NCLC tag labeling mechanism of SCRATCH

90 views
Skip to first unread message

thom jiji

unread,
Oct 14, 2021, 8:15:32 AM10/14/21
to SCRATCH forum
Hi all,

Would like to know what's the NCLC tag labeling mechanism of SCRATCH.

I'm transcoding some clips to H265 mov. Some of that transcoded H265 MOVs are complete 1-1-1 tags, Color Primaries, Transfer Characteristics, and Matrix Coefficients are BT.709. But some of the clips are missing Transfer Characteristics. My initial test today found that when the main output node color space is AP0, the rendered mov is 1-1-1, while when the main output node color space is Rec709 Gamma2.4, the rendered mov is missing Transfer Characteristics.

Want to know how I can properly add the 1-1-1 NCLC tag on SCRATCH rendering.

Thanks a lot!
thom

Guido

unread,
Oct 14, 2021, 8:49:27 AM10/14/21
to SCRATCH forum
In the Advance System Settings you can turn on "QuickTime: No gamma value" setting, that way the Rec709/gamma2.4 will be tagged as 1-1-1 instead of adding an exact gamma value atom.

Guido

Op donderdag 14 oktober 2021 om 14:15:32 UTC+2 schreef hxy4...@gmail.com:

Mazze Aderhold

unread,
Oct 14, 2021, 9:13:25 AM10/14/21
to scratc...@googlegroups.com
To add to Guido's reply:

tagging Quicktime files as 1-1-1 is in most cases incorrect, because image is not encoded
with a Rec709 curve, but most of the time gamma 2.4.
In order to write that value to the Quicktime file, you have to set the nclc tag to 1-2-1,
with "2" being "unkown". In addition to that, you have to write another atom, called the 'gama' atom
and set that to "2.4".

The resulting file will look exactly the same as in SCRATCH, when opened with Quicktime Player
or Safari on macOS. On Windows and browsers like Firefox, Chrome or players like VLC it does not
matter at all how the nclc tags are written.
So actually, the transfer characteristics are not "missing" - they are even more accurate this way ;-) .

Some ingest systems however might require that the nclc tags are 1-1-1 - even though technically not correct.
For this we have this advanced setting that Guido mentioned.

Now when flagging the output node as "ACES AP0":
Apart from it being a terrible idea to try and write ACES to an integer file format,
there is actually no nclc tag reserved for ACES. You cannot tag it.
Same as proprietary color spaces like Alexa Wide Gamut or RED Wide Gamut - there is no tag for those either.
In these cases, the best way is to indeed write 1-1-1 - and with that,
it will look at least close to what's expected on macOS systems.

But I'd love to know why you need the files to be tagged as 1-1-1, specifically?


Thanks,
Mazze




> Guido <guido.v...@assimilateinc.com> hat am 14.10.2021 14:49 geschrieben:
> --
> ASSIMILATE homepage - www.assimilateinc.com
> SCRATCH web - www.scratchweb.tv
> Facebook - https://www.facebook.com/assimilateinc
> Twitter - @assimilateinc
> ---
> You received this message because you are subscribed to the Google Groups "SCRATCH forum" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to scratch-list...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/scratch-list/029ee906-8d9c-46b9-aeae-b9777c314c5an%40googlegroups.com (https://groups.google.com/d/msgid/scratch-list/029ee906-8d9c-46b9-aeae-b9777c314c5an%40googlegroups.com?utm_medium=email&utm_source=footer).
>

alwes...@gmail.com

unread,
Oct 14, 2021, 5:43:13 PM10/14/21
to SCRATCH forum

So if I understand well, that means for example that by default, every h264 outputted as Rec709, gamma 2.4 from Scratch will be tagged as 1-2-1 (with the extra 2.4 gama tag)... and so displayed correctly on Quicktime player and Safari on Mac OS... as well as on VLC in Windows ? (I guess there will be some difference between "color managed" and "non color managed software" ?)

Mazze Aderhold

unread,
Oct 15, 2021, 11:58:34 AM10/15/21
to scratc...@googlegroups.com
Yes, in a nutshell that's correct.
One more variable in the equation is of course macOS' own color managment.
If that is not set up correctly, then all bets are off ;-) .


