Luminous Landscape Forum
Raw & Post Processing, Printing => Capture One Q&A => Topic started by: Rado on January 08, 2019, 01:03:32 pm
-
I haven't tested it yet but it seems like the color checker software can now create camera profiles for CO
https://xritephoto.com/CameraICC
-
The question is, exclusive to Capture1?
-
The question is, exclusive to Capture1?
Capture One uses industry-standard ICC profiles.
Any software used to create ICC profiles can be used with C1 and any other program that supports industry-standard ICC profiles.
-
Capture One uses industry-standard ICC profiles.
Any software used to create ICC profiles can be used with C1 and any other program that supports industry-standard ICC profiles.
I'm aware of the first point. And the 2nd. My question still remains unanswered.
I'll ping Xrite. ;)
-
According to Anders Torger and the documentation for Lumariver Profile Designer, C1 has a special way to handle the tone curve, where a part of it is contained in the ICC profile. According to this, a custom profile optimised for C1 is not necessarily a "Universal" profile.
You can read about it here:
C1 profile curves (http://www.lumariver.com/lrpd-manual/#c1_curves)
It would be interesting to know how the profiles from Lumariver Profile designer compara to those from the new X-Rite Tool
-
I haven't tested it yet but it seems like the color checker software can now create camera profiles for CO
https://xritephoto.com/CameraICC
The PDF that comes with the software refers to how to create profiles for capture one and how to prepare the original shot of color checker in Capture One for use in the the tool, it also has instructions on how to use the resulting ICC profile.
My first try is shown in pictures below, D700sun.jpg is with a color profile generated using the tool, D700Sun1.jpg is with the default CO D700 profile. The most noticeable difference is in the dark blue patch, which is very similar to dng profiles created using the software with Lightroom.
-
Futher to my last post here are 2 pictures only differing in that one has color profile generated by the Colorchecker software (the one ending CC2.jpg) and 0ne is the Capture One default (ending CO3.jpg). All other settings including color balance the same.
-
I haven't tested X-Rite's software, just looked at the video how it works, and looked at the train images above to make an A/B comparison.
As usual X-Rite software focus on ease of use which is a good thing for casual users, but in doing so they miss out a little on Capture One curve handling and possibilities to tune your general-purpose profile to taste. It seems like they really mean to use it for reproduction rather than general-purpose photography. For general-purpose I think it's better to have a more elaborate and configurable way to generate the profiles as you will generate profiles seldom and use it many times in many conditions.
Capture One profiles (or in any other raw converter for that matter) should be optimized for the curve you will use it with most, which for general-purpose photography is not the linear curve. That's why you export both linear and film curve when making a profile with Lumariver Profile Designer, one to get colorimetric data, the other to get the curve that will be used so the tone reproduction operator will work on the proper curve and optimize the tone reproduction for that. Another aspect is that Capture One expects some contrast to be stored in the ICC profile too, that's why Lumariver Profile Designer allows import of the bundled profile to extract curve from there, or you can design your own.
As you can see on the comparison shots the contrast of the X-Rite profile is lower as it doesn't take that into account.
When rendering a profile for linear curve and then using it with the film curve it will sort of work, thanks to that Capture One's working space RGB primaries are chosen well to minimize color shifts plus that the film curves lack shadow dip (further minimizes shifts). The shadow dip is supposed to be in the ICC profile itself though. So I wouldn't say it's a high-end way of making general-purpose profiles, and it's not how the Capture One bundled profiles are made either. They're not made for linear reproduction with a curve slapped on, they are made specifically to be used with the film curve, and then instead work a little bit worse with the linear curve, but that is how you want a general-purpose profile to be. For their digital backs Phase One have bundled separate reproduction profiles by the way.
Another aspect to look into, which is not evident from the train images, is how the profile makes gamut compression and roll-off bright colors to clipping. Capture One (or really virtually all raw converters) expects the general-purpose profiles to have some gamut compression built-in, and I'm not sure X-Rite is doing that or not from these images. If you make a reproduction profile you don't want any gamut compression at all of course, so I'm guessing they don't do any.
