Sony Megatron Colour Video Monitor

[MajorPainTheCactus]

Have you definitely checked all the colour settings are on ‘full colour’ in Windows display settings? Hmm another thought is maybe that display has an irregular sub pixel layout? Could you try BGR instead of RGB in the shader parameters. Also could you take a really close up shot of the display so I can see whats going on with the subpixels. Use the SMPTE-C plunge bars - red green and blue and get a shot of in between the bars so there is a bit of white, blue and green or red and green if you see what I mean.

EDIT: sorry just noticed you said it’s got full colour on

[MajorPainTheCactus]

Is it alright in windowed mode then? Also you need neutral sharpness - that maybe 0 or 50 usually dependent on your display. Also just to test try 60hz with VRR disabled.

Mine advanced display settings looks like this:

[BendBombBoom]

There is an info button the C1, I think it’s the 3 dots and it should tell you resolution, refresh rate, and if it’s in HDR. From my experience you need a GPU with a hdmi 2.1. I have a laptop with a 2060 and while I can turn on HDR in windows 10, it looked significantly worse than my 3080 with a 2.1 output.

[MajorPainTheCactus]

@nfp0 as another test and further to @BendBombBoom comment can you try in SDR mode - turn off all HDR both in Windows and RetroArch and turn on SDR in the shader. You may need to play around with the brightness, contrast, saturation and gamma settings to bring up the brightness if its particularly dark. This will rule out some sub pixel issue.

[yoZe]

@MajorPainTheCactus

Hi, the first question I have is what GPU do you have? This shader is one of the most efficient and will be faster at 4K than most other shaders at 1440p.

Resolution is key for TVL - 4K minimum for 600TVL, 1080p minimum for 300TVL. I’d get a 4K monitor that’ll go as bright as you can afford.

I’m looking for a 1440p monitor because my pc is mainly used to play recent games that require a lot of resources.

[MajorPainTheCactus]

Today I went back to the JVC Professional TM-H1950CG simulation as @DariusG had very kindly donated a better image than I had over on this thread (if anybody else has shot please donate generously!). Thanks again DariusG!

Calling all CRT owners: photos please!

JVC TM-H1950CG broadcast CRT (perspective corrected)

Anyway ignoring the aspect ratio in the image, I think I’ve got quite a bit closer.

JVC Professional TM-H1950CG

Sony Megatron Simulation

CRT Photo: No idea - ask DariusG!

LCD Photos: OnePlus 8 Pro Camera: Pro Mode, ISO 200, WB 6500K, Aperture Speed 1/60, Auto Focus, 48MPixel JPEG.

[Nesguy]

Fantastic!

The only things we haven’t seen yet (I think):

-low TVL aperture at 8K (for the 3 people that have an 8K display)

-low dot pitch dotmask (like an 80s computer monitor). I haven’t been able to find ANY reference shots of this, though.

[MajorPainTheCactus]

Ok so yes I haven’t implemented 300TVL on an 8K screen - I do need to do that I just couldn’t bring myself to type out the massive slot mask.

So I do have a low TVL dot mask but it may not be what you mean? If you select shadow mask and put it down to 300TVL it’ll show it. I’ll post some pics - I think last time you saw it you thought it was too low res.

As always close up photos of the displays would be of help if you have some?

My next major concern apart from yet another reorganisation of shader presets to support SDR out of the box is to focus on the signal side of things - chroma compression, NTSC comb filters and the like.

[MajorPainTheCactus]

Just added SDR versions of all my monitor presets and fixed all the presets after breaking them all last night .

A BIG BIG thank you to hizzlekizzle who I assume is @HunterK for all your hard work merging these pull requests almost every night. I will calm down soon!

Please delete your shaders_slang/hdr folder and update again - once the pull request has been merged - we’re on Sony Megatron v2.6 now.

[BendBombBoom]

I forgot to add this detail on my C1 OLED vs qn90a comparison. The Megatron shader on the qn90a has a buzzing, almost CRT flickering when applied. The OLED has none of that. (Which I prefer.)

[MajorPainTheCactus]

That is odd - backlight strobing creates a flicker in certain situations - nothing like BFI or backlight strobing was on? Was it only specific to the Megatron? If so then that would say something to do with backlight dimming but then I’m not sure that should be at high frequency i.e flicker especially if the image was static (or relatively static). You weren’t able to get any pictures of the sub pixels during the flicker?

[DevonCM]

I get flickering with certain settings too when just testing on my 1440p monitor (ASUS PG279) and it’s definitely not BFI or backlight strobing. I think it has something to do with how big the gaps are between lit pixels/subpixels and maybe how much screen coverage there is overall.

You might see similar weirdness with these patterns: http://www.lagom.nl/lcd-test/inversion.php#invpattern

I think the hard on/off grids (and even full screen on/off like BFI) can really mess with certain displays, and even lead to image retention. This would be one good reason to have some settings for a less than 100% mask strength or some type of bloom, so those pixels and subpixels that are normally sitting off for extended periods of time (right in-between others that are on full blast) can get some activity on them as well.

Maybe even just some kind of simple noise layer would help.

[MajorPainTheCactus]

Yeah I’m not sure I buy into this theory - there’s really little difference between this shader and displaying a pure red or a pure green or pure blue screen - just one out of four pixels are black in the horizontal direction is the only real difference.

I’m not sure there’s anybody that says don’t display a pure colour screen due to risk of burn in. In fact we’re turning off the majority of sub pixels so it’s even more unlikely to raise temps as to damage displays. LCDs in particular are lit by comparatively large lights behind them so they don’t really have the control to damage individual cells as in the above description.

In your particular case youre using an SDR IPS display and I’d imagine its pretty much impossible to have burn in on that display. Even an OLED probably wouldn’t have problems.

