. Not an ideal fix, but a temporary workaround maybe.

[Jamirus]

In real life, I would expect to see barely to no scanlines with high resolution content on standard slot masks TVs, certainly not with 300 TVL. Is this how the Megatron works though? I would think (but this is based on some testing with a lower res than 4k display) that, the more “breathing room” (scale factors) you have, the more the Megatron mimics the low res input.

To put this in another way, when using a standard slot mask preset (e.g. Bang Olufsen), you can expect it to look reasonably accurate for 240p content when your display has enough resolution, because the scanline dynamics where modeled according to that kind of content, but they don’t map to higher resolution, which would require a different basis.

Is my impression correct?

[hunterk]

Yes, and there’s no such thing as 480p+ on 300 TVL, AFAIK. There could be 480i, but I don’t think megatron supports interlacing simulation, does it?

Off-topic but relevant to the interests of folks in this thread, there was a big improvement to BFI in a recent commit, especially for screens capable of >120 Hz refresh rates: https://github.com/libretro/RetroArch/pull/16142

[Cyber]

Unfortunately that didn’t work in this particular scenario but lightening the scanline gaps by increasing the Scanline size and using the 8K setting seemed to have more or less eliminated the problem.

[MajorPainTheCactus]

Great thats good to know its fine with aperture grille. So the scanline algorithm is identical in both situations as the mask is applied last and has taken the same path through the code. Its not impossible theres a bug its just highly unlikely.

As I say the issue youre seeing is down to the fact we’re using discrete pixels rather than continuous phosphors. We have 10 pixels to draw the light and dark parts of each scanline and in the worse case we’re using 3 of those pixels for the horizontal part of the slot mask per scanline.

Then because we’re not exactly matching the mask with scanlines (i.e scanlines can be 9 pixels and sometimes 10 as we fit 224/240 etc of them to the pixel resolution of the display and the slot mask height might not be a multiple of that scanline height) they start to have different offsets and so you get these repeating moire like effects as edge cases appear every x scanlines depending on the display resolution, scanline resolution and vertical slot mask resolution.

You can minimise this effect with integer scaling and changing the scanline widths attack and mask height. Also upping the resolution would help - do you have an 8k tv!?!

[MajorPainTheCactus]

As I say the fundamental problem youre looking at is one of using discrete pixels to approximate continuous phosphors and the slot mask is highlighting this as it segments the scanline.

As an experiment try a integer scale with an exact multiple scaling of the slot mask. So in the above image it looks like the mask is 9 (or 8?) pixels high so put your screen scaling to that and see if you get a regular matching of the mask to the scanlines beneath it. Hopefully this should help with the moiré effects youre seeing.