Otherwise one can see that Capture One bundled renders warmer hues compared to X-Rite. This is because X-Rite is more correct on the hues, Capture One's bundled look is generally quite yellow, which you may or may not like. Personally I prefer more neutral rendering than Capture One's bundled profile, but I tend still to manually tune the residual errors so they go towards warm especially on the reds and blues. Deep blue is difficult by the way, if you keep that reproduction accurate it becomes very dark, so almost all general-purpose profiles lighten it considerably for better although slightly less accurate tone reproduction on screen and in print. Many of those little aspects are taken into account when making a good general-purpose profile. General-purpose profiles are weakly output-referred (ie they take into account that they will be displayed and printed, but are not strictly locked to a specific color space), while reproduction profiles are colorimetric scene referred to be able to copy artwork and similar. These are quite different.
To summarize, it seems to me that X-Rite's profile maker is best suited for casual reproduction workflows, while not ideally suited for making general-purpose profiles.
(Unfortunately Capture One v12 has currently a bug in the transfer function export so profile making doesn't work with that version yet, I guess X-Rite's software is affected too, so one have to use v11, the resulting profiles can still be used in v12 though)
-
Capture One uses industry-standard ICC profiles.
Any software used to create ICC profiles can be used with C1 and any other program that supports industry-standard ICC profiles.
The thing with ICC profiles in raw converters is that while the format itself is standard, raw converter color pipelines and where the ICC profile are applied within them is not. So to make an ICC profile that really works in concert with the ICC-using raw converter you must know how it actually uses the ICC profile. For example a profile maker that is solely designed for making ICC profiles for scanner software will not make ICC profiles that work well with Capture One.
In most raw converters using ICC profiles the ICC profile works on linear input and applies itself the tone curve if you want one. In Capture One the input into the ICC profile is not linear, and the curve is not supposed to be applied by the profile, but separately by Capture One. To further complicate things the ICC profile is still expected to apply a minor amount of contrast, most significantly in the shadow range. Oh, and gamut compression too. A profile maker truly made for Capture One needs to take all these aspects into account. If it only is supposed to make reproduction profiles, like traditional profile makers do, some of these aspects can be skipped, but if you want to make a high-end general-purpose profile all aspects need to be taken into account, just as for Capture One's bundled profiles, which as you know are truly excellent -- if your taste is in line with the rather subjective look they provide.
-
...
(Unfortunately Capture One v12 has currently a bug in the transfer function export so profile making doesn't work with that version yet, I guess X-Rite's software is affected too, so one have to use v11, the resulting profiles can still be used in v12 though)
May you explain what bug that is? I have v12 and v10. Therefore it would be no big deal for me to use an older version to create tifs for lumariver but it is good to know what the bug is.
To the x-rite software: It's good to see x-rite is taking C1 into account. But the software is still a big black box. The tethering shot with the CCPP in the video shows how serious they are about a good profile:
- shadows of the shoes on part of the target
- blue of the very close shoes will affect the very close rows
- the picture itself is underexposed
- the big softbox in the back will put a glare on the target (angle to the camera<45°)
-
May you explain what bug that is? I have v12 and v10. Therefore it would be no big deal for me to use an older version to create tifs for lumariver but it is good to know what the bug is.
Sure: v12 exports random broken curves in the TIFFTAG_TRANSFERFUNCTION tag, and sometimes even crashes on export. This tag is only included if exporting for profiling, and is only used by third-party profile makers. It's an undocumented feature of Capture One that has existed for ages, with the purpose of supporting accurate third-party profiling. So it's sad that it wasn't properly tested before they released v12.
The transfer function is required to know how to linearize the data. In theory you could use hard-coded "Capture One Gamma", but as the linear curve used inside Capture One is not fully linear due to highlight clip handling etc (and differs between cameras) it doesn't exactly match up with the Capture One Gamma, so the only safe way to make accurate profiling is to apply the provided transfer function. And with v12 that function is trash :(. This should not only affect Lumariver Profile Designer but any other software making profiles for Capture One the proper way, that is doesn't use hard-coded approximate curves.
Phase One knows about this bug, but it's certainly not the only bug that needs fixing in v12 so maybe it will be a while before it works again. I have quite a number of Capture One users that uses Lumariver Profile Designer so I've got several "it's not working in v12" emails every week since the v12 release. Unfortunately I can't do anything about it other than referring to using an older version when making the profile (the profile is then compatible with v12), and wait for a fix. If your camera is only supported in v12 the only way is to wait for a fix, or manually attach a sane transfer function but then you can't be sure it is matching your camera exactly.
