Blackmagic Forum

I am asking because the specifications on the X-rite support page for the Color Checker are wrong. All Color Checkers manufactured from November 2014 have patches with a new color formula. But no-one at X-rite cared to update the support page of the CC Passport.

Here X-rite talks about the new color specifications:
http://xritephoto.com/ph_product_overvi ... 4&catid=28

And here you can find the new CIELAB values for Colorchecker SG:
http://xritephoto.com/documents/apps/pu ... _l_a_b.txt

The Color Checker SG patches E2:E5 to J2:J5 correspond to CC PP 1:24.

You can convert the CIELAB values to RGB values here:
http://davengrace.com/cgi-bin/cspace.pl

For example:
Patch #15 red : was 175/54/60 (old) and is now 179/0/27 (new).

Code: Select all

1. dark skin  ... 115/82/68   vs. 111/63/41
2. light skin ... 194/150/130 vs. 199/140/120
4. foliage    ... 87/108/67   vs. 78/96/37
13. blue      ... 56/61/150   vs. 0/42/135
14. green     ... 70/148/73   vs. 55/141/52
15. red       ... 175/54/60   vs. 179/0/27
16. yellow    ... 231/199/31  vs. 245/194/0
17. magenta   ... 187/46/149  vs. 188/68/142
18. cyan      ... 8/133/161   vs. 0/130/164

So the question is, which values for the CC PP does Resolve use? The old ones, or the new ones?

I have a Colorchecker Passport with the old color patches, a Colorchecker Passport Video with different (muted) patches, and a Colorchecker SG with new color patches.
Comparing the old CC PP to the new CC SG the new primary color patches are more vibrant than the older ones.

So, no one has an answer to this?
Should I better contact BMD support for this?

My advice is to avoid the Color Checker and instead use the DSC charts. The problem with the X-Rite charts is that you can't rely on them for accurate saturation levels.

Note that it is possible to color-correct dailies completely by hand and without the use of charts entirely, as we did for many decades with film and digital. "In my day," we were lucky to just get a grayscale chart to at least give us an idea of what neutral white, gray, and black were.

Thanks for your answer Marc.
The question is, why can't you rely on their saturation levels? Maybe because there is a data mismatch (old vs new)?

Regarding the old days: I started with CC in the late 80ies and still do most of my work with curves. So no worry, I know how to bend colors to my like without needing luts or color targets

Marc Wielage wrote:My advice is to avoid the Color Checker and instead use the DSC charts. The problem with the X-Rite charts is that you can't rely on them for accurate saturation levels.

yes -- i also prefer other alternatives, because corrections concerning saturation and gamut compression can not be handled very well by utilizing CC24 targets.

but robert niessners request also sounds very plausible to me. resolves actual support for color targets is very limited and not always satisfying.

• when this feature was added, i immediately asked for IT8.7 support -- a very common choice for camera calibration --, and it's still not available.
  
• there is also no possibility to disable color patches and exclude them from the calculations or select only a subset of patches from one of the color charts with much more different test points.
  
• beside roberts request for support of different CC24 targets with deviating color specifications, i would also like to see the possibility, to load more precise measured reference values, as supplied for many test charts by their manufactures.

Robert Niessner wrote:The question is, why can't you rely on their saturation levels? Maybe because there is a data mismatch (old vs new)?

The various saturation levels for different color values are inconsistent. If they were specifically designed to hit a Vectorscope target -- as the DSC charts -- then they'd be useful. The DSC people have explained to me that they went through hell coming up with the precise hue/sat values for their charts, which requires a lot of precision and experimentation from their printing company. I would bet they have to toss out dozens of bad charts that don't meet their standards before they ship a chart that's correct.

> Does Resolve use OLD or NEW Color Checker values?
>
It's an 8 year old question, but anno 2025 the dropdown in DaVinci contains
- X-Rite ColorChecker Classsic
- X-Rite ColorChecker Classsic - Legacy

So I assume the 2014 model is still supported...
...while they surpassed 4x it's lifespan / expiration.

I guess that support can be dropped,
and I think it would be nice to see the X-Rite ColorChecker Digital with 150 tiles.

> no-one at X-rite cared to update the support page
>
Neither at Calibrite.
Nor do I spot a definition of these 150 tiles.
But maybe I'm searching in the wrong direction.

unicycle wrote:
So I assume the 2014 model is still supported...
...while they surpassed 4x it's lifespan / expiration.

I guess that support can be dropped,

No, that support can't be dropped, because when you have to work on older footage from back then where that one has been used, you need that values. Older footage doesn't just disappear.

unicycle wrote:and I think it would be nice to see the X-Rite ColorChecker Digital with 150 tiles.

That would be great as I have one but can't use it in Resolve.

unicycle wrote:Neither at Calibrite.
Nor do I spot a definition of these 150 tiles.
But maybe I'm searching in the wrong direction.