Greetz,
Mazze



> alwes...@gmail.com <alwes...@gmail.com> hat am 14.10.2021 23:43 geschrieben:
> > > ASSIMILATE homepage - www.assimilateinc.com (http://www.assimilateinc.com)
> > > SCRATCH web - www.scratchweb.tv (http://www.scratchweb.tv)
> > > Facebook - https://www.facebook.com/assimilateinc
> > > Twitter - @assimilateinc
> > > ---
> > > You received this message because you are subscribed to the Google Groups "SCRATCH forum" group.
> > > To unsubscribe from this group and stop receiving emails from it, send an email to scratch-list...@googlegroups.com.
> > > To view this discussion on the web visit https://groups.google.com/d/msgid/scratch-list/029ee906-8d9c-46b9-aeae-b9777c314c5an%40googlegroups.com (https://groups.google.com/d/msgid/scratch-list/029ee906-8d9c-46b9-aeae-b9777c314c5an%40googlegroups.com?utm_medium=email&utm_source=footer).
> > >
>
> --
> ASSIMILATE homepage - www.assimilateinc.com
> SCRATCH web - www.scratchweb.tv
> Facebook - https://www.facebook.com/assimilateinc
> Twitter - @assimilateinc
> ---
> You received this message because you are subscribed to the Google Groups "SCRATCH forum" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to scratch-list...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/scratch-list/1df7f00e-a729-4b8b-8fdd-4468af1e11f0n%40googlegroups.com (https://groups.google.com/d/msgid/scratch-list/1df7f00e-a729-4b8b-8fdd-4468af1e11f0n%40googlegroups.com?utm_medium=email&utm_source=footer).
>

thom jiji

unread,
Oct 19, 2021, 3:45:07 AM10/19/21
to scratc...@googlegroups.com

Hi Mazze,

Sorry for the late reply.

I think I may have misunderstood. Thanks for your detailed explanation. I did a lot of testing over the last few days, and like you said, if the NCLC tag value written is 1-2-1, then I get the same result in QuickTime and SCRACH (but here I have a new problem, more on that later), and if the NCLC written is 1-1-1, then it plays back in QuickTime much more greyer than SCRATCH. (Why I think it should be tagged as 1-1-1? I remember seeing the original suggestion of the 1-1-1 tagging thing as a solution to the problem of matching QuickTime to previews when using the Rec.709-A timeline colour space in DaVinci Resolve? But there's no Rec.709-A in SCRATCH, so there's no such thing.) I think this may be due to that the Gamma or transfer function of BT.709 is actually 1.96 or something like that? So it is technically incorrect to tag a Rec709 Gamma 2.4 as 1-1-1 which is BT.709. 😷

I have a new problem that has been bothering me for days. First, is it right to set monitor viewing colour space and EOTF as P3D65 Gamma 2.2 when I use MBP? I'm using a Mac mini with a sRGB display which is calibrated by DisplayCAL. The viewing colour space should be set to the colour space of the monitor you are using, right? But when I set it to the monitor's colour space which is sRGB Gamma 2.2, the colours in SCRATCH become very saturated. So I set to others, finding which colour space setting looks right for me. I set the normal and standard one: Rec709 Gamma 2.4, too saturated. At the same time I opened DaVinci Resolve as reference, using the same ACES configuration and DRT, and checked one of DaVinci's settings: Use Mac display colour profiles for viewers. It turns out P3D65 Gamma 2.2 is the right viewing colour space choose in my Mac mini SCRATCH, though it is a sRGB display. But when I export a ProRes 422 QuickTime file from SCRATCH, then open it in QuickTime player. It has a hue shift, and a little bit blacks increased.

I have tried to compare it with Baselight. When SCRATCH is set to P3D65 Gamma2.2, it matches to Baselight when viewing colour space set to Apple: 2.2 Gamma / P3 D65. (Baselight is in ACES workflow, working color space is ACEScct / AP1) But they are both different from DaVinci and QuickTime. DaVinci is just the same as QuickTime file exported from SCRATCH. Maybe it has something to do with Apple's global colour management and the color profile that DisplayCAL created? I’ve tried a lot of testing and comparison. It still doesn’t match to my ProRes 1-2-1 exports.

