Final Explanation of Gamma and Color Shift Problems

Apologies for my ignorance, but I'm still confused about a central point. Let's say one has done everything right - graded on a hardware calibrated monitor fed by the signal from an appropriate BMD box, then applied the proper NCLC tags. Will the resulting video file look the same when played back on platforms subject to Apple color management and on windows platforms not subject to Apple color management? I know that people will be watching on all sorts of messed up screens in all sorts of viewing environments, there's nothing we can do about that, but put that aside. If we take the same screen, viewed in the same environment, and feed it a video file played by an Apple color managed app running on an Apple device, and then feed it the same video file running on a non-Apple device, will the image on the screen look the same both times?

Apologies for my ignorance, but I'm still confused about a central point. Let's say one has done everything right - graded on a hardware calibrated monitor fed by the signal from an appropriate BMD box, then applied the proper NCLC tags. Will the resulting video file look the same when played back on platforms subject to Apple color management and on windows platforms not subject to Apple color management? I know that people will be watching on all sorts of messed up screens in all sorts of viewing environments, there's nothing we can do about that, but put that aside. If we take the same screen, viewed in the same environment, and feed it a video file played by an Apple color managed app running on an Apple device, and then feed it the same video file running on a non-Apple device, will the image on the screen look the same both times?

giving the right tags, it will be "close enough".

giving the right tags, it will be "close enough".

And with Display color managed player on both Side
VLC isn't Display color managed by default ➧ client's VLC state

Display Color managed = Taking account of NCLC tag's to colormanage to Display "ColorSpace/Profile"

If color in not managed then pure fact that Mac displays are about P3 gamut gives you oversaturated colors.
It has to be color managed preview.
You can't have perfect preview without color management on the display as we are not anymore just within Rec.709 gamut. Content and displays use bigger gamuts today, so it's about impossible get 1:1 mapping without management.

giving the right tags, it will be "close enough".

My project was graded in a dim room on a monitor calibrated to Rec.709 Gamma 2.4 with a brightness of 100 cd/m2 (it also looks fine viewed in a brighter environment on the same monitor calibrated to Rec. 709 Gamma 2.2 with a brightness of 120 cd/m2).

Is there any consensus at this point about which tags would be properly read by the largest number of applications within both Apple and Windows environments to generate an image that wold be "close enough"? Is the best bet the 1-13-1 tag suggested by Danielle Siragusano, which I believe identifies an SRGB image with 2.4 gamma? My understanding is that he recommends this tag because it will look correct in color-managed applications, and is also more likely to look ok in non-color-managed applications since most screens out in the world are SRGB.

1-13-1 as any other tag is basically pointless as YT will overwrite it to 1-1-1 anyway.
Grade to calibrated screen to 2.4 or 2.2, export with standard 1-1-1 tag and that's it. No much more you can do if you want "global settings".
On Mac you can suggest to users to watch it over copy of ref BT.1886 mode with brightness which suits them (100nit is way too dim for home use).
I don't think most screen today are sRGB. TVs won't be, Macs are not and no idea about PCs.

Will the resulting video file look the same when played back on platforms subject to Apple color management and on windows platforms not subject to Apple color management?

Due to Apple's color managed interpretation of Rec.709 tagged video and the lack of comparable color management in SDR on the PC/Android side, it is not possible to get a Mac and a PC to deliver the exact same image on YouTube. As others have noted, getting relatively close is as good as it gets. In general, the same YouTube image on macOS/iOS displayed via a color managed browser/app, is generally going to appear with slightly higher gamma and slightly less saturation relative to its appearance on an unmanaged sRGB display on PC.

1-13-1 as any other tag is basically pointless as YT will overwrite it to 1-1-1 anyway.

Thanks Andrew! But I'm primarily interested in commercial streaming platforms other than YouTube. Are other platforms likely to overwrite or ignore a 1-13-1 tag? I'm thinking not only of the bigger ones like Netflix, Amazon Prime, and Tubi, but also smaller niche streamers.

In general, the same YouTube image on macOS/iOS displayed via a color managed browser/app, is generally going to appear with slightly higher gamma and slightly less saturation relative to its appearance on an unmanaged sRGB display on PC.

