Let’s keep it cordial in here guys.
ideally, we would have a calibration card in the periphery of the shot (to isolate room/camera effects), and also another shot of the display showing its own calibration pattern (smpte color bars or whatever; to isolate CRT/white point effects) with the same calibration card in the shot (to isolate camera+CRT effects).
So, keep that in mind when you post pictures of your CRTs, gang 
This is good, useful information. Thanks!
Shifting tastes and later universal adoption of D65 may offer an indication as to which one is objectively superior 
related to this: https://jov.arvojournals.org/article.aspx?articleid=2135743
I still can’t find the obscure documentary Alex was talking about, or the quote from Yamauchi explaining why anything but 9300k is wrong, but I’ll keep looking- I’m about 45 minutes into a documentary about Nintendo in the 1980s, looks promising
Ah, here we go, the 1984 XBR Trinitrons did, indeed, have the Trinitone setting:
https://www.sony.com/electronics/support/res/manuals/W000/W0006213M.pdf
While at least this non-XBR 1984 model did not: https://www.sony.com/electronics/support/res/manuals/W000/W0006164M.pdf
I think it would be a mistake to think of it that way. D65 “won” because it was part of Rec. 709, which was adopted as the global compromise standard color gamut, displacing all of Rec. 601-525, NTSC-J, and Rec. 601-625.
As would this. It would be more accurate to suggest that anything but D93 is (generally) revisionist for (most) Japanese games made prior to the 1996-1998ish.
I am personally strongly of the opinion that the grand majority of Japanese Famicom games look their best when using the CXA2025AS palette with a D93 white point, but i’m sure there were some Japanese folk who immediately hit that Trinitone button for D65 back in the 80s and 90s when they set up their new TV. (Just as there were countless USians who stuck with D93 despite D65 being the more correct option for most media made outside Japan.)
This combo actually does look good, yes, because it pushes the reds harder than most palettes. And you’re touching on another important factor, the variance in TV decoders - which definitely makes it difficult (impossible?) to pin down any single, intended look. (the sky isn’t purple with that combo, though! 
)
I’ll grant that 9300k may be “generally correct” for Japanese games pre 1998, but it’s difficult to say that other temperatures are “wrong” when the games were being marketed globally.
Well the documentary I was watching ended without the quote from Yamauchi that I was expecting, oh well. “The Story of Super Mario Bros 3” if anyone is curious. Still pretty good.
I think the colors shown here are absolutely definitive:
That’s the thing, it isn’t that it is supposed to be purple, it’s that it uses the Famicom’s “purpleish blue”, as opposed to the other available options, which is distinct from actually being purple.
But it is quite easy to say that D93 is more likely to reflect the creative intent, and color temperatures other than D93 are less likely to reflect the creative intent.
Keep in mind that video was itself created for Japanese broadcast in 1986, and thus would have also been intended for viewing at D93. (It has also been through who knows how many generations of copying and who knows what processing along the way before it was finally uploaded to Youtube.)
I know, it’s that ambiguously purple blue. The video actually shows it pretty well, I think, even if it’s not reference material.
Ok, I’ll grant you that. It’s more likely to reflect the creative intent if such intent actually exists- I guess that’s the part I’m questioning. We don’t really have a smoking gun in the form of a developer quote establishing that 9300k is essential. If it was, I think such a quote would be easy to be find. Besides color temperature, different tv decoders can lead to wildly varying colors, they knew it would be displayed on millions of different TVs, etc.
I know man, I was just having fun with that one.
No one ever bothered to create an actual standard for display gamma until BT.1886 in 2011 because literally all CRTs could be presumed to physically have functionally identical gamma.
We don’t tend to talk about or create standards for things that just Are, and that is one of the ways we end up with these sorts of gaps in understanding.
I think there’s still a leap being made here from a standard that was being used at the time to an essential part of an intended look, if it exists. I think it’s a reasonable assumption that IF they cared about color temperature, they probably wanted it to be 9300k. But then if you throw in the TV decoder it gets more complicated. “Only Sony 9300k is acceptable” seems less likely than “the games were designed to look acceptable on a range of displays, with a variety of color temperatures” How is a single intended look even possible when TV decoders vary so much?
The first shots are 6500K the second ones 9300K (also using sRGB to NTSC correction), notice how much more “correct” it looks at 9300K. That teal gradient in intro image at 6500K makes no sense, if not ugly.
At the second picture it looks completely flat at 6500K, notice how balanced it looks at 9300K.
I’ve drawn some pixels myself in the past, the artist probably wanted to use a gray-ish blue to simulate distance (objects appear less saturated in distance) and 6500K bumps up green, messing the original. That is some state of the art pixels there, i wouldn’t drop that teal there and ruining it, not even with a gun in my head lol.
Ok, I will concede. 9300k is the only correct color temperature! Everyone else is doing it wrong and completely butchering the artist’s work, spread the gospel!