-
Sure: v12 exports random broken curves in the TIFFTAG_TRANSFERFUNCTION tag, and sometimes even crashes on export. This tag is only included if exporting for profiling, and is only used by third-party profile makers. It's an undocumented feature of Capture One that has existed for ages, with the purpose of supporting accurate third-party profiling. So it's sad that it wasn't properly tested before they released v12.
The transfer function is required to know how to linearize the data. In theory you could use hard-coded "Capture One Gamma", but as the linear curve used inside Capture One is not fully linear due to highlight clip handling etc (and differs between cameras) it doesn't exactly match up with the Capture One Gamma, so the only safe way to make accurate profiling is to apply the provided transfer function. And with v12 that function is trash :(. This should not only affect Lumariver Profile Designer but any other software making profiles for Capture One the proper way, that is doesn't use hard-coded approximate curves.
Phase One knows about this bug, but it's certainly not the only bug that needs fixing in v12 so maybe it will be a while before it works again. I have quite a number of Capture One users that uses Lumariver Profile Designer so I've got several "it's not working in v12" emails every week since the v12 release. Unfortunately I can't do anything about it other than referring to using an older version when making the profile (the profile is then compatible with v12), and wait for a fix. If your camera is only supported in v12 the only way is to wait for a fix, or manually attach a sane transfer function but then you can't be sure it is matching your camera exactly.
Thanks you Anders for the information. It will also affect BasIIColors Input and this is supported by C1 or at least it is recommended to be used in the public color guide DTDCH.
Is it only with the export function? Or does it happen with the "edit with..." function too? And what about "linear scientific" curves? I'll do some tests this weekend.
Can you write us a bug description? I will report the bug too then. If more people ask they may give it more priority then.
-
Thanks you Anders for the information. It will also affect BasIIColors Input and this is supported by C1 or at least it is recommended to be used in the public color guide DTDCH.
Is it only with the export function? Or does it happen with the "edit with..." function too? And what about "linear scientific" curves? I'll do some tests this weekend.
Can you write us a bug description? I will report the bug too then. If more people ask they may give it more priority then.
For critical work like professional reproduction, like most BasICColor users are working with as far as I understand, I would not recommend to upgrade immediately to big new major releases, at least not while keeping the old software and workflow working so there's a fallback. Capture One is far from unique when it comes to having some bugs in the "dot zero" major release, I've been a bit guilty myself in that regard :-[. But hopefully there will be some patch releases soon that starts fixing stuff users have reported in.
The bug description:
"When exporting a TIFF file for profiling the embedded transfer function (TIFFTAG_TRANSFERFUNCTION) contains seemingly random curve segments rather than the actual transfer function. This means that any third-party profile maker that depend on the transfer function to be able to linearize the image data can no longer make profiles. In my case the profile maker is XXX."
Be sure to test first though, I'm not 100% sure that the bug is manifested on all system+camera combinations, and I only want people that actually have a problem to send a report to Phase One.
If you are using Lumariver Profile Designer there's more information here http://www.lumariver.com/lrpd-manual/#c1_troubleshooting for example how you can use the tone curve dialog to actually plot the shape of the embedded transfer function, so you can see for sure that it's broken.
Edit: as shown below not all profile makers need the transfer function (X-Rite can do without). I was assuming that everyone did as the information has always been there. If the profile maker doesn't make use of the transfer function it must in some way estimate it, and it can do this using the target's neutral patches from black to white. I think that process is unnecessary when the information is there and is exact, but if Phase One is slow in fixing this bug I may need to reconsider...
-
Quick update, I just went by the office and briefly tested X-Rite profile maker. It is not dependent on the transfer function, so it can make profiles regardless. In fact I tested to strip the TIFF of both ICC and tags that even said it was generated by Capture One and it still made a profile, so it probably has no hard-coded information either. With no information of what linear is or what software that made the file I guess they make some estimate it from the color checker grayscale step in the image itself, which you of course can do if accuracy is not that important, curve fitting a gamma function would probably yield a quite good result, which they indeed do get. They probably have a large amount of smoothing involved too which make the profiling more robust for any kind of input errors.
The profile maker has no options whatsoever, so you don't really know what it does or what the intent of the profile is. However I made a reproduction profile with Lumariver Profile Designer for reference and compared it to what X-Rite makes, and indeed it does have the look of a reproduction profile as you would expect (this is what traditional profile makers have done in the past, linear reproduction, nothing else). Of course using such a profile with a curve does not yield a particularly good or perceptually accurate result, but how much it shows depends on subject material and I'm sure some will like it anyway. And for casual reproduction work it will make more accurate results than with the bundled Capture One profiles which are designed for general-purpose use and not reproduction, but of course you don't have the diagnostics and controls of more advanced software.