As to why you’re getting flickering I’m not sure - do you get flickering if you display a red, green or blue screen? My IPS is fine at displaying all those patterns that you linked so it might be hard for me to correctly diagnose this - it does feel like a display level issue. I presume you’re using the SDR version in all this?

[DevonCM]

Here’s an example of a paused screen where it’s particularly apparent, note that most of the screen is black here.

This is the crt-sony-megatron-sony-pvm-2730-sdr preset with the TVL set to 1000. It’s less obvious on 800, but about the same on 600. It doesn’t seem to happen at 300, presumably because the chunks are so big at that point that the signal isn’t spiking back and forth as rapidly.

I’m not sure if it happens during movement, whether it’s less obvious or the fact that it’s not still actually prevents it. I don’t actually use this shader since my resolution isn’t appropriate for it, but this flickering was something I noticed when testing it out and playing around with the settings so I just wanted to corroborate what BBB was saying.

If you give me a preset with some added bloom or lower mask strength or whatever I can try it out and see if it still happens, but there’s probably someone far more qualified to look into it further.

(Also it’s only letting me attach one image per post and they still require approval for some reason, don’t mean to shit up the thread and doublepost…)

[DevonCM]

I said image retention, not burn-in. It has nothing to do with displaying a section of pure color, or temperatures. It’s about staggered pixels, especially on still/constant images (like when paused), and it’s more apparent when only some portions of the screen are filled.

Here’s an example of how displaying one of those patterns on only a portion of the display can affect the rest of the line that it’s on: (Only the green portion is the actual pattern)

This is not uncommon, it happens on both of my completely different monitors, and I’m sure many others.

I’m also not suggesting that this shader causes the same issue or behaves in the same way, just that staggering pixels can absolutely cause monitors to do weird shit.

[hunterk]

sorry, i approved those in reverse order

[MajorPainTheCactus]

Ah ok you’re using the 1000TVL because you’re on a 1440p display which makes sense as that gives roughly 660TVL. Ok so that makes a little more sense in terms of a high frequency pattern. The 1000TVL option uses a magenta-green mask as in there are no pixels that are off on a pure white screen just alternating magenta green lines. This may play havok with any image processing the screen is doing and I’d just turn off any image processing.

However fundamentally I’m not sure LCD technology has any inherent problem with display alternating colours - we all used to use Windows not too long ago where 1 pixel lines were all over the place for instance. I do see terrible problems occur when chroma compression is employed to compensate for low bandwidth. I note your display is a 1440p with 165Hz refresh rate - are you running it at that hz and res? Mind you that might explain the green image in the bottom photo but not the flickering.

I’d like to try and repeat that green pattern my end as I have a few monitors here - what exactly are you doing in the bottom picture?

Can you see what is actually flickering? Is it the whole screen or is certain areas or at the pixel or sub pixel level? Is it irregular or is it a fizz? Does it go away when you display a full white screen with that preset and 1000TVL option set (test 240p Suite has one)? Another thing you could try is the black white mask on option 3 for CRT type.

Although you could well be right and what you see above is the same problem, BBB is also using a 4K HDR screen and will undoubtedly be using the 600TVL mask which is very different to the 1000TVL one youre using and then add on to that he’s using a TV and youre using a monitor which are too very different beasts and the end result is probably who knows with the myriad options at play.

I’d like to repro this and I think I may have just the monitor bear with me - its certainly displaying my shader as green through Parsec but I thought that was just Parsec compressing the image but maybe it isn’t. I’ll have a play around with it.

[MajorPainTheCactus]

So I’ve added a ‘crappy TV’ preset that has a AEG CVT 4800 VT simulation with a NTSC composite signal simulation (basically vanilla NTSC-adaptive .slangp).

I played around with all the settings in NTSC adaptive and the defaults seemed good enough with a tweak to slangp scaling.

Anyway most of the details are over on this thread:

How crappy of a TV do you need for composite video artifacts?

Here’s something else this time from my 2006 set, PS2 composite vs VGA-RGB Scart with some filters/shader on top. FCE Ultra/FCEUMM default palettes. PS2 NTSC: [PS2_CP_AEG_NTSC] PS2 PAL: [PS2_CP_AEG_PAL] VGA-RGB: [RGB_AEG] With core composite filter activated: [RGB_CP_FCE] with Blarrgs SNES composite filter instead: [RGB_Blarrg_SNES_CP] with NTSC adaptive shader default settings: [RGB_NTSC_Adapt]

but here’s a video of the results here (as always play full screen and 4K to see the details properly):

Sony Megatron CVM - AEG CTV 4800 VT - 300TVL Slot Mask

[rafan]

Hi @MajorPainTheCactus , I’m interested to verify that RA’s output is utilizing my screens 10-bit color depth.

My screen is in HDR (400) with 10-bit color output. (verified with control panel settings).

I enabled logging in RA with write to file, and after loading the HDR PVM 2730 preset, the log file shows:

[INFO] [Vulkan]: Using BGRA8888 format.
<...>
[INFO] [slang]: Using render target format R8G8B8A8_UNORM for pass output #0.
[INFO] [slang]: Using render target format A2B10G10R10_UNORM_PACK32 for pass output #1.
[INFO] [slang]: Using render target format A2B10G10R10_UNORM_PACK32 for pass output #2.

Just to verifiy with you, is the above confirmation of final RA output being in 10-bit color depth?

[MajorPainTheCactus]

Good question! I’m not sure where or when the first message is printed out but it would be wrong if RA is using a HDR swapchain (back buffer) at that point. It’s difficult to know whether it’s talking about the swap chain format as well. It should be saying A2B10G10R10 for a HDR format - let me check my output and I’ll confirm (how do you enable logging to a file btw)

All the slang messages are correct however