Thanks Jamie! However, I thought it was the other way round. If I compare the same video file on my Mac played on the QuickTime viewer (color managed by Apple Color Sync) and played on VLC (not color managed), it appears a bit darker and with significantly higher contrast when viewed on VLC. Apologies if I'm confused; you know a lot more about this stuff than I do.

Thanks Jamie! However, I thought it was the other way round. If I compare the same video file on my Mac played on the QuickTime viewer (color managed by Apple Color Sync) and played on VLC (not color managed), it appears a bit darker and with significantly higher contrast when viewed on VLC. Apologies if I'm confused; you know a lot more about this stuff than I do.

What you see in VLC on macOS on Apple's display hardware is not what you would see on a PC. But even then, what you are seeing tracks — the same Rec.709 tagged file when color managed on macOS appears brighter and less saturated than when it is not color managed on the same display (though some of the extra saturation in the case of comparing managed to unmanaged on macOS is also a result of Apple displays being P3, which adds to the appearance of saturation when Rec.709 is displayed unmanaged).

Although these videos have already been posted more than once in this thread, I will post them again and highly recommend watching both as they are the clearest and most detailed explanation of SDR video display on both macOS and unmanaged PC/Android.

1-13-1 as any other tag is basically pointless as YT will overwrite it to 1-1-1 anyway.
Grade to calibrated screen to 2.4 or 2.2, export with standard 1-1-1 tag and that's it. No much more you can do if you want "global settings".
On Mac you can suggest to users to watch it over copy of ref BT.1886 mode with brightness which suits them (100nit is way too dim for home use).
I don't think most screen today are sRGB. TVs won't be, Macs are not and no idea about PCs.

I will follow this advice: 1-1-1, and forget about it.

Someone once told me: “if Apple, Disney and netflix did not fix this problem, you at home are not going to fix it.”

Someone once told me: “if Apple, Disney and netflix did not fix this problem, you at home are not going to fix it.”

I completely agree that as individuals we have little hope of solving all this.
As for Disney, Apple and Netflix, those companies don't have to worry about even trying to fix web video because they have their own mobile/TV apps and other OTT delivery systems to be able to control the image on consumers screens. As far as I've seen, those apps deliver a correct image on Apple devices.
It is those delivering only (or mainly) to YouTube who get bitten by these problems of NCLC tags and web/computer video.

What % watches Netflix etc. on TVs vs laptops?
I don’t even have TV in my new house

What % watches Netflix etc. on TVs vs laptops?
I don’t even have TV in my new house

I still do, i have a 65” but i will get a 77” with full 5.1 sound…

It’s nice, couch, beer and all…

What % watches Netflix etc. on TVs vs laptops?
I don’t even have TV in my new house

According to the most recent stats I could find, an overwhelming majority are watching Netflix on a TV, with the remainder a roughly even split between phones/tablets vs desktops/laptops.
https://truelist.co/blog/netflix-statistics/

On that browser based viewing, I have to run a test to be sure, but it looks to me like Netflix is adding a transform to the stream sent to the browser to account for the difference between BT.1886 and the 2.2 gamma of virtually all computer displays, including Apple's. That doesn't fully account for the transform that ColorSync applies to 1-1-1 input, but its not near as much difference as what happens when people are managing for 2.4 gamma in the Resolve GUI and viewing that image directly against 1-1-1 on YouTube in a browser on Apple devices.

I still do, i have a 65” but i will get a 77” with full 5.1 sound…

It’s nice, couch, beer and all…

As it should be. Right on. For the 77" are you going with tried and true WOLED, or with one of the newer QD-OLED?

What % watches Netflix etc. on TVs vs laptops?
I don’t even have TV in my new house

I still do, i have a 65” but i will get a 77” with full 5.1 sound…

It’s nice, couch, beer and all…

TV lies and there is about nothing new and nice on Netflix

I still do, i have a 65” but i will get a 77” with full 5.1 sound…

It’s nice, couch, beer and all…

As it should be. Right on. For the 77" are you going with tried and true WOLED, or with one of the newer QD-OLED?

Not sure yet, I like the new sony a95L, but also the C3 LG are really good…

TV lies and there is about nothing new and nice on Netflix

Did you tune in FoxNews?