-
The PDF that comes with the software refers to how to create profiles for capture one and how to prepare the original shot of color checker in Capture One for use in the the tool, it also has instructions on how to use the resulting ICC profile.
My first try is shown in pictures below, D700sun.jpg is with a color profile generated using the tool, D700Sun1.jpg is with the default CO D700 profile. The most noticeable difference is in the dark blue patch, which is very similar to dng profiles created using the software with Lightroom.
The dark blue patch is a quite distinctive look of a reproduction profile. That blue patch on the color checker really is very dark so if you are making reproduction work (eg photographing a painting on canvas and then printing it on canvas) it should be that. However if you do general-purpose photography the deep blues will be so dark that tone reproduction suffers significantly, it's all just all to dark to make subtle gradations visible. Therefore almost no bundled profiles from a good quality raw converter renders blues like that, but instead lightens them significantly often paired with a little desaturation. It's just not a good idea to slap a curve on top of a linear reproduction profile and call it a general-purpose profile, and the overly dark deep blues is just one reason.
But well I'm getting on my hobby-horse now as I constantly talk about the importance of differing between linear reproduction and general-purpose when it comes to profiles. If you do want a neutral rendering with contrast, which product photographers want for example, the answer is still not a linear reproduction profile with a curve on top like X-Rite does here, that will over-saturate among other problems. The profile maker must take the final contrast into account to make a percetually neutral rendering.
-
The dark blue patch is a quite distinctive look of a reproduction profile. That blue patch on the color checker really is very dark so if you are making reproduction work (eg photographing a painting on canvas and then printing it on canvas) it should be that. However if you do general-purpose photography the deep blues will be so dark that tone reproduction suffers significantly, it's all just all to dark to make subtle gradations visible. Therefore almost no bundled profiles from a good quality raw converter renders blues like that, but instead lightens them significantly often paired with a little desaturation. It's just not a good idea to slap a curve on top of a linear reproduction profile and call it a general-purpose profile, and the overly dark deep blues is just one reason.
But well I'm getting on my hobby-horse now as I constantly talk about the importance of differing between linear reproduction and general-purpose when it comes to profiles. If you do want a neutral rendering with contrast, which product photographers want for example, the answer is still not a linear reproduction profile with a curve on top like X-Rite does here, that will over-saturate among other problems. The profile maker must take the final contrast into account to make a percetually neutral rendering.
Thanks for that explanation, the darkening of the blue patch is something that has puzzled me for years when using the x-rite software.
Dave
-
For critical work like professional reproduction, like most BasICColor users are working with as far as I understand, I would not recommend to upgrade immediately to big new major releases, at least not while keeping the old software and workflow working so there's a fallback. Capture One is far from unique when it comes to having some bugs in the "dot zero" major release, I've been a bit guilty myself in that regard :-[. But hopefully there will be some patch releases soon that starts fixing stuff users have reported in.
The bug description:
"When exporting a TIFF file for profiling the embedded transfer function (TIFFTAG_TRANSFERFUNCTION) contains seemingly random curve segments rather than the actual transfer function. This means that any third-party profile maker that depend on the transfer function to be able to linearize the image data can no longer make profiles. In my case the profile maker is XXX."
Be sure to test first though, I'm not 100% sure that the bug is manifested on all system+camera combinations, and I only want people that actually have a problem to send a report to Phase One.
If you are using Lumariver Profile Designer there's more information here http://www.lumariver.com/lrpd-manual/#c1_troubleshooting for example how you can use the tone curve dialog to actually plot the shape of the embedded transfer function, so you can see for sure that it's broken.
Edit: as shown below not all profile makers need the transfer function (X-Rite can do without). I was assuming that everyone did as the information has always been there. If the profile maker doesn't make use of the transfer function it must in some way estimate it, and it can do this using the target's neutral patches from black to white. I think that process is unnecessary when the information is there and is exact, but if Phase One is slow in fixing this bug I may need to reconsider...
The issue is present with my generated tiffs too. Beside Lumariver I tested BasICColor Input 5 too and it doesn't work either. Therefore I wrote the support of BasICColor and Phase One. Thanks for your explanations. Let's hope Phase One fix this soon.
