Wayland Color Management Protocol Posted For Weston

[oiaohm]

HDR10 and dolby PQ are both color spaces, HDR400/600 etc, get very close to being colorspaces, as HDR400 defines srgb support, 8bit color, brightness, and at least a PQ transfer curve,

This means you are now over-bright. PQ P3 is over bright srgb/ Rec. 709

don't assume that REC2020 is the only HDR spec

To be correct is not assume by me. REC2020 for SDR parts of the specification references REC709. I did check what REC2020 says SDR should be.

DCI-P3 65 is the wrong color space. The define points being the 3 corners and the white point of srgb and Rec 709 absolutely align with each other there is only a nits difference. DCI-P3 65 white point lines up but the 3 corners don't line up with REC709/srgb at all.

https://uxplanet.org/understanding-colours-in-screens-the-story-of-srgb-and-dci-p3-colour-profiles-2109485f7cfe?gi=349b7f178bfe

HDR10 is not a color space it s certification.

https://en.wikipedia.org/wiki/HDR10

HDR10 color space is ITU-R BT.2020 that can also be written as REC2020. With certification/calibration performed SMPTE ST 2084 standard but that different to what REC2020 says should be used. Now we have a problem SMPTE ST 2084 allows for a larger margin of error than REC2020 certification/calibration method says is acceptable.

HDR400/600 are not certified to output srgb correctly. They are certified to output srgb as per PQ P3 that is srgb/reg709 rendered washed out because the 3 points have been moved in the color space.

If you are not use to seeing srgb rendered correctly you would think that PQ P3 is correct when it not comes very important when doing color matching and other things.

Basically HDR400 HDR600 if you test against pure REC2020 they are all wrong.

This leads us to a bigger problem. Movie industry like this incorrect form of REC2020 that uses SMPTE ST 2084. This means you cannot watch movies that are right and have srgb/Rec709 that is right at the same time. Srgb/Rec709 being washed out on monitor in HDR mode is in fact rendering correctly as per monitor certification in most cases.

Yes people complain that their movie rendered too dark in places on pure REC2020 certified monitor is also correct. Yes pure REC2020 is darker colors than the REC2020/SMPTE ST 2084 hi-bred in the center area due to how the bits align in the color space.

This is a calibration and color space problem. Yes Quackdoc if you are using HDR400 monitor and you are claiming srgb/REC709 is rendering correctly while HDR content is displayed like it or not you are talking out your ass those monitors are not in fact able todo that. You need to break out a color meter and correctly check you will find the srgb is values incorrect all in the direction of being too bright/washed out.

Problem is the DCI-P3 65 is not srgb/REC709. Now if you were claiming HDR400 monitor can do DCI-P3 65 content correctly you would be correct but do remember that is over bright and looks washed out to those who are use to looking at properly calibrated srgb/REC709 monitors because DCI-P3 65 is the wrong color space.

This is why we are checkmated. We need to accept for most HDR monitors we will not be able to display srgb/rec709 correctly while displaying HDR content because the monitor is incapable of doing it. If you have HDR monitor capable of displaying srgb/REC709 correctly while displaying HDR content the HDR content is going to be not displayed as intended in lots of cases. This is a true rock and hard place.

[mSparks]

it's important to remember, that the amount of luminance levels that you can actually visibly see, changes when you apply a better transfer.

Its important to remember that if you dont evenly space the quatisation levels, the dynamic range decreases because the minimum value you can measure/emit increases
e.g. lets say instead of 1 nit for the lowest value of 255:1 you use 0.3 nits
all of your other spacings now have to increase to 1.0028 nits per level and your dynamic range is now 255:1.0028 < 255:1
254*1.0028+0.3=

255:max(0.3,1.0028)

Just like perpetual motion, it doesn't matter how you try and fake it, you can't put more information into 8 bits than 255 levels, dynamic range and bits measure the same properties of information, but ones analogue, the other is digital.

That, btw, is exactly how you downscale the "real world" with generally a 1,000,000:1 dynamic range, to an SDR image with a 255:1 dynamic range, and why darks/lights end up "washed out" depending on which end of the scale you clip on.

[Quackdoc]

This is a mistake. Those are not color specification specs they are certifications specifications. REC2020 and REC2100 are the color spaces and what is encoded to go over the displayport/hdmi cable.

