Sony Megatron Colour Video Monitor

[MajorPainTheCactus]

Just uploaded my Sammy Atomiswave simulation to Github. This is probably one of my most favourite cabinets and screens as its large, got scanlines and has a slot mask.

Again possibly the colours are a little off but we’re close.

As always you can find it all under the ‘hdr’ folder in the zip (once its been merged):

slang-shaders-master.zip

Here’s the cabinet I took photos of:

Absolutely amazing cabinet I’m sure you agree and here’s my side by side comparisons of the CRT (first) vs my simulation shader. Note the ISO is now the same on both the CRT and LCD.

Please do click on each to see larger more detailed photo.

CRT Photo: OnePlus 8 Pro Camera: Pro Mode, ISO 200, WB 5000K, Aperture Speed 1/60, Auto Focus, 48MPixel JPEG.

LCD Photo: OnePlus 8 Pro Camera: Pro Mode, ISO 200, WB 5000K, Aperture Speed 1/60, Auto Focus, 48MPixel JPEG.

[hunterk]

we should probably consider using/developing a reproducible methodology for getting pictures of a test image (SMPTE color bars or whatever) to generate color transform LUTs for the various CRTs, too.

[Nesguy]

Fantastic!

Do you know the specs on this monitor? It looks medium-high TVL.

It occurred to me that you can use the 8k mask (RRGGBBX) to get a consumer-grade/average arcade look. TVL would be around 300.

[MajorPainTheCactus]

Yes! I’ve got some pictures of a 350TVL PVM I intend to simulate.

[MajorPainTheCactus]

Yes so one issue above and why I’ve kind of been side stepping colour for the time being is that the colour in the photos above are different to what you actually see by eye. They are probably better to match though than what I’ve done above.

The other thing to note is that my LCD display is just not bright enough - it needs another couple of hundred nits. A lot of the colour differences caught above I think are due to brightness. If I’d used ISO100 on the cabinet photos I reckon they’d be much closer.

Another other thing is that the LCD photos don’t look the same colour as my LCD. I should probably get a decent camera rather than use my phone camera.

Having said all that I think we should try and remove as many layers of differences as we can and LUTs are the way to go.

[MajorPainTheCactus]

I’m going to do an experiment and retake the photo of say Cammy above with higher ISO on my LCD and see what effect reaching the limit of my camera has on colour.

[MajorPainTheCactus]

As for the CRT it appears this cabinet is a rebadged Taito Egret 3 which you can read about here:

Taito Egret 3

[MajorPainTheCactus]

Time to implement horizontal deconvergence…

[Syh]

Love me some deconvergence!

[Jamirus]

It’s a former Taito Cabinet turned into Atomiswave, running a CPS 2 game?

[Nesguy]

Is there a way to avoid blurring the individual triads while adding deconvergence? Maybe @guest.r knows?

In the CRT shot you can still see distinct RGB phosphors but they get blurred together in the LCD shot.

From what I can see in the crt shot, deconvergence is something that only happens at the edges of objects(?)

For some reason the vertical lines are emphasized more in the LCD shot, as well.

[DevonCM]

Yeah that really shows off the beauty of the CRT, the phosphors that are partially lit and fading out really smoothly and in some cases into super fine spikes. Probably gonna need 16K to get close to that.

@Nesguy

From what I can see in the crt shot, deconvergence is something that only happens at the edges of objects(?)

Well no, it’s just that the colors aren’t all perfectly aligned over a given part of the image, but you only notice it at the edges where it runs off by a certain amount. You don’t see it in the middle cause they’re already overlapping there anyway. I’m sure there’s a far more technical way to describe it though.

@Nesguy

For some reason the vertical lines are emphasized more in the LCD shot, as well.

Yeah it seems like the signal is way too sharp, as if it’s not using the right type of NTSC pass or something. Also noticeable on the red parts on Cyclops, and the black lines between the yellow and blue on his thighs turning into hard stairs. I know arcade monitors would be getting a cleaner signal than consumer TVs (they’re RGB or VGA or something, right?) but I think it’s still much different than just passing the raw pixels through. Although I could be completely wrong and it’s from something else entirely.

I gotta say though, getting as close to that arcade look as possible is the dream and I’m really excited to try it out eventually. I hope we start to see some nice RGB grid gaming OLEDs this year, and not just those curved ultrawide ones they’ve announced so far…

[MajorPainTheCactus]

@Nesguy

Is there a way to avoid blurring the individual triads while adding deconvergence?

I suppose it depends what kind of deconvergence you’re talking about but the answer is yes as my first attempt will simply be an offset which by looking at the images is all I need and is all I’ve done for my vertical deconvergence. Let’s see though.

[MajorPainTheCactus]

@Nesguy

In the CRT shot you can still see distinct RGB phosphors but they get blurred together in the LCD shot.

That’s probably just poor focus from my camera - I haven’t added any blur at least kernel based blur.

[MajorPainTheCactus]

@Nesguy

From what I can see in the crt shot, deconvergence is something that only happens at the edges of objects(?)