News? It's not news anymore- it's politics
There is a news and on every channel it sounds differently based on who supports/rules given station.
I'm sure USA isn't any different, neither is BBC etc. It's all regime TVs, just a modern version

Life without a TV is better place

Sure, with all that YT crap?

I see us all going to get OT, so we can soon celebrate the 20th page of this thread with the word "final" in its title

(Because we can't let this thread die out until it reaches at least 20 pages...)

As mentioned previously, I'm preparing to export a feature-length documentary to a streaming aggregator. I obviously have no control over or knowledge of how the resulting file will be viewed - via a web browser or via a proprietary app, on a PC, on a Mac, on a TV, etc etc.

Having spent a lot of time with this thread, after grading on an Eizo monitor hardware calibrated to Rec 709 Gamma 2.4 viewed in a dim room and fed by an Ultra Studio Monitor 3G box, I'm planning to simply export a ProRes 422HQ Rec 709 Gamma 2.4 file tagged 1-1-1. It will look like cr*p when viewed on any Mac application managed by Apple ColorSync, but that's just the way it goes. Before I send out the file however, I wanted to check with more knowledgeable folks that the discrepancy I'm seeing between the file played via Quicktime on a Mac and the file played via Resolve or virtually any PC application is no worse than what you'd expect. If there does seem to be a bigger discrepancy than usual, then there may be a problem with my workflow.

Here's a screen shot from the exported file played on Quicktime on my Mac. It appears comparatively washed out as one would expect:

Screen Shot from exported file played via Quicktime on a Mac

Screen Shot 2024-04-11 at 5.20.16 PM.jpg (711.99 KiB) Viewed 4080 times

And here's a screenshot of the same file played in Resolve. This matches almost exactly what I see through the UltraStudio 3G box into my reference monitor, as well as when I view the file on a PC plugged into my reference monitor using Windows 11 Media Player or uploaded to Youtube and viewed through any Windows browser.

Screen Shot from exported file played via Resolve on a Mac

Screen Shot 2024-04-11 at 11.41.27 PM.jpg (662.71 KiB) Viewed 4051 times

It's a bit hard to tell given the small sizes of the images and the bright white surround (and yes the foreground figures are supposed to be in near-silhouette), but does the the QuickTime screen shot look flatter than you would expect, or would you say the difference is about normal?

Thanks for any feedback!

Thanks to everyone who contributed to this 563-post thread. It seems clear there is not a perfect solution, but there are at least settings that can allow you to get the Resolve preview and rendered file in Quicktime on Mac to match (even if it may look different on Windows). That's good enough for me for current needs.

But my question is what should you do if you have a feature-length movie that is already graded under Resolve default settings (Rec 709 Gamma 2.4, with "Mac display color profile" un-checked)? If I follow any of the suggested approaches here it would require re-grading 1000 shots to get back to the look I had. In the end, I applied a suggestion I found elsewhere:

Select P3-DCI color space tag and Rec 709 gamma tag in render settings. As I understand it, this is basically compensating for the fact that I did the grading on a P3 display on Mac without "Mac display color profile" checked. The results are pretty darn good compared to other things I've tried, but not perfect. It's a little brighter, less rich. Any suggestions are much appreciated it!

Hi I wonder if you have found any better way than this since this post? I also found this workaround from a Youtube comment, and it's been the closest thing I can get to things looking similar in Resolve and in VLC and Quicktime etc. Things still look less saturated and less contrasty in Quicktime but it's better

Are you still using this method of choosing P3DCI for Color space tag and Rec709 for Gamma tag in the advanced settings when on the deliver page? I was using Rec709-a, but not sure if I should. I tested both and saw no difference

My main concern is how it looks on the web, Vimeo, Youtube etc

Thanks

(Because we can't let this thread die out until it reaches at least 20 pages...)

As mentioned previously, I'm preparing to export a feature-length documentary to a streaming aggregator. I obviously have no control over or knowledge of how the resulting file will be viewed - via a web browser or via a proprietary app, on a PC, on a Mac, on a TV, etc etc.
....

