CRT simulation, feedback and reference pictures needed

A place where you can chat about anything that isn't to do with games!
Post Reply
User avatar
nimitz
Posts: 875
Joined: Thu Jan 10, 2008 5:05 am
Location: Québec

CRT simulation, feedback and reference pictures needed

Post by nimitz »

Hey guys,

I've been working on a CRT simulation shader on and off in the past year, since the ones available were in my opinion not doing a very good job of simulating the actual look of a decent arcade-style CRT monitor. At this point, most of the work is done, the single pixel gradius stars look good, same with the various dither-based transparency effects (Sonic waterfalls, Jurrassic park snes, street fighter alpha health bars).

I have done the most I can as far as color correction goes with the reference images I could find on the net and from fellow member trap15 (thanks again!), but I still need quality reference pictures to do the final tweaking of the phosphor shape (and resulting apparent scanline thickness), color correction and gamma correction.

So what I need is quality pictures of real hardware on an arcade monitor or a good CRT (PVM and such) via rgb, 3 to 4 feet away from the screen, directly in front (no steep angles), in a well lit room and ideally without light reflecting off the screen (diffuse light or light behind the screen).

Ideally, the images would have one of these in the picture (if you're willing to go through the trouble you can pm me or ask me in this thread and I can arrange to get you one):
Image



Failing that, if you have a decent camera (the newer high end phones have pretty good ones) you can use manual/Pro settings and set it to:
ISO 200 Film Speed
Exposure Value (EV) 0
5000k or daylight white balance


Other than that, if you have a lot of recent playtime on actual arcade hardware and monitors you can take a look at the screenshots and provide constructive criticism. If a lot of criticism points the same way, I'll modify the effect accordingly.



Here are a few screenshots of the current latest version (0.9), If you want more screenshots and details about the filter, see this page: http://stormoid.com/Cathode/cathode.html
Image
Image
Image
Image

For comparison here's a 4x scaled unprocessed screenshot of Alien Soldier
Image
Last edited by nimitz on Wed Nov 01, 2017 6:47 am, edited 1 time in total.
User avatar
Skykid
Posts: 17655
Joined: Sun Nov 18, 2007 2:16 pm
Location: Planet Dust Asia

Re: CRT simulation, feedback and reference pictures needed

Post by Skykid »

I'm unable to help, but I just wanted to chime in and say I love the work people put into simulating CRT displays. With today's and eventually future technology it will become easier to authentically replicate a multitude of CRT looks without having to fork out incredible amounts for PVM/BVM sets or scanline generating hardware. And the best part is no enormous box or associated picture issues that go with a real CRT.

It's looking great so far, keep fighting the good fight.
Always outnumbered, never outgunned - No zuo no die

User avatar
emphatic
Posts: 7984
Joined: Mon Aug 18, 2008 3:47 pm
Location: Alingsås, Sweden
Contact:

Re: CRT simulation, feedback and reference pictures needed

Post by emphatic »

Skykid wrote:I'm unable to help, but I just wanted to chime in and say I love the work people put into simulating CRT displays. With today's and eventually future technology it will become easier to authentically replicate a multitude of CRT looks without having to fork out incredible amounts for PVM/BVM sets or scanline generating hardware. And the best part is no enormous box or associated picture issues that go with a real CRT.

It's looking great so far, keep fighting the good fight.
+1
Image | My games - http://www.emphatic.se
RegalSin wrote:Street Fighters. We need to aviod them when we activate time accellerator.
User avatar
Bananamatic
Posts: 3530
Joined: Fri Jun 25, 2010 12:21 pm

Re: CRT simulation, feedback and reference pictures needed

Post by Bananamatic »

I thought the whole point of crts was low lag, not making the screen ugly as heck
User avatar
GaijinPunch
Posts: 15845
Joined: Mon Jan 31, 2005 11:22 pm
Location: San Fransicso

Re: CRT simulation, feedback and reference pictures needed

Post by GaijinPunch »

God damn that is nerdy and awesome!
RegalSin wrote:New PowerPuff Girls. They all have evil pornstart eyelashes.
User avatar
austere
Posts: 680
Joined: Mon Mar 22, 2010 10:50 am
Location: USA

Re: CRT simulation, feedback and reference pictures needed

Post by austere »

Awesome work as always, nimitz! :D
Bananamatic wrote:I thought the whole point of crts was low lag, not making the screen ugly as heck
The human eye isn't a simple raster engine. Much of it works in a differential manner between adjacent groups of [retinal] ganglion cells, so when you see "scanlines" you may* actually end up absorbing more details about the image. But simple scan lines just don't have the same effect as a high quality CRT.

As far as I can see nimitz has it close to perfect. The closest attempt to this (in my opinion) was MooglyGuy's MAME modification, but that had way too many knobs for most people to get the most out of it.