On my CRT it’s an offset - a misalignment of the electron beams resulting in lighting up the wrong phosphors. This gives a blur effect to your eyes.

[MajorPainTheCactus]

@Nesguy

For some reason the vertical lines are emphasized more in the LCD shot, as well.

Yup that’s precisely why I need to add horizontal deconvergence.

[MajorPainTheCactus]

@DevonCM

Yeah that really shows off the beauty of the CRT, the phosphors that are partially lit and fading out really smoothly and in some cases into super fine spikes. Probably gonna need 16K to get close to that.

I’m not so sure chucking more pixels at this is the solution - let me add the horizontal deconvergence and see where we get to. I’m not convinced there’s an awful lot of luminance difference across a phosphor as the beam goes across it. Sure there maybe very subtle differences but I’m not sure you’d be able to really tell - certainly at a distance at least.

@DevonCM

Yeah it seems like the signal is way too sharp, as if it’s not using the right type of NTSC pass or something. Also noticeable on the red parts on Cyclops, and the black lines between the yellow and blue on his thighs turning into hard stairs. I know arcade monitors would be getting a cleaner signal than consumer TVs (they’re RGB or VGA or something, right?) but I think it’s still much different than just passing the raw pixels through. Although I could be completely wrong and it’s from something else entirely.

I think this is mostly down to the horizontal deconvergence again - it gives a much softer image. As for Cyclops there’s a whole load of stuff going on there not least the camera - in the full image you’ll see the highlight on Juggernaut’s boot has been completely removed/clipped out so it’s probably just a poor photo. I also need to do some more work on the Sammy Atomiswave simulation as I noticed I’ve got Cyclops belt wrong. I don’t think much of this has very much to do with the input signal as I’ve mostly proven in my PVM shader, it’s largely a screen problem. But I do see some overly aliased pixels so I could be wrong. I’m going to try and implement the deconvergence tonight as it should be relatively simple to add - famous last words.

[MajorPainTheCactus]

I’d say not quite as you describe it as in, in the factory they stuck a Sammy Atomiswave logo on the cabinet instead of a Taito Egret logo. Quite what decides its a rebadged Taito cabinet rather than a rebadged Sammy cabinet I’m not sure. But yes you are right its running a CPS2 game - its probably run a lot of different games in its time. It is an absolutely amazing screen to see in person though.

[MajorPainTheCactus]

Just a small thing to add from this photo I think this might be a PAL system. It is in UK after all - I set my camera to 1/60 shutter speed and you can see black lines - I think if instead I’d set it to 1/50 those lines would go because of better synchronisation.

However I dont know if you get PAL arcades but then again aren’t NTSC and PAL based off of different A/C electric systems? I may well have just made that up in my head - certainly the UK has a different voltage to both mainland Europe and the USA.

[DevonCM]

@MajorPainTheCactus

I’m not so sure chucking more pixels at this is the solution

Sorry I just meant that to match the close-up look of the spiked and faded phosphors as well as their rounded corners it would need way more.

@MajorPainTheCactus

I’m not sure you’d be able to really tell - certainly at a distance at least.

Of course you lose most of it at normal viewing distance, but still I bet a lot of the subtle smoothness and blending comes from the way they’re fading and only partially lit, even just barely in some cases. I mean just looking at that picture of the timer there are a dozen different phosphor shapes based on how they’re being lit.

@MajorPainTheCactus

I think this is mostly down to the horizontal deconvergence again…

…I don’t think much of this has very much to do with the input signal

In this case it might be the deconvergence, but based on your full shot of the cabinet the alignment/calibration seems off so it’s probably not good to try to match that exactly, at least for the base setting. Though it’s also hard to tell how much of that is from geometry bloom… I think I would much rather the blending and softness come from the signal’s consistent properties than “flaws” of the individual display’s setup.

But does this mean you are or aren’t doing the NTSC signal emulation? Because just in my limited experience playing with CRT Royale, tweaking the NTSC settings makes the biggest difference in “believability” of the resulting image. Where the base version without the NTSC passes just looks like a mask pasted on top of an emulator.

I don’t mean to be presumptuous and I only have a very limited and surface-level understanding of how this stuff works, so apologies if I’m way off. Relying on just deconvergence for blending reminds me of how a lot of modern games will use heavy chromatic aberration to hide poor anti-aliasing or just generally insufficient resolution, rather than as a subtle post effect. And it’s interesting because that’s supposed to just be an undesirable trait of a particularly bad lens, maybe similar to using a cheaper monitor and then also not keeping it properly calibrated. Like that post you linked mentions they downgraded from the nicer Toshiba to a cheaper and inferior model, for example.

It seems better to address the source where possible and then season the display-specific stuff to taste, but that could be difficult without overwhelming users with settings. A lot of that will also come down to whether someone wants to reproduce the best case scenario possible at the time, or what the majority of people would have actually experienced. Of course I think it would be nice to be able to try both.