Here are screenshots of my settings.

SCRATCH main output node set to ACES AP0, then I put a transform node appended to, and set that transform node to Rec709 Gamma 2.4. Finally, I put ProRes encoder node and rendered it.

Thanks,

thom


Mazze Aderhold <ma...@colormeup.de> 于2021年10月14日周四 下午9:13写道:
4_Baselight Color Space Journey.png
3_Resolve ACES setup.png
1_SCRATCH preview-compressed.png
2_QuickTime player-compressed.png

Mazze Aderhold

unread,
Oct 21, 2021, 12:20:11 PM10/21/21
to scratc...@googlegroups.com
Hi Thom,

so this will be a bit difficult to troubleshoot.
But generally, if your display is sRGB, then you should also flag the monitors in SCRATCH as such.
It also depends on what your display is set to in the macOS display calibration settings.
If for some reason it's set to P3D65 there, then it will of course do a transform to that.
However, if your display cannot display P3D65... then it will show incorrect colors.
Generally for SCRATCH: It bypasses macOS color management. There is however a catch to it.
If you change the display color space in the macOS system settings while SCRATCH is running,
you'll see pretty dramatic changes of the entire UI. However, if you then restart SCRATCH,
it will be back to normal. So if you're experimenting with macOS color management, please only
change its settings while SCRATCH is closed and then compare.

Next, you should use a Rec709 test chart that you know exactly what it looks like and then load that into
the various different softwares and just display it without any color management.
Make sure that the display is set correctly to what it actually can display (sRGB) in macOS' settings.
Also remove the calibration LUT - make the whole setup is dumb and simple with as less variables as possible.
Then see if you're seeing the same image in all 3 apps.
Then try the whole rendering part - again, without any color management whatsoever.
Next, add the individual CMS of each software into the mix.
And lastly throw in the calibration LUT.

That will then tell you where the culprit lies.


Cheers,
Mazze


> thom jiji <hxy4...@gmail.com> hat am 19.10.2021 09:44 geschrieben:
> > > Facebook - https://www.facebook.com/assimilateinc
> > > Twitter - @assimilateinc
> > > ---
> > > You received this message because you are subscribed to the Google Groups "SCRATCH forum" group.
> > > To unsubscribe from this group and stop receiving emails from it, send an email to scratch-list...@googlegroups.com.
> > > To view this discussion on the web visit https://groups.google.com/d/msgid/scratch-list/029ee906-8d9c-46b9-aeae-b9777c314c5an%40googlegroups.com (https://groups.google.com/d/msgid/scratch-list/029ee906-8d9c-46b9-aeae-b9777c314c5an%40googlegroups.com?utm_medium=email&utm_source=footer).
> > >
> >
> > --
> > Facebook - https://www.facebook.com/assimilateinc
> > Twitter - @assimilateinc
> > ---
> > You received this message because you are subscribed to the Google Groups "SCRATCH forum" group.
> > To unsubscribe from this group and stop receiving emails from it, send an email to scratch-list...@googlegroups.com.
> > To view this discussion on the web visit https://groups.google.com/d/msgid/scratch-list/859160459.4232382.1634217202711%40communicator.strato.de.
> >
>
> --
> ASSIMILATE homepage - www.assimilateinc.com
> SCRATCH web - www.scratchweb.tv
> Facebook - https://www.facebook.com/assimilateinc
> Twitter - @assimilateinc
> ---
> You received this message because you are subscribed to the Google Groups "SCRATCH forum" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to scratch-list...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/scratch-list/CAD75tOWmm0dPW1Fdz7khSt-5Ye3W5OMK5SP7HGBQEF47pFsJsw%40mail.gmail.com (https://groups.google.com/d/msgid/scratch-list/CAD75tOWmm0dPW1Fdz7khSt-5Ye3W5OMK5SP7HGBQEF47pFsJsw%40mail.gmail.com?utm_medium=email&utm_source=footer).
>
Reply all
Reply to author
Forward
0 new messages