If you have Mac with reference profiles then once you choose BT.1886 reference profile preview, YT/Vimeo will look correct. You can also copy this profile and then have control over brightness.

On that browser based viewing, I have to run a test to be sure, but it looks to me like Netflix is adding a transform to the stream sent to the browser to account for the difference between BT.1886 and the 2.2 gamma of virtually all computer displays, including Apple's. That doesn't fully account for the transform that ColorSync applies to 1-1-1 input, but its not near as much difference as what happens when people are managing for 2.4 gamma in the Resolve GUI and viewing that image directly against 1-1-1 on YouTube in a browser on Apple devices.

Hello everyone

I can confirm that Netflix in ColorSync menaged browser compensates for the difference between BT.1886 and the Apple P3 sRGB screen. I had the opportunity to grade one of its productions. The print screens I got from Chrome were identical in terms of image tonality with the preview files tagged as (1-13-1), sRGB transfer function. They only differed slightly in color, Netflix was little bit more greenish that what I graded. This difference was only visible side by side.

It would be way easier if they simply changed "decoding" math for 1-1-1 tag

That would be a real miracle!

It would be way easier if they simply changed "decoding" math for 1-1-1 tag

... but then we wouldn't have had a chance to soon celebrate a 5 year anniversary of this thread

It would be way easier if they simply changed "decoding" math for 1-1-1 tag

... but then we wouldn't have had a chance to soon celebrate a 5 year anniversary of this thread

On that browser based viewing, I have to run a test to be sure, but it looks to me like Netflix is adding a transform to the stream sent to the browser to account for the difference between BT.1886 and the 2.2 gamma of virtually all computer displays, including Apple's. That doesn't fully account for the transform that ColorSync applies to 1-1-1 input, but its not near as much difference as what happens when people are managing for 2.4 gamma in the Resolve GUI and viewing that image directly against 1-1-1 on YouTube in a browser on Apple devices.

Hello everyone

I can confirm that Netflix in ColorSync menaged browser compensates for the difference between BT.1886 and the Apple P3 sRGB screen. I had the opportunity to grade one of its productions. The print screens I got from Chrome were identical in terms of image tonality with the preview files tagged as (1-13-1), sRGB transfer function. They only differed slightly in color, Netflix was little bit more greenish that what I graded. This difference was only visible side by side.

Greenish tint sometimes happens during fast/poor 10 -> 8 bit conversion.

... but then we wouldn't have had a chance to soon celebrate a 5 year anniversary of this thread

Five years and how many hundreds of messages? It's endless!

Newcomers and younger people don't believe it, but we had these problems in the 1970s, 1980s, and 1990s, mostly because we couldn't convince clients to calibrate their displays. NOW, it's possible to provide clients with at least something "resembling" a calibrated monitor, even if it's just an $800 iPad Pro. That at least is in the ballpark of reality. Their $300 cellphone will not be even close.

Quite a few professionals over here do that as well.
Definitely helps to show how far off other screens can be.

On that browser based viewing, I have to run a test to be sure, but it looks to me like Netflix is adding a transform to the stream sent to the browser to account for the difference between BT.1886 and the 2.2 gamma of virtually all computer displays, including Apple's. That doesn't fully account for the transform that ColorSync applies to 1-1-1 input, but its not near as much difference as what happens when people are managing for 2.4 gamma in the Resolve GUI and viewing that image directly against 1-1-1 on YouTube in a browser on Apple devices.

Hello everyone

I can confirm that Netflix in ColorSync menaged browser compensates for the difference between BT.1886 and the Apple P3 sRGB screen. I had the opportunity to grade one of its productions. The print screens I got from Chrome were identical in terms of image tonality with the preview files tagged as (1-13-1), sRGB transfer function. They only differed slightly in color, Netflix was little bit more greenish that what I graded. This
difference was only visible side by side.

Can someone confirm my observation?
I also invite you to Liftgammagain, where I created a thread related to this topic.
"macOS - previewing SDR content files"

... but then we wouldn't have had a chance to soon celebrate a 5 year anniversary of this thread

Five years and how many hundreds of messages? It's endless!