-
I had a reply from BasICColor support. They are aware of the issue and are in touch with Phase One which told them: "We have already addressed this bug and there will be a fix in a coming C1 release for this issue." :D
-
In the end of last week I asked Phase One support for progress on the issue and they gave me the same reply that the v12 bug will be fixed in their next patch release which is due out "soon". Interesting to hear about BasICColor. I hadn't tested their profile maker but I assumed they were affected too, which surely has helped in getting priority on the issue as their software is pretty big in professional reproduction so I'm sure they've got their fair share of support emails...
-
I've heard the version 12.0.1 released a few days ago fixed the issue with the transfer function export so BasICColor and Lumariver profilers should work fine again, but I haven't had time to test for myself yet.
Edit: just tested, it works now. :)
-
I haven't tested it yet but it seems like the color checker software can now create camera profiles for CO
https://xritephoto.com/CameraICC
The new ColorChecker for CO looks like it could save me a lot of time in reproducing art photos in the future. When I tried it though (following the PDF instruction) the icc profile that ColorChecker created and saved could not be found by CO. I tried dropping the profile into other various folders associated with color profiles, restarting CO every time i did this. No luck. It seems that CO is very picky about which profiles get listed in the "Other" section of the icc profile section.
I'm on CO 11.1, ColorChecker 1.2 and Mac High Sierra
-
The new ColorChecker for CO looks like it could save me a lot of time in reproducing art photos in the future. When I tried it though (following the PDF instruction) the icc profile that ColorChecker created and saved could not be found by CO. I tried dropping the profile into other various folders associated with color profiles, restarting CO every time i did this. No luck. It seems that CO is very picky about which profiles get listed in the "Other" section of the icc profile section.
I'm on CO 11.1, ColorChecker 1.2 and Mac High Sierra
you are creating a "camera profile" for a specific camera model, are you not ... why in the world you are aiming for it to appear in the "other" section ? instead of making a proper file name for it to show up for your camera model ?
-
Because it says so in the manual. See attachment.
-
Because it says so in the manual. See attachment.
I downloaded the manual, and thank you for the link. I'm puzzled by these statements on page 2:
Capture One Workflow to Apply Profile
1. Re-Start Capture One software to recognize the new
profile.
2. Open the image where the profile will be applied.
NOTE: If images in the collection need to be white
balanced, white balance using the image with the
ColorChecker in it. Then manually use that white
balance setting in the rest of the images.
3. Apply the profile to desired image. The profile will
appear under the Other profiles list option.
Number 2 has me confused! Does it say to white balance using the color pick in CO and then apply the profile? Am I reading this correctly? Then what would be the point of creating a new profile.
One more thing, can I adjust the exposure in the color checker file before I create the TIFF? For example, let's say I over or under exposure the color checker and make the corrections in CO before exporting the image. Will that affect the overall look and accuracy of the ICC?
I'm not entirely sure how to use it after it's done. In the Color panel I select the new ICC in the Base Characteristics ICC Profile drop down box, but do I leave the Curve section to Auto or Linear?
Please help, dazed and confused.
Paul
-
Hi Paul, I was confused at first until I understood that the tool has a particular purpose. It is aimed at photographers photographing say an original painting who want as accurate a colour match as possible. They would set up the lighting etc and then take an initial shot of the colour checker followed by a series of shots of the painting(s) being copied. Then when processing use the colour checker picture as though it was a grey card to set colour balance for the whole series and then the profile to get the best colour match possible.
The tool is not realy intended for generating general purpose profiles.
Dave
-
I downloaded the manual, and thank you for the link. I'm puzzled by these statements on page 2:
Capture One Workflow to Apply Profile
1. Re-Start Capture One software to recognize the new
profile.
2. Open the image where the profile will be applied.
NOTE: If images in the collection need to be white
balanced, white balance using the image with the
ColorChecker in it. Then manually use that white
balance setting in the rest of the images.
3. Apply the profile to desired image. The profile will
appear under the Other profiles list option.
Number 2 has me confused! Does it say to white balance using the color pick in CO and then apply the profile? Am I reading this correctly? Then what would be the point of creating a new profile.
One more thing, can I adjust the exposure in the color checker file before I create the TIFF? For example, let's say I over or under exposure the color checker and make the corrections in CO before exporting the image. Will that affect the overall look and accuracy of the ICC?
