"Correct" Color: Some Evidence Related to the NES

RetroArch Additions Shaders
#1

The current approach to “authentic color” in retro gaming consists of researching which standards were in common usage, and the characteristics of TVs commonly in use.

The problem is, pixel artists probably never actually adhered perfectly to any set of standards.

This is Takashi Tezuka, the sprite designer for Super Mario Bros 3. Check out the colors on that small monitor.

That is a decidedly warm color temperature. It’s not 9300k at all. It’s 6500k, 100%. What else have we gotten wrong?

Probably the best we can do to determine which colors are “correct” is to find a photo like the one above.

Another bit of evidence: SMB sky should be purple as confirmed by direct quotes. It’s purple with a warm temp, blue with a cool temp on almost any TV decoder.

2 Likes
#2

I think that’s really the key here: there’s not much that can be pinned down as right/wrong, correct/incorrect.

Like the second link shows, despite intentionally picking the “purplish” color, later Mario games, which most definitely included input from Miyamoto, went with pure blue. So, inferring the “artist’s intention” in either direction is going to leave plenty of room for (valid) argument to the contrary.

I know a lot of people don’t like that and would rather be able to tell people they’re doing it wrong–or at least breathe a sigh of relief, confident in the knowledge that they’re doing it right. Unfortunately, life is messy sometimes :man_shrugging:

5 Likes
#3

I’m not sure how directly Miyamato was involved after the SNES days. But, the point still stands, as each of those palettes was chosen by an artist working at Nintendo.

I guess my general point is, if chasing color accuracy makes someone feel good, great. I myself don’t really consider color accuracy to be essential for retro gaming, and just go with what feels right. I’m somewhat skeptical of the idea that there was widespread adherence to a color standard within the video game industry, particularly during the early days.

#4

I think he’s pretty active to this day. I recall reading him being directly involved with Mario Maker, for example. Like, not writing code, but steering the vibes.

But yeah, I’m with you. I tend to tune out from the infinite discussions of gamma lol

2 Likes
#5

Maybe the guy was compensating for ambient light from cold neons in the office.

Personally, I’m fine as long as skin, sky, grass and whites are looking somewhat natural (or not, if that’s the intention). That’s how I set my TV back in the days. Left hand raised at the screen as a reference, tweaking with the right. Still doing that with shader settings nowadays.

Same with sound. I’ve got monitors for mixing with flat response in a little homestudio, but ultimately I’m still using an old Marantz with the loudness and EQ set the way I like in the living room.

1 Like
#6

not only for NES, but before it’s too late and the devices are lost (or deterioration), maybe we need a database for how original consoles with CRT TVs (no mods or emus) work IRL using professional photography that shows the colors and details well (kinda like how CRTpixels do it), and there should be still images (like PNG since it’s RGB and lossless), and to show analog effects we need animated one (like APNG)

also its a good idea if there are many images for the same console and TV but using other connections such as S-Video and RGB (only if they are officially supported and with original connections)

and it will be better if there are additional images that come from analog capture card in PC or MAC for those original connections from the console :slight_smile:

2 Likes
#7

In most cases, yes, they even had dedicated hardware and software like the one shown in the first image.

I don’t think it’s a good reference.
This photograph was taken with an analog Reflex Camera, with settings for office use (not a product), photographic development, scanned and published in print, and then a digital photo was taken of the print. All of this alters the tones of the image.

The color palette of the NES is a special case, but the screens are additive colors and the colors can vary (on any system) depending on the camera settings.

I have noticed that these colors are especially sensitive to the white point.

image

If your camera has automatic settings, they can change in real time, as shown in this video.

Captura desde 2025-10-14 08-40-31

You can see it from minute 2:49.

3 Likes
#8

The first TV I had was a beat up little black and white set. It was also the first TV I played NES games on, so the color palette looked really different than it does in a lot of these screenshots. :thinking:

1 Like
#9

There probably is a warm shift to the photo, but that’s not enough to account for the colors we’re seeing on the monitor.

There’s no way you get that color yellow background and that color blue for the logo with a cool color temp.

That doesn’t necessarily mean that they were targeting the standards that are popularly assumed to be what they were following.

I’d love to see some direct quotes or solid evidence regarding color temperatures. The connection between broadcasting standards in place at the time and whatever pixel artists were doing always seemed a bit tenuous to me.