* Has not been proven in a scientific study but would gladly fund one. :mrgreen:
<RegalSin> It does not matter, which programming language you use, you will be up your neck in math.
User avatar
nimitz
Posts: 875
Joined: Thu Jan 10, 2008 5:05 am
Location: Québec

Re: CRT simulation, feedback and reference pictures needed

Post by nimitz »

Added a unprocessed screenshot of Alien Soldier in the top post for comparison.

It might help the younger crowd understand how little information there is to work with to create the final picture. The result will never be a HD-like image as these have up to 25 times more pixels and in turn color information to work with. That being said, the content was created while looking at the output through CRT monitors, which had many characteristics (scanlines are just a small part) which allowed to convey more lighting and coloring information to a human viewer than the unmodified color data would otherwise allow.

In short, the goal is simply to make those tiny images look as good as possible by rendering them in an simulation that is akin to the feedback the artists were getting when creating them.


austere: do you have any pictures of that "MooglyGuy's MAME modification", I would be very surprised if it comes close in terms of simulation accuracy as I remember the source and being a relatively naive implementation of crt simulation.
Last edited by nimitz on Wed Nov 01, 2017 7:26 am, edited 1 time in total.
User avatar
austere
Posts: 680
Joined: Mon Mar 22, 2010 10:50 am
Location: USA

Re: CRT simulation, feedback and reference pictures needed

Post by austere »

nimitz wrote:austere: do you have any pictures of that "MooglyGuy's MAME modification"
Good question! I might just still have the configuration on a portable HDD. It's the only one that comes close but that's more because the others are equally naive implementations, just saying that you surpassed the previous public "best". Will hopefully post a comparison so people can see.
<RegalSin> It does not matter, which programming language you use, you will be up your neck in math.
User avatar
austere
Posts: 680
Joined: Mon Mar 22, 2010 10:50 am
Location: USA

Re: CRT simulation, feedback and reference pictures needed

Post by austere »

I couldn't find the configuration. so I went off memory. It's been ... half a decade, so this probably isn't close, nevertheless:

Image
Image

Click the images to get the lossless versions. When you put the two together, it's clear your one is a lot better.
<RegalSin> It does not matter, which programming language you use, you will be up your neck in math.
User avatar
DoomsDave
Posts: 575
Joined: Thu Jul 03, 2014 5:47 pm
Location: Australia

Re: CRT simulation, feedback and reference pictures needed

Post by DoomsDave »

Skykid wrote:I'm unable to help, but I just wanted to chime in and say I love the work people put into simulating CRT displays. With today's and eventually future technology it will become easier to authentically replicate a multitude of CRT looks without having to fork out incredible amounts for PVM/BVM sets or scanline generating hardware. And the best part is no enormous box or associated picture issues that go with a real CRT.

It's looking great so far, keep fighting the good fight.
+1

So many awesome people keeping stuff like this alive, I really appreciate it and these screens are looking fantastic.
User avatar
Zen
Banned User
Posts: 1072
Joined: Fri Jul 29, 2011 4:36 pm

Re: CRT simulation, feedback and reference pictures needed

Post by Zen »

The skull in the background in the Marvel Super Heroes screen, has that nice three dimensional pop-out going on. Looking forward to seeing this shader in motion.
Sad to think many younger gamers don't realise how good these games looked.
Image
User avatar
BryanM
Posts: 6389
Joined: Thu Feb 07, 2008 3:46 am

Re: CRT simulation, feedback and reference pictures needed

Post by BryanM »

Has anyone done one of these on a per-pixel simulation basis? It always drives me a little crazy the three bars of color aren't visible. Where there's spacing vertically or horizontally, but never both. Always these clean regular lines, when there should be gaps of varying shapes and sizes along the stacks of pixels.
User avatar
jepjepjep
Posts: 974
Joined: Tue Nov 18, 2008 7:42 pm

Re: CRT simulation, feedback and reference pictures needed

Post by jepjepjep »

I took some CRT pics a while ago: https://flic.kr/s/aHsm6VJ8Rv

It's hard to get good photos because of reflections off the glass, moire artifacts, exposure settings, focal length, etc.

Imagemm2_1 by jepapac, on Flickr

Imagemegaman2zoom by jepapac, on Flickr

I think this is a great project, nimitz, but a difficult one. Also keep in mind that CRT images can differ quite a bit just due to the contrast/brightness settings. A lot of TVs really overdrive the contrast and you end up with a really bright saturated image where scanlines are faint or nonexistent especially in bright areas. BryanM brings up a great point about varying gaps. You can kind of see the effect in the image above if you consider the gaps around the white of megaman's eyes and compare them against the gaps around the red bricks.
User avatar
qmish
Posts: 1592
Joined: Sun Oct 26, 2014 9:40 am

Re: CRT simulation, feedback and reference pictures needed

Post by qmish »

Bananamatic wrote:I thought the whole point of crts was low lag, not making the screen ugly as heck
240p game on 1080p LCD is ugly unless you play it windowed at 100%-200% size which is very small

jepjepjep, this photo is legendary! Btw maybe neutral filter can help with reflections (if you use SRL).
User avatar
Stevens
Posts: 3867
Joined: Thu May 01, 2014 11:44 pm
Location: Brooklyn NY

