Please show off what crt shaders can do!

RetroArch Additions Shaders
#4621

Yes I think its wrong. There’s no way that NTSC is nearly the same colour space as AdobeRGB. It doesn’t make sense does it? Why would the standards committee go down in colour gamut space - especially by that much. (EDIT: I now see why they went down in colour space: because no TV of the time could reproduce the colours in the 1953 NTSC colour space.)

All the images I see are much more like this, which makes far more sense to me:

CG318-4K-color_modes

Yes thats exactly right and what I’m saying. Whether you’d be able to really notice the difference between EBU (PAL) and 709 I’m not sure (my guess is that I doubt it with the naked eye - PAL vs NTSC (SMPTE-C) more so though)

#4622

I’m not sure where that NTSC colour space has come from but what I can say is that all the colour gamut transforms everybody is using (Grade, Guest and Sony Megatron) conform to the EBU (PAL/rec. 601 625 line) and SMPTE-C (NTSC/rec. 601 525 line) colour spaces. Just look and compare the numbers in matrices in the shaders vs those in the Khronos Groups pdf - they all match up and aren’t anything to do with this much larger NTSC colour gamut. I dont know :man_shrugging:

EDIT: Ah-ha! I see where that larger colour gamut is coming from (from the Khronos Groups pdf):

14.5 NTSC 1953 color primaries The following chromaticity coordinates are defined in ITU-R BT.470-6 and SMPTE 170m as a reference to the legacy NTSC standard: Rx = 0.67 Ry = 0.33 Gx = 0.21 Gy = 0.71 Bx = 0.14 By = 0.08 Wx = 0.310 Wy = 0.316 (Illuminant C)

This is much larger - they were getting well ahead of themselves with this colour space and thats then why in the 1970’s everybody moved to the EBU and SMPTE-C colour spaces that the TV’s of the time could actually produce with their phosphors. Obviously 1980’s/1990’s consoles were way past the point of using the 1953 NTSC standard. Everything would look very wrong if we used that input colour gamut.

1 Like
#4623

That does appear to be what happened, though.

"The sRGB color gamut covers about 72% of the NTSC gamut."

https://www.eizo.com/library/basics/lcd_monitor_color_gamut/#:~:text=The%20sRGB%20color%20gamut%20covers,for%20applications%20involving%20still%20images.

This diagram shows the same thing:

image
image900×500 58.4 KB

and this one

image
image800×800 85.8 KB

https://color.viewsonic.com/explore/content/Color-gamut_205.html

1 Like
#4624

Yes I think it started with color spaces based mostly on pure theory and over time they became more empirical and based on what displays could actually do.

1 Like
#4625

The Reason why NTSC changed is right in the wiki article:

NTSC - Wikipedia

TL;DR early phosphors were crappy and so they decided to sacrifice saturation for brightness. NTSC was never very popular unlike SMPTE C

2 Likes
#4626

Yes you’re right it did happen - I just edited my post - see “EDIT:” and was because they way over stretched back in 1953 (some might say the same is occuring with brightness in the rec.2020 spec). As I say in the edit no TV could actually display that and so the rec. 601 standard came out that brought everything back more inline with what could be done with the technology of the time. Its a colour gamut nobody was using much past 1970 it would seem.

1 Like
#4627

yep now we’re slightly back to theory with the rec2020 spec since displays might peak at 90% coverage.

Ok so the grade CRT gamuts and the GDV CRT gamuts are all within rec 601?

In short, does this part of the readme for guest-dr-venom need to be revised?

"

"

@guest.r

1 Like
#4628

You could argue its technically correct as in those EBU, SMPTE-C and Rec. 709 triangles on this diagram aren’t perfectly aligned but whether you’d notice the difference is arguable. :man_shrugging: EDIT: It does look like SMPTE-C is completely contained inside rec. 709 I’d have to look at the actual numbers.

CG318-4K-color_modes

2 Likes
#4629

601 is completely within 709 afaik.

I’m confused, I think there must be something else going on.

Have you checked the x/y rgb values in the readme for guest-dr-venom to see if these are all within sRGB? Start at # CRT Color Profile #

It seems like the only ones (out of those in guest-dr-venom) that would be clipped would be the last two, the calibrated profile for Phillips CRT monitors and the calibrated profile for Sony Trinitron monitors, and it’s just the greens that are clipped…?