I was speculating that one of the reasons for the shift to a blue sky in later iterations of SMB could be that they were using photographs of CRTs as a reference- every time I try to take a photo of that purple sky, it always comes out blue. That and just, you know, the sky is blue.

1 Like
#10

O_O

If you didn’t understand the above, it’s going to be too complicated to explain why, but…

The photo has flash, goodbye TV colors.

#11

The tones in the rest of the image are not shifted significantly enough to indicate the kind of shift that would be required to shift cool colors on a monitor to the warm colors we’re seeing. Regardless of whatever camera shift is occurring, that’s warm to start with.

The purple sky quote is even more damning, though- it’s purple with a warm color temp.

It’s kinda moot though- as I’ve acknowledged, we have multiple “official” palettes now, pick whatever you like. I like the original colors :smiley:

#12

Monitors are not calibrated for temperature. NTSC-J has a standard of 9300K. And the TVs shown in the photo are not professional models; one is a Trinitron and the other is a Sharp.

Could it be that they are not addictive colors like a TV screen?

This isn’t about me; I’m trying to explain why that image isn’t a good color reference.

But I think you’re overcomplicating things. That image is in the production stage; they could have put it there and then changed it.

2 Likes
#13

So you’re assuming 9300k was the standard. I know this is retro gaming gospel among some, but show me the solid evidence. Just because this was a broadcasting standard doesn’t mean it was in use by the pixel artists. It’s a huge leap without much to support it. A single quote from a developer would be great. I haven’t been able to find anything.

Meanwhile, we have the purple color quote (you don’t get that color with a cool temp), and the photo can be considered corroborating evidence of warm temperature usage.

I can see the shifts happening to the monitor but I think it’s a stretch, and too convenient, to claim that we started with 9300k and the camera somehow made it 6500k. :smiley:

Anyway, I don’t think retro gaming is a very color-sensitive thing in the way movies are.

1 Like
#14

What’s even messier is trying to judge these things using photos. Have you ever gotten the white balance to match exactly what you’re seeing on the screen when trying to take a photo or make a video recording of CRTs or CRT shaders running on a modern display?

That alone skews things at least slightly.

#15

true, and that why it has to be done Professionally with post-processing adjustments to make the images the same colors and image as CRT as they appear to the eye. It might also be a good idea to put two images, one without any post-processing modification and the other with post-processing modification

1 Like
#16

IMO they tested for three major markets, thats Japan, USA and the rest (european countries, Australia…). So it might depend which game version is used, since the cartridges contained different “content”.

3 Likes
#17

A camera shot is not a good reference for sure. That photo could have been taken from a magazine that did extra processing etc that altered colors. Only if you have a real japan nes, japan cartridge and pro equipment (that won’t alter colors again)to capture what’s really going on.

2 Likes
#18

If anything the shift would be toward blue, because that’s what always happens when you try to photograph CRTs. It’s very rare that I’ve seen CRTs appear warmer in a photo compared to IRL.

I think a lot of people are just very invested in the idea that “9300k is correct,” with no hard evidence to support this. It’s all pure speculation among retro gamers. It was a broadcasting standard and it was how games happened to be displayed in that region. Whether it was part of an intentional look is highly speculative.

The OP was meant to be somewhat tongue in cheek, guess that didn’t come across.

I’m still very confident that is a warm color temp on that monitor. Starting with a cool color temp and warm shifting the photo simply will NOT produce that color blue for the logo, period. I’ll die on this hill.

Have fun, do what feels right.

2 Likes
#19

All the posts you open asking questions, and when they answer you, you say, “I don’t know, I don’t think so, I’m not sure, convince me.”

Son, do what we all do and search the internet.

#20

I have no idea what you’re talking about. What other posts are you even referring to? Maybe you’ve misinterpreted or misunderstood something that was said to give you this impression.

And I have searched the internet - extensively- that’s part of the point I’m making. The hard proof that 9300k is part of an intended look simply doesn’t exist. Show me otherwise and I’ll happily eat my words. I know you’re very invested in being right about this 9300k thing but it’s all speculative.

In fact, I never post a question without first Googling the topic, and only if I think it will lead to a fruitful discussion - and it often does. I think there are others who would agree. Do you seriously think composite video is “solved?” (I assume that’s the topic you’re referring to?) There currently is no clear consensus on how all this stuff is working among the emulation community right now- these are HIGHLY relevant and important conversations to be having.