Re: CRT simulation, feedback and reference pictures needed

Post by Stevens »

Top notch work sir.
You're sure to be in a fine haze about now, but don't think too hard about all of this. Just go out and kill a few beasts. It's for your own good. You know, it's just what hunters do! You'll get used to it.
User avatar
Xer Xian
Posts: 886
Joined: Sun Feb 06, 2005 3:23 pm
Location: Italy

Re: CRT simulation, feedback and reference pictures needed

Post by Xer Xian »

Pictures can definitely give an idea of the CRT mask effect/overlay but I wonder if they are reliable for color accuracy as well? I've always had a significant variation in color rendering on the few CRT pics I've taken.. Granted, I'm not into photograpy and my cellphone camera is mediocre - but as already pointed out there are also other factors at play, like monitor calibration or even the type of phosphors used, which may complicate things a bit.

Since there are standards that were defined for SD television (I think this pdf set the european standard), wouldn't it be better referring to them? Just saying, I'm definitely not an expert on this.

Regarding the CRT mask effect, you may want to take into consideration the dot structure of shadow mask monitors as well? They are more likely to have been used inside cabinets than aperture grilles.. here's a couple of shots from a JVC TM-H:

Image Image

(there's a button on the webpage to download the pictures)

Link crop:

Image
User avatar
nimitz
Posts: 875
Joined: Thu Jan 10, 2008 5:05 am
Location: Québec

Re: CRT simulation, feedback and reference pictures needed

Post by nimitz »

Thanks for the positive comments guys!

BryanM: My algorithm (and many others) are actually done on a per pixel simulation basis, and the three "bars" of color are in fact simulated in my filter. The issue is that when working at 4x to 6x resolution there just isn't enough pixel to completely reproduce the small details of the crt screen. The goal then becomes to have an image that looks the same from some distance, say 3 to 4 feet away from a 19 inch CRT screen and under those conditions, I'm not sure the said color bars are visible at all. That being said, I can try to change the decay curve and especially have different functions for horizontal and vertical decay, which I am currently experimenting with.

jepjepjep: Thanks for the pictures, these are definitely better than most of the reference images I have lying around!
User avatar
zinger
Posts: 1385
Joined: Sun Jan 30, 2005 10:58 pm
Location: Sweden
Contact:

Re: CRT simulation, feedback and reference pictures needed

Post by zinger »

I found these on an old flashcard recently (Astal running on a PVM monitor). Sorry if they're not exactly what you're looking for, but hopefully they can be of some help.

http://soundshock.se/dump/P8211456.JPG
http://soundshock.se/dump/P8211460.JPG

I remember taking loads more screenshots of games running on this monitor and my cabinets, and I'll get back to you if I can find them, but I'm afraid they're all lost.
SOUNDSHOCK
User avatar
Skykid
Posts: 17655
Joined: Sun Nov 18, 2007 2:16 pm
Location: Planet Dust Asia

Re: CRT simulation, feedback and reference pictures needed

Post by Skykid »

nimitz wrote: BryanM: My algorithm (and many others) are actually done on a per pixel simulation basis, and the three "bars" of color are in fact simulated in my filter. The issue is that when working at 4x to 6x resolution there just isn't enough pixel to completely reproduce the small details of the crt screen.
Do you have a version of this that would work on a 2x scale? If so I'd love to beta test it.
Always outnumbered, never outgunned - No zuo no die

hmph
Posts: 2
Joined: Mon Nov 13, 2017 6:03 am

Re: CRT simulation, feedback and reference pictures needed

Post by hmph »

OP, your shader looks absolutely incredible. I hope to get to use it someday!
User avatar
evil_ash_xero
Posts: 6245
Joined: Thu Jul 12, 2007 6:33 am
Location: Where the fish lives

Re: CRT simulation, feedback and reference pictures needed

Post by evil_ash_xero »

It's looking really good.

The one thing I always notice about CRTs is the kind of "glow" they put out. Which is shown in that Mega Man pic. That's never duplicated in almost any filters I've seen. It might be difficult to get right.
User avatar
linko9
Posts: 497
Joined: Wed Sep 10, 2008 6:01 am
Location: Nouvelle-Jersey

Re: CRT simulation, feedback and reference pictures needed

Post by linko9 »

evil_ash_xero wrote:It's looking really good.

The one thing I always notice about CRTs is the kind of "glow" they put out. Which is shown in that Mega Man pic. That's never duplicated in almost any filters I've seen. It might be difficult to get right.
Yeah that's something that probably can't be imitated without an extremely high resolution screen. Basically you need the brighter sections to "bleed" more into the blank space between the scanlines, so that white portions of the scanline are actually thicker than darker parts. Obviously can't be achieved with a 1080p screen, and even 4K isn't enough to get it to look right, I would guess. Will be neat to see what people can whip up years down the line when 8k screens or whatnot become common.
Post Reply