Issue with ColorChecker Profiles
Has anyone else trouble getting those profiles created with a ColorChecker Passport (Photo2) to work correctly with Capture One?
I'm not able to get true blacks with these Profiles - the rest of the color seems to be in order, but the images look slightly less contrasty and no matter what I do, I can't get a true black (0 0 0 0) value with them.
With the regular camera profile it's no issue at all...
Happens on both Windows and OSX with current versions of Capture One and ColorChecker Camera Calibration...
Camera: Fujifilm GFX 50s
I'm not able to get true blacks with these Profiles - the rest of the color seems to be in order, but the images look slightly less contrasty and no matter what I do, I can't get a true black (0 0 0 0) value with them.
With the regular camera profile it's no issue at all...
Happens on both Windows and OSX with current versions of Capture One and ColorChecker Camera Calibration...
Camera: Fujifilm GFX 50s
0
-
I think I have found the problem:
Capture One is not using "Black Point Compensation" during ICC Profile conversion.
So any custom camera profile (such as the ones creates with the X-Rite Software) is not correctly translated into other ICC profiles such as Adobe RGB or ProPhoto RGB (etc) which results in a shift of black values, which prevents true blacks from being rendered/displayed!
I contacted support and I suggest you check your own profile conversion if you are using a custom profile!!0 -
According to the Lumariver Profile Designer software, "Capture One has a quite complex way to handle tone curves, so there are a few more steps to follow..." I've tested both and Lumariver seems to have resolved this issue IMHO. 0 -
Lumariver may have a correct black level but the colors are quite a bit off. A reddit user created a Lumariver profile with the same RAW and the colors were muted and badly seperated.
With Xrite the colors are on point but the black levels are bad - so if Xrite and/or PhaseOne can get it together and work things out it would be the best solution.0 -
C-M-B wrote:
Lumariver may have a correct black level but the colors are quite a bit off. A reddit user created a Lumariver profile with the same RAW and the colors were muted and badly seperated.
it all depends on how you build profile and that starts with how you shoot your target ... free dcamprof or commercial LrPD (both from Lumariver) are just a tool with many possible inputs and options (so "A reddit user created a Lumariver profile with the same RAW and the colors were muted and badly seperated." - says nothing at all about the tool, but a lot about the user) and unlike X-Rite software they allow (for a qualified user) to do so much more ...0 -
True but I was just making an observation.
The reddit user also mentioned that it was just a quick and straightforward profile - and I have to say that most users probably won't need anything but that (especially for free).
As I said; if Xrite/Phase One can work it out, it'd be the best solution for me0 -
I created a profile with Xrite and the colors are great (accurate). Black levels...bad.
In my quest for greater overall image fidelity, I sprung for Lumariver Profile Designer this weekend ($113). The blacks are much improved but the colors are bleh...skin tones in particular. I followed Martin Bailey's instructions here: https://vimeo.com/268585170 as well as very methodically walking through the Lumariver instructions. After creating 6 or 7 profiles, I couldn't get colors that looked as good as my Xrite profile.
So I wasted the better part of an afternoon (and probably $113). I suppose operator error could be in play here, but the instructions are not that complicated. I may take another shot when I get time but as of now, disappointment lurks.1 -
Richard, please get in contact with X-rite. I have contacted them as well and they seem somewhat unfamiliar with this issue.
The more users report this, the sooner a fix is going to happen!
If I use "embed camera profile" as the Output ICC Profile in the "Process" settings, the colors and black levels are correct! Then I can use Photoshop to convert it to another Profile with 0 issue.
But if I use any other output ICC Profile (like AdobeRGB or ProPhoto or sRGB) I am unable to get a true black value of "0 0 0 0" anywhere in the image - even if I totally underexpose or clip the blacks with levels and curves - it does not get even remotely close to a "0 RGB" value...
Of course it's not an option to use the x-rite Profile for further editing or printing or sending images to clients, they need something commonly used and widely spread like AdobeRGB or (sometimes just) sRGB. Maybe ProPhoto.0 -
Windows 10 x64
I went to this link...
www.xritephoto.com/cameraicc
...and filled in the form.
I wasn't then given a link to download the beta software.
What am I missing? I'd like to try it out.
Thanks.
D.0 -
Try: https://www.xritephoto.com/ph_product_o ... reID=2059#
and click on "PC Version".
Then you need the Target and you can take a photo an create a profile using this target.
However as I said these profiles are flawed or just broken - there's a definite shift happening with the black values when processing the files to a different ICC profile like Adobe RGB. Something that happens with no other profile, neither the generic profile nor one created via lumariver.