You are correct - those data can't be found anymore.

Robert Niessner wrote:No, that support can't be dropped, because when you have to work on older footage from back then where that one has been used, you need that values. Older footage doesn't just disappear.

Ah, yes. I wasn't thinking when I wrote that.
Also, I wrote 150 - it's 140.

Robert Niessner wrote:

unicycle wrote:Neither at Calibrite.
Nor do I spot a definition of these 140 tiles.
But maybe I'm searching in the wrong direction.

You are correct - those data can't be found anymore.

So I searched harder. Tried the archive.org - with success.
So now have a bunch of links to exact data; from unofficial AND official documentation.

But as newbee I cannot drop links, despite that each post here is manually moderated, and should catch spam from valid pointers.
And laufbildkommission.com doesn't disclose any contact details.
(but it still makes me proud to see happy users of TinyURL for your blog in english).

Also the bottom of the ColorChecker page at WikiPedia provide links to the better sources.
At least 3 paths of 2 clicks that shows these chips are surprisingly exact defined, in 16 bit CIE L*a*b* values.
If you convert those values recklessly, then the very slight discrepancies are lost.

Reading all of that -and especially the legal print- makes wonder if the 2014 change was to recreate copyright, as I cannot find a patent. Although I am able to find patents based on the chart, like US7728904B2 - and even that has expired. But not finding a patent sounds like they protect I.P. using copyright instead.

The history of the GretagMacbeth ColorChecker goes back to 1976, and looks like originating in the UK.
49 years is half a century = 2½ * patent expiration. So, I'm wondering wether the legacy values are now public domain / can be reproduces without licence, or still are copyrighted for another two decades
(the old basic values that is - not the new Digital SG version).

Robert Niessner wrote:

unicycle wrote:and I think it would be nice to see the X-Rite ColorChecker Digital with 140 tiles.

That would be great as I have one but can't use it in Resolve.

Which would require licening between Blackmagic and Calibrite...
as Calibrite wants to make money on both selling their product AND of possible use of their product (...).

...unless DaVicini's had a tool allowing -just like LUT Maker Pro- defining your own personal targets. Such that DaVinci users can use the data of ColorChecker Digital SG according their copyright: "allowed for personal use".
Enabling users to create such personal tools also allows the skin-tone specific chart of greywhitebalancecolourcard.co.uk to be used in DaVinci.

And while addressing this tool:
b] use AI to automatically set the initial position the matrix.
c] make it possible to use it on fisheye lensens without undesired corrections for distortion.

Robert Niessner wrote:

unicycle wrote:Calibrite ColorChecker Digital SG with 140 tiles.

I have one but can't use it in Resolve.

Question: do the recent versions still come with a glossy layer?

As a new user you can post links, I think after some time and 5 posts that restriction is gone.
You can send me a private message or an email at
my first name.surname @ mydomain

I don’t recall the targets having a glossy layer.

I would say it is kind of irrelevant, the tool has been broken for years;

viewtopic.php?f=21&t=176392&hilit=+color+checker

Marc posted above recommending not using it and discusses in my post as well.

Sadly I don't think you really have a manual route either? Xrite Details how to manually use Hue-Hue and Hue-Sat curves at 5:00 here;

youtube.com/watch?v=2W2sRPqZRCs&t=303s

But that is NOT classical MacBeth Chart. My understanding is that on the video charts they intentionally picked the 6 color chips to match up with the vectorgram. And these are not the same as the 6 on the Macbeth?

The Red on my color checker passport video is 51, 41, 19 (LAB). Vs the numbers you have above "Patch #15 red : was 175/54/60 (old) and is now 179/0/27 (new)."

unicycle wrote:> Does Resolve use OLD or NEW Color Checker values?
It's an 8 year old question, but anno 2025 the dropdown in DaVinci contains
- X-Rite ColorChecker Classsic
- X-Rite ColorChecker Classsic - Legacy

I wish they would support the Calibrite (new owners of X-Rite) Color Checker Video charts, which are a lot more accurate than the previous charts and were made in the last 5 years. The older Macbeth charts (widely used for film in the 1970s, 1980s, and 1990s) were also bad for digital video, in my experience. The Digital SG chart is also very good, and I know people who test LUTs who use that regularly.

The values are only vaguely correct at best. To me, you either use a calibrated chart, or just look at the scopes yourself and get the vector values in the ballpark and not obsess over it. I think getting the grayscale values (particularly peak white and black) is a lot more critical. Color/tint will change depending on color temp, lighting, and so on.

Here's a visual explanation of how to interpret the DSC charts on a vectorscope:

Color Match does work, but you have to do it in a certain way.

From RAW footage of a ColorChecker Video in sunlight converted to Rec709 without color match:


And with color match:


Off topic: Can you guess the camera?