HDR10 and dolby PQ are both color spaces, HDR400/600 etc, get very close to being colorspaces, as HDR400 defines srgb support, 8bit color, brightness, and at least a PQ transfer curve, HDR500 requires 10bit, dci p3 of at least 90% or greater etc. I can send an sRGB gamut and a pq transfer without issues to my monitor and it displays them properly.

8bit LCD being used output HDR is in fact invalid by the color encoding. REC2020 is clear the coding is either 10 or 12 bits per color. Display port and HDMI standard is clear the encoding is 10 to 12 bits per color for HDR.

don't assume that REC2020 is the only HDR spec

Lot of monitors people call HDR are going to have viable issues they don't have the 10 or 12 bits of color per pixel. One of these issues is washed out SDR.

not it's not, if the HDR looks fine, it's a color management issue

That is not the case. Perfect graded monitor for REC2020 SDR appears almost right with Windows out the box on HDR screen without any adjustment so not washed out at all. Problem is it over 90 percent of monitors out there are not this.

you have a source for this? i've seen many well calibrated displays where this is not the case. on the contary, a color managed setup almost never looks wrong even when mixing sdr and hdr. in fact I would argue that if this was the case, then everyone involved in color is a bloody retard and needs to be fired

You are absolute right there are lot things called HDR. But once you get to what is being transmitted down cable and the color space that is meant to be there is not many things that are HDR.

where is the spec for this?

Do you know why there are so many different certifications for HDR monitors that allow garbage it simple. Vesa Dolby and so on get paid for every monitor that has their HDR logo on. Its in their best interest to make certificates total garbage HDR monitors can pass because it equals more dollars for them.

feel free to post an somewhat industry recognized spec for this. I have yet to see one. and I wont take opinion as fact

yeah, the "proper" measurement is
10log(80/0.31372549019) = 10log(255/1)

the "proper measurement" changes depending on what transfer you use since it influences what gets sent to the monitor. it's important to remember, that the amount of luminance levels that you can actually visibly see, changes when you apply a better transfer.

[oiaohm]

you guys are neither right first and fore most HDR is not a spec, there are multiple specs that provide an HDR like HDR10(+), Dolby, one of the various HDR400, 600, etc. not every spec shares the same necessities.

This is a mistake. Those are not color specification specs they are certifications specifications. REC2020 and REC2100 are the color spaces and what is encoded to go over the displayport/hdmi cable.

HDR10 in fact mandates the monitor can do 10 bit color as the REC2020 color space say you should be using.

8bit LCD being used output HDR is in fact invalid by the color encoding. REC2020 is clear the coding is either 10 or 12 bits per color. Display port and HDMI standard is clear the encoding is 10 to 12 bits per color for HDR.

But then you can get a certification certificate and call your monitor/tv HDR when the LCD is only 8bits per color with zoned backlights to attempt to hide the fact you are missing 4 bits of color control.

By what is valid to send over a HDMI or Displayport cable 8bit HDR does not exist.

this is not true AT ALL, SOME HDR specs require more then 8bit color depth, and you SHOULD have more then 8bits or else you can have visible issues, but many monitors allow you to trigger "HDR" modes using 8bit color

The note of many monitors allow you to trigger HDR modes using 8bit color is so correct. There are a lot of things claiming HDR that when you look at what is transmitted is not HDR. Monitor using 8bit color in HDR mode does not have correct REC2020 display.

Lot of monitors people call HDR are going to have viable issues they don't have the 10 or 12 bits of color per pixel. One of these issues is washed out SDR.

This is rarely ever the issue, even if you had a perfect grading monitor, unless you are properly using a color managed setup it will always look washed out.

That is not the case. Perfect graded monitor for REC2020 SDR appears almost right with Windows out the box on HDR screen without any adjustment so not washed out at all. Problem is it over 90 percent of monitors out there are not this.

Washed out is in fact side effect of the poor quality HDR monitors certification standard. Even vesa has made some total garbage level HDR certifications.

https://displayhdr.org/#tab-1400

Increased color gamut (95% DCI-P3 65) compared to all other current DisplayHDR tiers

I like this one from vesa documentation.

Not that hard to spot. Where is the certification color space "DCI-P3 65" that not REC2020 or REC2100 color space that is the data that should be traveling over your displayport or hdmi cable.. Of course being certified for "DCI-P3" instead of what in REC2020 is going to make all you SDR content correctly coded into REC2020 as per REC2020 washed out.

You are absolute right there are lot things called HDR. But once you get to what is being transmitted down cable and the color space that is meant to be there is not many things that are HDR.