I've talked to support lvl1 and lvl2 and they escalated the whole issue to "Product Management" because they don't know what's wrong. They get the same results.
So these profiles are NOT for professional use.
I highly recommend using the generic profiles instaead until they've figured this out - which is really really annoying since I paid good money for the target which is now useless.0 -
I followed the instructions in this video (better than X-Rite instructions).
https://youtu.be/n4kO6HndQpc
I presume that the rest of you are doing similar?
Thanks.
D.0 -
Are you not experiencing the issue with the black level shift?
Are you able to achieve a 0 0 0 RGB value with these profiles?0 -
In what way are you’re trying (& failing) to achieve a 0,0,0, RGB value?
I’m asking because I would not expect any 0,0,0, reading from the colour checker.
So, I’m assuming you’re making adjustments (to Exposure?) in an effort to make the black patch 0,0,0, and not achieving this?
But, I suspect I’m misunderstanding what you’re at.
If you could walk me through your workflow, I’d be grateful.
Thanks.
D.0 -
I'm trying to achieve 0 0 0 ANYWHERE in the image.
So for example if I set the following parameters to an image:
Exposure to -4
Contrast: 50
Shadow/Black Level: 255
That would have to result in 0 0 0 0 - a "true" black level, no matter what.
But when I export the image to a common RGB profile like Adobe RGB or ProPhoto RGB that value shifts. With one xRite profile it's 18 12 14 14 (Red Green Blue Brightness). With a different xRite profile it's 13 0 0 4. And yet another gets me to 9 0 0 3.
So no only does it shift the "true" black levels to unuseable levels which can not be corrected properly, the shift also varies between each profile and also the shift between AdobeRGB and ProPhoto and sRGB is differnt!
BUT! When you put the output ICC to "embed camera profile" (= no conversion from the x-rite profile to another profile) the blacks stay perfectly fine.
However nobody would think of using a custom profile to send to clients or to be printed.
Also when you use generic profiles or profiles generated with Lumariver the blacks don't shift.0 -
Windows 10 x64
Finally back at a computer, having been away, so a few observations...
1. The colours I get in C1 on the Colour Checker, after making a profile, are bang on.
2. But, as has been pointed out, it is impossible to get a true black in C1. e.g. in Levels, drag the shadow slider as far to the right as you like. You will not get 0, 0, 0, nor will your black/shadow warning blue kick in.
3. After profiling, the black patch on the X-Rite CC (within C1) reads 53, 53, 53, (approximately, +/- 1 point) which is what I would expect. (I use the green value in the green patch as a gamma for the purposes of preparing the profile file for export from C1. So, after choosing Linear Response/No Colour Correction, I increase Exposure to get about 150 for the green value of the green patch, before exporting the file)
4. If I now export this profiled image of the CC, in Adobe RGB, and open it in Photoshop and create a Levels adjustment layer, dragging the shadow slider 53 points to the right renders the black patch 0, 0, 0, which is what I would expect.
So, the inability to create a 0, 0, 0, value in C1 would appear to be a C1 issue alone.
While this may be a problem for those seeking to create punchy profile recipes for general use (though your contrast issues can be addressed afterwards in Photoshop or similar) I think that the workflow is usable for repro work.
Any further thoughts appreciated.
D.0 -
This is unuable for repro work!
First of all the profiles are not coherent in their black level shift:
Profile A created with a tungsten light source (approx. 3090K) and an old camera lens results in a black level shift to 11 0 14 5 (Red Green Blue Brightness) when exported to AdobeRGB.
Profile B created with a flash light source (approx. 5800K) and a modern camera lens results in a shift to 13 8 12 10 when exported to AdobeRGB.
I also tested this with daylight at 12:00 and the result was pretty much the same.
And if you use ProPhoto it shifts to a completely different value altogether.
So it's impossible to predict the behaviour and if a shift to 13 8 12 10 occurs it would mean a lot (!) of correction which would ruin the whole concept of profiling. You'd be better off letting a bunch of blind monkeys take the image and then hammer on your keyboard in photoshop than using the x-rite profile and later correcting it.0 -
C-M-B wrote:
This is unuable for repro work!
First of all the profiles are not coherent in their black level shift:
Profile A created with a tungsten light source (approx. 3090K) and an old camera lens results in a black level shift to 11 0 14 5 (Red Green Blue Brightness) when exported to AdobeRGB.
Profile B created with a flash light source (approx. 5800K) and a modern camera lens results in a shift to 13 8 12 10 when exported to AdobeRGB.
I also tested this with daylight at 12:00 and the result was pretty much the same.