I think if you showed those two images to a group of random people about half would select image A and half would select image B- that teal color could be ambient light from the monitors, that’s how I always interpreted that. Honestly I do prefer the colors in the first images 
It’s sick and wrong, I know. Second image in particular looks just dim and dull and desaturated, that’s usually how 9300k feels to me. I feel like it just sucks all the color out of everything. I know 9300k is correct and I’m wrong, it’s a serious personal problem I need to work on.
What is this even doing?
It tries to project colors as they would look like on an NTSC TV.
Right, I get that part, but I’m not sure how it’s actually accomplishing this- is this the “NTSC colors” shader you’re referring to?
I think part of the confusion in the 9300k vs 6500k debate is that it doesn’t work the same on digital displays, the difference is really dramatic on LCD because the backlight has a bias. I guess the CRT phosphors have a bias, too, but it seems a lot more flexible because you can fine tune the electron beams.
Also- white is white and not tinted in the 6500k shots… but that’s my problem, the tinted white is correct.
I think it’s worthwile to mention that arround 1987 SMPTE-C standard superseded the old 1953 NTSC standard in official reccomendations.
Some consoles outputed colors, optimized for the old standard, and this could be an explanation why there is a discrepancy if viewed on an older ntsc 1953 or later smpte-c tvs.
So where I’m currently coming from is that 9300k may be “generally intended” or whatever on a CRT, but things work a bit differently on our modern, backlit displays These are (almost) all produced and manufactured with D65 in mind including the backlight- you can force a cooler temperature but this is just filtering out more of the red and green, washing out that portion of the image in order to fake it.
A typical LCD has a native white point, unlike a CRT’s fully adjustable white point
With an LCD, what was “generally intended” regarding color temperature no longer really applies…
I guess OLED would give you the same white point flexibility as CRT.
Hmm…i’m not entirely sure about this, since white points are also related with color temperatures, which also bring specific tinting.
From my understanding, a “white point” is strongly correlated with a hypothetical light source, which for example is “shining” on a “white” surface. Ofc. the light source can affect all color combinations, which produces specific D#### hues.
The definitions is something else, connected with light emmitings of heated materials…
I found some sources about this, the reading is not too heavy:
https://www.eizoglobal.com/library/basics/color_temperature_on_an_LCD_monitor/index.html
In other words: RGB(255,255,255) looks a bit blueish on a D95 calibrated display, but D65 or “6500K” is also within the triangle of representable colors, but is not to be reproduced by (a proportional of) RGB(255,255,255).
This is an another one of those gaps in understanding I was talking about. Rec. 601-525/SMPTE C was already the de facto standard by the time even the Magnavox odyssey was released in 1972, it just didn’t have a written standard behind it until Rec. 601 in 1982.
The NTSC 1953 gamut displays were only really produced until 1958 when the original phosphors were displaced by P22 all-sulfide phosphors, which were themselves displaced in 1964 by P22 sulfide/vanadate phosphors, and then finally the Conrac/RCA gamut phophors in 1968-1969, which both Rec. 601-525 and SMPTE C were later based on).
NTSC 1953 continued to be used as a broadcast container gamut until ?1982ish?, similar to how Rec. 2020 is used as a container for various other gamuts such as P3 today. Different displays supposedly handled that in different ways, some with so called color correction circuits that introduced their own issues, but none of that should have been a significant factor for “modern” displays by the mid-80s as far as i understand it. (Frankly, i would love to take colorimeter readings off still working 60s and 70s TVs and see what they might have really been doing in actual practice).
PAL-land adopted their own preferred phosphors (EBU) in the late 60s, and Japan theirs (NTSC-J) in the early 70s.
Sony Trinitrons also had their own somewhat distinct phosphors (generally called Apple RGB nowadays if it’s called anything at all, due to the famous Apple 13" RGB monitor.)
This is probably a significant factor, yes. I personally change color temperature directly using my C1’s settings rather than using shaders to do it, even. D93 on a CRT has a certain shimmer and shine to it that i haven’t been able to replicate when adjusting color temperature with shaders alone, there is something that is just flat out missing in the process i guess. But changing it directly on the C1 display absolutely does the trick. It looks just like my FW900 set to D93 in that regard.
On my setup the 9300K Shots look pinkish in the whites in the text and the grey alien planet fog background. The whites and greys look neutral in the “6500K” shots, supposedly the first ones.
The Super Metroid title text looks too red near the bottom in the “9300K” shots.
So those look a bit off to me, while the 6500K shots look perfect colour wise.
Subjectively, the shade of green on the spacecraft’s windshield makes much more sense to me in the 6500K (first) shots vs the second.
I don’t agree with the statements above, actually I find the opposite to be true from my perspective.
This reasoning and logic doesn’t necessarily apply to the images that we’re comparing or to what was going on in the heads of those who programmed the game decades ago.
It might be better to try to track down some real developers at this point as well as those close to industry, for example in education and game development training or associates of developers who might have been privy to conversations and had actually been there to more easily find answers to these types of questions.
I always thought it was supposed to represent fog. The whitish of the fog, with the blueish of the room gave that teal hue.
That’s exactly what it evokes in me, as well, without even realizing it until you said it.