DisplayHDR HDR400 to HDR1400 from vesa are different levels of garbage.

Do note the something else about that Vesa site the true black monitors stop at 600 nits. Yes another way to attempt fake up HDR with garbage hardware is increase the brightness that the monitor to attempt to burn your eyes out your head.

Good question how do you get 50 times the dynamic range with the same signal. You notice the DisplayHDR true black 400 from visa is mean to have 50x the dynamic range DisplayHDR 1000 that exactly what happened here where both monitors will be getting the same signal by hdmi or displayport. Yes 600 lower nits of output higher quality of output. True black testing starts getting close to REC2020 certification with the Vesa tags.

Yes very few of the monitor standard certifications in fact mandate that the monitor can do a decent job rendering REC2020 content. This leads to having to do color correction over a massively defective monitor.

Your general HDR monitor in HDR mode like it or not is garbage caused by.
1) Being certified against the wrong color space.
2) Using backlight to cover for lacking enough bit per color.

Do you know why there are so many different certifications for HDR monitors that allow garbage it simple. Vesa Dolby and so on get paid for every monitor that has their HDR logo on. Its in their best interest to make certificates total garbage HDR monitors can pass because it equals more dollars for them.

[mSparks]

again, post where it is defined in relation to video and light stimulus, keeping in mind that we are working with, and trying to emulate light, non a linear representation of it

yeah, the "proper" measurement is
10log(80/0.31372549019) = 10log(255/1)

"still equal"
The contrast ratio however is different, because the smallest value can be equally tiny, and the largest value grow as large as you have watts.

What I think is interesting, is now with VR - especially when eye tracking is more widespread - you can get very close to "real world" dynamic range by bypassing the compression in dynamic range you would normally get from your iris/pupil. but for displays in normal lighting, if you want to capture and display a high dynamic range, you need the bits to store and transfer the information, else you will always end up with either a florescent light as bright as the sun, or the sun no brighter than a florescent light.

[Quackdoc]

every signals course for like the last 200 years, but
e.g.

https://www.ophiropt.com/en/n/understanding-dynamic-range

So to correct myself from earlier.
if the smallest signal you can get is 0.3 nits, and the largest is 80nits the dynamic range 80/0.3 = 255:1

and if the largest is 255 and the smallest is 1 the dynamic range is 255/1=255:1

actual value might be out because dBs are a pita.

again, post where it is defined in relation to video and light stimulus, keeping in mind that we are working with, and trying to emulate light, non a linear representation of it

[mSparks]

where is this defined? and by what body?

every signals course for like the last 200 years, but
e.g.

https://www.ophiropt.com/en/n/understanding-dynamic-range

So to correct myself from earlier.
if the smallest signal you can get is 0.3 nits, and the largest is 80nits the dynamic range 80/0.3 = 255:1

and if the largest is 255 and the smallest is 1 the dynamic range is 255/1=255:1

actual value might be out because dBs are a pita.

[Quackdoc]

dynamic range is a general signals term that infers the quantization effects in a signal, it doesn't have a unit.

where is this defined? and by what body?

[mSparks]

you forget the fact that we aren't working in linear, and neither do our eyes when you compress the maximum luminance range, this means for a lot of people you are compressing the range of brightness we can actually physically see. Everything we do in color in the end is trying to mimic reality.

and the name for that is contrast ratio, not dynamic range.

dynamic range is a general signals term that infers the quantization effects in a signal, it doesn't have a unit.

But high dynamic range sounds cooler than high contrast ratio (because it is), and marketeers have never had a problem with liberal use of terms to sell their snake oil until they get hit with a class action.

We are mostly passed much of that now, HDR10 is a real thing, I learnt a lot of this from the guys and gals at Laminar Research who are now using 16bits per colour in their graphics engine.
nice write up of that here:

https://developer.x-plane.com/2021/11/photometric-lighting-what-is-it-and-why-do-we-need-it/

[Quackdoc]

255 steps of 0.3nits per step (0-80 nits), is identical dynamic range to 255 steps at 1nits per step (0-255nits)

What changes there is the contrast ratio
"The contrast ratio (CR) is a property of a display system, defined as the ratio of the luminance of the brightest shade (white) to that of the darkest shade (black) that the system is capable of producing."

you forget the fact that we aren't working in linear, and neither do our eyes when you compress the maximum luminance range, this means for a lot of people you are compressing the range of brightness we can actually physically see. Everything we do in color in the end is trying to mimic reality.