Newcomers and younger people don't believe it, but we had these problems in the 1970s, 1980s, and 1990s, mostly because we couldn't convince clients to calibrate their displays. NOW, it's possible to provide clients with at least something "resembling" a calibrated monitor, even if it's just an $800 iPad Pro. That at least is in the ballpark of reality. Their $300 cellphone will not be even close.

Problem is elsewhere.
First you need a "technology" which doesn't mess with your files. Then we can start worry how to provide better "screens".
For that long industry keeps ignoring simple problem which could be solved quite easily. Problem is that there is no money to be made, so no one bothers.

... but then we wouldn't have had a chance to soon celebrate a 5 year anniversary of this thread

Five years and how many hundreds of messages? It's endless!

As they say, time (and posts) fly when you are having fun ... Ummm ... We _ARE_ having fun, right? Right?

Editors should NEVER use the word "final" to describe ANYTHING! This topic is proof!

Although these videos have already been posted more than once in this thread, I will post them again and highly recommend watching both as they are the clearest and most detailed explanation of SDR video display on both macOS and unmanaged PC/Android.

[/quote]

They are still totally legit though. I'm sending them to everybody who has a problem with this

Although these videos have already been posted more than once in this thread, I will post them again and highly recommend watching both as they are the clearest and most detailed explanation of SDR video display on both macOS and unmanaged PC/Android.

They are still totally legit though. I'm sending them to everybody who has a problem with this [/quote]


Those are essential for me, of course.

For those who still don't believe, I prepared a little exercise.
I uploaded 3 video files to Frame.io with different combinations of NCLC tags, which are usually used in SDR workflow.
When you open the frame.io link, by default all of these files will look the same. The actual tags were overwritten by (1-1-1), the standard Rec.709 profile, which on macOS manifests itself with the famous "washed out" look.
That happens with this kind of uploads, eg. YouTube, Vimeo, of course.
Now switch the resolution of video in Fame.io player and set it to Original.
Analyze the difference, than watch 3 other folders with print screens made of all 3 video files in their Original state (resolution).
Those print screens were made under control of 3 different operational system.
Tell me please which one of the video files corresponds the best to the rest of the pictures - print screens.
Also tell me please, why we should not use (1-2-1) taged files on macOS system, I mean 2.4 gamma.
Do not comment on Frame, please.

This exercise is only valid for macOS systems!
Standard Display P3 ICC profile.

https://f.io/34H8g9rb

Ok, guys. My exercise, maybe it wasn't clear enough.
So, I've found the great way do demonstrate to a client, how his video should look like and how it will look on macOS -YouTube, Vimeo and some others online platforms. It is only valid for the client who is watching this on macOS, ColorSync managed browsers - Safari, Chrome, etc.
With just a single Frame.io review link!

On Resolve, Deliver Page, chose Gamma Tag - sRGB. It will result with (1-13-1) NCLC tag for your file.
Upload file to Frame.io platform and grant permission to download this. It is essential for viewing video with proper metadata NCLC tags there.

Send your client a review link and explain that this is how the image will look on YouTube, Vimeo on MacOs.
Then asked him to change default Frame player resolution (probably 1080p) to Original (lower right corner).
Explain to him that this is how the image will look like on broadcast, YouTube, Vimeo on Windows, Android (probably) and iOS devices.
It will look like this on Netflix on MacBooks also.

To prove this, I share 3 directories with print screens taken on macOS, iOS and Windows in Frame.io. Images from iOS and Windows look identical to a correctly tagged sRGB(1-13-1) file on macOS Frame player, in its Original quality setting.

Could someone check how it works on Android?

https://f.io/34H8g9rb

I'm also curious about what your grades look like regarding NCLC tags, on Netflix, Disney, Max, Prime, etc.
On macOS managed browser, of course. Are they "washed out", or not?

Guys you are chasing ghosts here.

Let's say I visit 5 different friends, and watch the same clip on their computer/TV(Netflix), tablet, phone ... whatever they have - it will NEVER look the same - not even close.

Way to many different (and not calibrated) displays, to many different display technologies, to many different individual settings (that are probably way off).

And there is nothing you can do about it - get over it.

For the record - screencaptures mean nothing, they are captured way before the signal hits the display.
They are not actually "screencaptures", unless you take a photo of your screen, and that's an whole other can of worms.