And if you use ProPhoto it shifts to a completely different value altogether.
So it's impossible to predict the behaviour and if a shift to 13 8 12 10 occurs it would mean a lot (!) of correction which would ruin the whole concept of profiling. You'd be better off letting a bunch of blind monkeys take the image and then hammer on your keyboard in photoshop than using the x-rite profile and later correcting it.
So, why am I getting 53,53,53 for the black patch in C1 and the exact same value in Photoshop when I export it as an ARGB Tiff file?
D.0 -
Because you're measuring the "black patch" which does not represent a true black tone of 0 0 0 RGB. That one shifts to a certain value (dependent on your profile).
Try it yourself and set the Exposure to -4, contrast to +50 and clip the black levels at 255. That will give you an image that is 100% black with a value of 0 0 0 RGB.
A friend of mine is currently examining these profiles and it seems like x-rite developers made some errors - they only set the L(0) (lightness) value as black and not L*a*b(0,0,0). That would explain everything.
Only problem is that I'm getting close to 0 feedback from X-rite. They bumped the issue to product development and they are not reacting at all.0 -
Thanks.
But, my point is that while I can’t force an image to appear as a pure 0,0,0, as you rightly state (because if I drag the Levels shadow slider all the way to the right, the image goes completely black, not completely blue coz the shadow warning doesn’t activate. Also, that fully black image doesn’t read 0,0,0, but more like 0,5,12,) since I am only making small changes, does this apparent bug have practical implications? In terms of RGB readings of the patches and the visual appearance of the image on my calibrated monitor, it doesn’t appear to be detrimental.
This bug only manifests in the blacks. If I blow the highlights, I get a fully red (highlights warning) screen and a 255,255,255, readout, which is what I’d expect.
But, within the range of 20,20,20, to 240,240,240, within which I’m working, everything seems spot on.
In short; image looks perfect and RGB patch readings are also perfect when the x-rite profile is applied. The only thing I can’t do is force a 0,0,0, image/reading, something I’ll never need.
D.0 -
Of course it matters!
When I'm editing images and sending them to clients or to a lab for printing them I will always have to edit the images afterwards to get to a true black tone. And because different profiles (for different lenses) have different black levels (some have 4 0 0, other have something like 9 7 7!) there is no efficient and quick way to edit them to look the same. Which is the whole point of profiling each lens and camera.
1. This causes additional time and effort in order to correct the mistake X-rite made with their profile software.
2. This always has the potential for mistakes, oversight and error when trying to correct this mistake.
3. This causes shifts in brightness/contrast and probably also color when you try to correct those images. So the whole point of creating a profile is completely moot.
I think your case (20 20 20 - 240 240 240) is much too special for this topic. Most photographers work within the full range of available RGB values when they're using regular common RGB profiles, otherwise the images would certainly look very dull, especially when printed on glossy paper or when viewing them on a screen.
ps.: One of my profiles results in a black level of rgb(18, 12, 15) - this profile has been created exactly as required by x-rite and using full spectrum daylight (at about 12:00) as a light source. Another one, created using the same parameters but using a different lens results in rgb (10 0 0 1).
It is impossible to use these profiles and have the images look even remotely similar, especially with darker images. Imagine a photo of a landscape at dusk/dawn without the blacks starting at RGB 18, 12, 15. And the people with X-rite are still debating whether this is a bug/mistake or a desired result. Good god. 😂0 -
Working perfectly for me now.
Not sure if it's the Windows 1903 update, which I installed yesterday, or the C1 12.1.1 update, which I also installed yesterday.
Anyway, I'm now getting a solid blue (shadow/black warning) on the entire image when I drag the Levels shadow slider all the way to the right. Previously, the image had been turning black not blue (Highlight/Shadow warning was switched on) and the blacks didn't look right. Now they're perfect.
D.
Ps. The Levels/Shadow Slider drag gives me a solid blue on any image. Moving Exposure to minus 4 does not except on already dark images.0 -
Did you set the output ICC to "embed camera profile"? Because then yeah, you'll get black.
But with any other RGB Profile like AdoberRGB you won't get it.
I have tried it a few days ago with Windows 1903 with C1 12.1.1. and it didn't change anything. Same thing with current Mac OSX.
Also that couldn't change due to a Windows/OSX or Capture One update, since the x-rite profiles themselves are faulty. So they would have to change their program and new profiles would have to be created.
You can't get the missing L*a*b (0,0,0) coordinates from a profile by updating Windows or Capture One. Those were never the issue.0 -
I'm now a bit confused as to what you're trying to achieve.