1 Like
#4630

sRGB is the same as rec. 709 in terms of colour primaries it only differs in the gamma curve that is applied (see the Khronos Group .pdf). I think this was because sRGB was aimed at the dektop with web pages vs 709 that was aimed at HDTV.

As for Guest-Dr-Venom I have no idea where those last two colour space matrices come from - I’m sure @guest can remember though? I’d be very interested.

2 Likes
#4631

Yeah I’m just somewhat baffled by the readme, it suggests a wide color gamut is necessary but the first 3 specs are all contained within sRGB, it’s just the “calibrated” profiles that are slightly clipped, unless I’m not reading things right. This color stuff drives me nuts, lol.

2 Likes
#4632

Color profile matrices were submitted by dr venom, other matrices were calculated from xy and color temperature specs using a calculation tool.

Color temperature matrices are relative to sRGB, but this shouldn’t matter too much since the xy transformation is neutral.

I would like to explain once more that the colorspace transformations are not absolute, but relative. They more or less produce xy primaries and color temperature shifts from what your calibration is and for best results a lut (for example) can be produced from you display’s icc profile to sRGB. DisplayCAL is a great tool for this.

Wider gamut of your monitor comes handy here as ‘gdv’ color profiles can be used with more accuracy.

4 Likes
#4633

Is this in lieu of using the color space setting? Or in addition to?

2 Likes
#4634

It’s about accuracy of color reproduction. A slightly inferior alternative would be to calibrate your display to sRGB. The problem is that various displays have different color gamuts and it’s nearly impossible to cover this from the shader.

If you use a lut which converts your exact gamut to sRGB, then, first, there is no clamping to sRGB boundries, you get sRGB as a starting point to use color profiles like P22, SMPTE-C…and very authentic color reproduction.

Edit: you can also use a lut to convert your display’s gamut to P22…directly. Currently up to 4 lut png’s are supported with gdv.

3 Likes
#4635

So the most accurate results would come from using the color profiles along with the appropriate color space on a wide gamut monitor? I’m still not clear on this part

It looks like all the coordinates given are within the sRGB color space except for the greens in the two “calibrated” profiles, is this correct? The other 3 look like they’re completely contained within the sRGB space. So is a wide gamut monitor necessary for the other 3?

Secondly, in theory, would a display calibrated to display 100% sRGB be as accurate as using a LUT to transform whatever the actual gamut is to sRGB?

Let me know if this is correct:

  1. Most accurate:
  • use a calibration device to create a LUT that turns whatever the display’s actual gamut is into 100% sRGB (I guess starting from the monitor’s built-in sRGB mode would be the thing to do?)
  • leave “display gamut” setting at default (sRGB)
  • apply any of the following CRT profiles: EBU, P22, SMPTE-C
  • if a wide color gamut CRT profile is desired (Sony or Phillips), create a LUT for the appropriate color space, then select that color space as the “display gamut” in the shader settings.
  1. Not as accurate
  • skip the LUT and just choose the color space setting that matches your display.

I’m assuming that since the changes we’re talking about are so subtle, we really want to make a LUT. Otherwise the profiles will not work as advertised.

This is another question I was about to ask but you read my mind, lol.

2 Likes
#4636

If you calibrate your display to sRGB then the minor differences between sRGB and P22…will be clamped to sRGB and not taking advantage of your display’s wider gamut. For best accuracy you should use the lut option, second is using your displays sRGB mode.

Edit: calibration device is not widely available, best and common option is to use a display specific .icc profile, which can be also used for lut generation.

2 Likes
#4637

Yo everyone, hope you’re all enjoying your day. Just dropping in to say if you downloaded my shader pack please grab it again, did a quick edit on them, you can grab it right here. As always hope you have fun

9 Likes
#4639

So I looked at the chromaticity diagrams again and sure enough, all the color profiles are ever so slightly outside of sRGB in varying ways (just like it says in the readme!). It’s not just the Sony and Phillips profile that are outside of the sRGB gamut, it’s all of the CRT profiles. It’s mostly extremely subtle but yes, you do need a wide color gamut for these to work as advertised, otherwise you’re no closer to the intended colors than if you were to just leave it at sRGB and not use a profile. In some cases it may be hard to tell the difference though… @MajorPainTheCactus please let me know if your findings don’t match up.

2 Likes
#4640

Please share with us your shader/preset, it looks absolutely authentic to crt

#4641

Will do - my wife is giving birth to our second son in the morning so the investigation may get delayed. :rofl:

7 Likes