I'm not entirely sure how to use it after it's done. In the Color panel I select the new ICC in the Base Characteristics ICC Profile drop down box, but do I leave the Curve section to Auto or Linear?
Please help, dazed and confused.
Paul
ippolitois
so the newly created profile does appear in the "other" section in your version of CO? Because here it is still nowhere to be found.
-
ippolitois
so the newly created profile does appear in the "other" section in your version of CO? Because here it is still nowhere to be found.
Yes, it appears in the Others folder as specified in the PDF. I'm on Win 10, build 1809 with all the updates.
Hope this helps.
Paul
-
You have to name the profile in a specific way to get it to appear in the area you want. It doesn't have to be in the Other folder. See existing profiles for guidance.
-
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.
-
You have to name the profile in a specific way to get it to appear in the area you want. It doesn't have to be in the Other folder. See existing profiles for guidance.
I tried a few things like putting the profiles in different color sync related folders (user and system levels). Also I tried renaming them, as you suggest. But how? For instance a profile the is named CNNIRC09 gets listed in the "Others" folder, whereas a profile named CNN819RA (which rests in the same color sync folder) is not.
As ippolitois mentions that he got it working on a WIN 10 platform, I tend to believe it is an OS related problem (I'm on Mac, High Sierra)
-
But how?
for simplicity - see how P1 names camera profiles for cameras and where they put them ...
-
The dark blue patch is a quite distinctive look of a reproduction profile. That blue patch on the color checker really is very dark so if you are making reproduction work (eg photographing a painting on canvas and then printing it on canvas) it should be that. However if you do general-purpose photography the deep blues will be so dark that tone reproduction suffers significantly, it's all just all to dark to make subtle gradations visible. Therefore almost no bundled profiles from a good quality raw converter renders blues like that, but instead lightens them significantly often paired with a little desaturation. It's just not a good idea to slap a curve on top of a linear reproduction profile and call it a general-purpose profile, and the overly dark deep blues is just one reason.
But well I'm getting on my hobby-horse now as I constantly talk about the importance of differing between linear reproduction and general-purpose when it comes to profiles. If you do want a neutral rendering with contrast, which product photographers want for example, the answer is still not a linear reproduction profile with a curve on top like X-Rite does here, that will over-saturate among other problems. The profile maker must take the final contrast into account to make a percetually neutral rendering.
Thanks, that is useful to know. So in your opinion for the kind of product (and food) studio photography you are talking about is the most convenient way to a good result for a specific lighting condition still to tune the canned C1 profiles (in my case Fuji GFX 50s and legacy P+ backs) using the Color Editor (and calibrated/profiled monitor)?
Tony
-
The thing with ICC profiles in raw converters is that while the format itself is standard, raw converter color pipelines and where the ICC profile are applied within them is not. So to make an ICC profile that really works in concert with the ICC-using raw converter you must know how it actually uses the ICC profile. For example a profile maker that is solely designed for making ICC profiles for scanner software will not make ICC profiles that work well with Capture One.
In most raw converters using ICC profiles the ICC profile works on linear input and applies itself the tone curve if you want one. In Capture One the input into the ICC profile is not linear, and the curve is not supposed to be applied by the profile, but separately by Capture One. To further complicate things the ICC profile is still expected to apply a minor amount of contrast, most significantly in the shadow range. Oh, and gamut compression too. A profile maker truly made for Capture One needs to take all these aspects into account. If it only is supposed to make reproduction profiles, like traditional profile makers do, some of these aspects can be skipped, but if you want to make a high-end general-purpose profile all aspects need to be taken into account, just as for Capture One's bundled profiles, which as you know are truly excellent -- if your taste is in line with the rather subjective look they provide.
Thank-you Anders for the very concise explanation(s) of what it takes to properly produce profiles for C1, very informative.
-
For those really interested in camera profiling, Lumariver's user manual http://www.lumariver.com/lrpd-manual/ (http://www.lumariver.com/lrpd-manual/) is a wealth of information and a must read. Regarding C1 profiles I found it easier to make my own notes after reading through the manual's sections on how that raw converter handles it's profiles and the Tone Curve settings. Next for me is to get explore (and get my head around) the advanced Optimization workflow.
-
Here's a bit of an "Update" on the ICC profiles created with the ColorChecker and the X-Rite software:
They are in fact all broken and unusable for professionals.
What do I mean by that? Well - they are all missing the L*a*b (0,0,0) coordinate and therefore you can't properly convert the images to any other profile like AdobeRGB, ProPhotoRGB or even sRGB without a noticeably shift in the black points.