If I open an image of an X-RIte CC in C1, with ICC Profile saying Generic (in my case, Canon 5D MkIV), and with View/Proof Profile set to Adobe RGB 1998, and with the Curve set to Linear Response; I get RGB values that are quite a bit off - and that's with Exposure bumped up to bring the grey patches into line.
However, if I create a profile using X-Rite (with Embed Camera Profile) and then apply that profile to the same target image (also with Curve/Linear Response, and with Exposure boosted to raise the grey values) I get X-RIte CC colour patch values in the image that are almost bang on in Adobe RGB. (When they are off for a given patch, they are off relatively or equally across R, G and B.)
If I switch View/Proof Profile from Adobe RGB to, say, ProPhoto, the readings are similarly in the ball park.
The real tell-tale is the Cyan patch, which tends to visibly change the most on-screen as I switch between viewing profiles. So, for example, in ProPhoto RGB, I get a reading of 75, 103, 148. If I switch to Apple RBG, the Cyan patch colour changes noticably, and the RGB reading is 0, 110, 156. In both cases, the RGB values are acceptably close to what they should be.
You seem to want to be able to input, for example, Adobe RGB into the X-RIte software output and get a perfect ARGB profile in return. I'm saying that what I'm seeing in colour spaces that I would use (ARGB and ProPhoto) is pretty much exactly that, and that what you're trying to do is not necessary. If fact it's wrong, and that's why the X-Rite software instructions say Embed Camera Profile and not any other profile. The software creates a camera profile for a given lighting/camera/lens combination that gives RGB readings for any colour space I've tried that are very accurate.
The proof (quite literally) of the pudding is in the RGB readouts of the different profiles from the same image.
D.0 -
No.
I'm taking a photo.
In the Base Characteristics I set the X-rite profile as the ICC Profile and set the curve to "Film Standard" or "Film High Contrast". Both Film Standard" and "Film High Contrast" use the X-rite Profile - camera specific curves (for example the Fuji curves) ignore the x-rite Profile and use the generic one.
If I set the OUTPUT ICC profile (in the Process Recipe) to "embed camera profile" it looks fine.
If I set the OUTPUT ICC profile (in the Process Recipe) to any common RGB profile (AdobeRGB, ProPhoto, sRGB,...) the black levels change and get elevated and I can't get to a true black tone. No matter what. It always stays well above 0 0 0. And that is not correct behavior when it comes to profile and color conversion.
Even if I keep the "embed camera profile" and then convert it in Photoshop to something usable like AdobeRGB, the shift still happens. That's because the profiles are broken and they are missing a noticeable bit of the L*a*b (0,0,0) coordinates. So when a conversion happens there's no way to achieve a true black tone since the program that's doing the conversion can only use a L*a*b point form within the original profile.
A friend of mine (who is a bit more tech- and color-savy than I am) suspects that during profile generation the x-rite software only looks for a value close to L(0) without any regard for the a* and b* axis. And at that very point the software sets the lowest value which of course would be incorrect an a huge mistake.
Now whether this is due to the flat linear curve for the target TIFF from the calibration image or due to something else - I have no idea. I only know that the black values get screwed up.
The colors themselves are indeed fine! They look great, better than the generic profile. But it simply lacks contrast and that's the reason why. This has already been noted by a lot of people (especially compared to profiles generated with other software solutions) and I think that's why the x-rite profiles have not yet caught on. People see the result and are disappointed in the lack on contrast and the loss of black and they just stop using it or return it. You can find dozens of comments about this but It seems like they didn't talk to x-rite about this.
The middle and upper value are on point because the middle and upper values are available as L*a*b coordinates in the profile. They are correct. But the black and (as a consequence) similar dark values are wrong.0 -
I think you need to talk to your tech friends.
The X-Rite software takes RGB numbers that read incorrectly off a CC Color Checker image in any colour space you view it in (using C1’s generic camera profile) and produces a profile that gives near perfect RGB numbers from the CC patches in any common colour space that you view them in.
There is no issue with contrast and/or a solid black unless you (incorrectly) fail to Embed Camera Profile in the output.
Please post links to comments by those who think otherwise.
Thanks.
D.
Ps. And when you say the black value is wrong, which black value are you referring to? The black patch on the CC? Whichever black you’re referring to, what RGB value should it be?0 -
Dinarius wrote:
I think you need to talk to your tech friends.
The X-Rite software takes RGB numbers that read incorrectly off a CC Color Checker image in any colour space you view it in (using C1’s generic camera profile) and produces a profile that gives near perfect RGB numbers from the CC patches in any common colour space that you view them in.
There is no issue with contrast and/or a solid black unless you (incorrectly) fail to Embed Camera Profile in the output.