You seems to don't understand the issue.
It has been stated many times that screen accuracy has nothing to do with this problem.
Screens accuracy is a totally different subject and can be discussed once this is fixed.

If you're talking about screen accuracy/preview difference then you don't need to go to your fiends.
Put two different brand 30K$ reference screens in one room and they won't look the same In some case they will look like day and night regardless of 5h calibration proces done for both.

Gee, just turn the knobz on monitor until they match! Its called calebriton, look it up!
Any relative match can be achieved with appropriately mismatched monitors.

It has been stated many times that screen accuracy has nothing to do with this problem.

Jokes aside, this ^

No idea what you mean.
You can use 10K probe and both will show you perfect calibration yet they will look different to the eye. In some cases it will be very visible difference in others very subtle.
Welcome in real world
In studio an A grade ref screen is treated as correct, so some studios never put 2 different ref screens in the same room.

You can use 10K probe and both will show you perfect calibration yet they will look different to the eye. In some cases it will be very visible difference in others very subtle.

That will be metameric failure.....


when you calibrate a display, there is little expectation of a 100.00000% match in a side by side environment (even just the factor that you might be at a different angle and/or distance, will show), they are close within a reasonable tolerance.

I always make my engineer mad by putting two calibrated X300 in black and white and they never perfectly match. But by the same token, I understand tolerances and cut some slack...

You seems to don't understand the issue.

Since I'm doing this since over 30 years now, I would say I totally understand the issue.
Back in the CRT days it was pretty easy. You had a class A monitor and your scopes sitting on your desk, and that was it. Since there was only one display technology everybody saw pretty much the same image (unless they did screw with their TV settings).

I also think we solved the so called "gamma shift" problem a 100 times already.
So what's left is different displays/players/technologies at your customers. Unless you send them a locked down iPad, there is nothing you can do. And still .. all the end users with their wacky displays will see something different.

It's just the nature of the beast - you can only do so much.

With a well calibrated monitor and a Decklink, make sure it looks good on your side, everything else is out of your control anyway.

And your client for a TVC may still say that the color from his/her CI (corporate identity) doesn't look right. So, it's a pretty good idea to regularly, like every 3 months or so, do a calibration and keep the protocol. Of course, this should be for the display that your client saw when signing for the final result.

It's pretty much the only way you can prove to their lawyers that you didn't screw up.

You can use 10K probe and both will show you perfect calibration yet they will look different to the eye. In some cases it will be very visible difference in others very subtle.

That will be metameric failure.....


when you calibrate a display, there is little expectation of a 100.00000% match in a side by side environment (even just the factor that you might be at a different angle and/or distance, will show), they are close within a reasonable tolerance.

I always make my engineer mad by putting two calibrated X300 in black and white and they never perfectly match. But by the same token, I understand tolerances and cut some slack...

Funny that white+ black + grey are great way for checking color monitor

Guys you are chasing ghosts here.

Let's say I visit 5 different friends, and watch the same clip on their computer/TV(Netflix), tablet, phone ... whatever they have - it will NEVER look the same - not even close.

Way to many different (and not calibrated) displays, to many different display technologies, to many different individual settings (that are probably way off).

And there is nothing you can do about it - get over it.

For the record - screencaptures mean nothing, they are captured way before the signal hits the display.
They are not actually "screencaptures", unless you take a photo of your screen, and that's an whole other can of worms.

"screencaptures mean nothing, they are captured way before the signal hits the display."

I don't agree with that statement. Could you elaborate?
I'm pretty sure that they are captured after some ColorSync or any other color management math operation.
Actual pixels values could be altered or not, and these screenshots show it clearly.

I don't agree with that statement. Could you elaborate?
I'm pretty sure that they are captured after some ColorSync or any other color management math operation.
Actual pixels values could be altered or not, and these screenshots show it clearly.

Just try it. Change you monitor settings to something totally out of whack (like totally blue and contrasty), than do a screen capture, it will still look fine, cause it is captured before it hit's your monitor.

Can we have a party on October 23rd to celebrate five years of Gamma/Color Shift confusion? (And again, note this is not a Resolve problem -- it's a problem of different operating systems, different players, and lack of calibration.)