Please post links to comments by those who think otherwise.
Thanks.
D.
I have already taked to X-rite and they CONFIRMED this behavior.
Again: this issue is ONLY with black and darker values of the image. NOT with mid- and high-tones.
A few comments regarding the contrast and X-rite ICC profiles:
https://forum.luminous-landscape.com/in ... c=128465.0As you can see on the comparison shots the contrast of the X-Rite profile is lower as it doesn't take that into account. (torger)
I have created three or four camera profiles with the ColorChecker Passport for Capture 1. While I think the profiles for my cameras used in Lightroom are excellent and make a HUGE difference, the profiles, from C1, in a word, suck. I keep thinking it's me and I keep remaking the profiles thinking I did something wrong, but each time the color is flat, unsaturated and just plain bad. I had high hopes, but at this point, I'll use the canned generic profiles bundled with C1. (nemophoto)
https://fstoppers.com/capture-one/x-rit ... ent-447784Has anyone else had problems with this? I have tried it with a 5D Mk III and a 7D Mk II. After creating the profiles and applying them, the skin tones and other color look great but I completely lose the blacks. See examples below. Nothing I can do will bring the blacks back. Any thoughts? Scott Bolster - January 11, 2019
https://www.youtube.com/watch?v=3DsrO6u ... RFt4AaABAg(...)
- contrast is too flat, no chance to add or modify a tone curve (dcamprof/Lumariver can do that)
(...) (engibda)0 -
Dinarius wrote:
Ps. And when you say the black value is wrong, which black value are you referring to? The black patch on the CC? Whichever black you’re referring to, what RGB value should it be?
I'm talking about any point in an image that should be "0 0 0" RGB. Anywhere on any image.
So if you take a photo during night with no moon and the lenscap on with ISO 50 and 1/1000s at F22, then you underexpose in Capture One, set contrast to +50 and set the shadows in the levels-tool to 255 - it should be completely black with an RGB value of 0 0 0.
Using the generic profile as camera profile and AdobeRGB as an output profile that results in 0 0 0 RGB.
Using the xrite profile as camera profile and AdobeRGB as an output profile that does not result in 0 0 0 RGB. Instead you get values ranging from 4 0 0 to 18 11 15.
Do you understand what the problem is?0 -
Read all the comments on the X-Rite YouTube video link.
Overriding impression is that people have not a clue what having to use Curve/Linear (when creating the profile) implies for the finished profile. The comment from nemophoto above is a perfect case in point: “...the colour is flat and unsaturated...†Eh...precisely. I get the impression that people are expecting Instagram-ready filters and not profiles.
The other criticisms do not detail how they are creating the profile - incorrectly, I suspect.
D.0 -
Do you understand the problem that it is impossible to get to a true black tone (0,0,0) RGB because the profile does not cover L*a*b(0,0,0)?
I am not asking whether it's an issue for your personally,only if you understand that it is a problem for others who take images with (0,0,0) RGB blacks.0 -
hi C-M-B:-) and everybody else.
its not a big science to show what is C-M-B is talking about and apparently x-rite admited already the mistake in calculating a colorspace.
you just need a visualisation software for colorspaces, like colorthink or gamutvision, the last one is freeware. the homepage of gamutvision seems to be offline, but maybe you are lucky and able to find it somewhere. i used gamutvision, which is less advcanced as colothink but enough to find the answer...
if you load a x-rite colorprofile and zoom into L*a*b(0,0,0) you will see that the body of that colorspace doenst reach that point. on the other hand every other colorspace created with x-rite is somehow trying to reach into coordinates around (0,123,123) which is also out of visible spectrum, though you are profiling from a chart for a visible spectrum... i assume x-rite did think of correcting the blacks in calculation after creating a profile but on the wrong spot. anyway, the result of this colorspace is - if you shoot something black like C-M-B did describe, your data will never contain information with coordinates L*a*b(0,0,0) but something like (5,0,0) using this profile.
so, the real problem accures whe you are changing your colorspace into something like adobebrg or srgb just for web - the problem becomes obvious - your L*a*b coordinates will become in rgb something like (11,13,12).
every paper colorprofile is missing blacks even more than the ones you create with x-rite and they are only for softproofig but not as a starting point for working on your pictures.0 -
I previously owned Pentax and the colors were off. I now have a Sony A7iii I do not have any color issues with the lenses (Samyang, Rokinon, Pentax) or any color cast due to lighting that was not correctable with white balance. But it is good to know that I can correct color using Capture One with my Sony camera and the Color Checker 0
Please sign in to leave a comment.
Comments
100 comments