This shift can be very minor (0 0 0 turns into 3 0 5 rgb) or massive (0 0 0 turns into 13 11 15 rgb). Since this shift happens during profile conversion while exporting a RAW image, it can't be countered by any setting or adjustment in Capture One and of course it also happens in Photoshop. When you try to correct this during or after the conversion in Photoshop it'll impact the contrast and colors of the whole image - which of course renders the whole point of profiling completely moot.
Yes, the X-rite profile has nice colors but due to the crappy contrast and no true black levels it can't be used for repro work (or any serious critical work for that matter).
I notified x-rite about this 3 months ago and it took them weeks to even recognise and confirm that this happens. I got handed through the various support tiers and they even informed product management about this issue in July(!).
So far: no changes, no updates. No press release to inform photographers who use it and may create problems they probably don't know about. Not even a note on their product page.
It seems like they hope it goes away if they just ignore it.
-
Can someone explain how a small number of colour patches that cover a very small part of the 3d colour space reliably create an accurate profile for general photography use. It would seem it's akin to trying to do a 3d survey of a large house using a 12" ruler. Even with the 140 patch SG it's still a small part of the gamut. Is this what PhaseOne use to create their camera profiles?
-
Can someone explain how a small number of colour patches that cover a very small part of the 3d colour space reliably create an accurate profile for general photography use. It would seem it's akin to trying to do a 3d survey of a large house using a 12" ruler. Even with the 140 patch SG it's still a small part of the gamut. Is this what PhaseOne use to create their camera profiles?
All ICC profiles are based on extrapolation. Even if you could measure something close to 16.7 million device values, the resulting profiles would be larger than many images they "Define" by colorimetrics.
-
All ICC profiles are based on extrapolation. Even if you could measure something close to 16.7 million device values, the resulting profiles would be larger than many images they "Define" by colorimetrics.
That's what I assumed and why I never found it worked for anything other than flat copy work. I assume camera makers and RAW processing developers use something more sophisticated that can create control points spread throughout the color space so they are just filling in the gaps and limiting the errors?
-
That's what I assumed and why I never found it worked for anything other than flat copy work. I assume camera makers and RAW processing developers use something more sophisticated that can create control points spread throughout the color space so they are just filling in the gaps and limiting the errors?
I wouldn't make such an assumption and there are different kinds of camera profiles besides ICC profiles which are usually based on a rendered output rather than say .dcp profiles based more on the raw data. So much of this isn't about accuracy and more about producing a starting point for pleasing colors.
-
Anyone care to share their experiences of creating X-Rite profiles on C1 20?
Unless my eyes are deceiving me, something is better.
Contrast is better and I'm getting 0,0,0, on one file and 3,0,0, on another, without having to radically adjust the Curves shadow to the right.
I made two profiles on two shots of the same X-Rite Passport CC.
When I applied the profile that was giving me 0,0,0, to the other card (whose profile was giving me 3,0,0,) I also got 0,0,0, on that card.
After creating a profile, I am zeroing out changes to the CC image file and just choosing Linear and then boosting Exposure until the green in the green patch is about 148/150. What I'm seeing is much better.
Would be interested to hear if others see an improvement using files that weren't great previously.
I'm using version 2.0.1 of the X-Rite software - PC/Windows 10.
Thanks.
D.
-
For critical work like professional reproduction, like most BasICColor users are working with as far as I understand, I would not recommend to upgrade immediately to big new major releases, at least not while keeping the old software and workflow working so there's a fallback. Capture One is far from unique when it comes to having some bugs in the "dot zero" major release, I've been a bit guilty myself in that regard :-[. But hopefully there will be some patch releases soon that starts fixing stuff users have reported in.
The bug description:
"When exporting a TIFF file for profiling the embedded transfer function (TIFFTAG_TRANSFERFUNCTION) contains seemingly random curve segments rather than the actual transfer function. This means that any third-party profile maker that depend on the transfer function to be able to linearize the image data can no longer make profiles. In my case the profile maker is XXX."
Be sure to test first though, I'm not 100% sure that the bug is manifested on all system+camera combinations, and I only want people that actually have a problem to send a report to Phase One.
If you are using Lumariver Profile Designer there's more information here http://www.lumariver.com/lrpd-manual/#c1_troubleshooting for example how you can use the tone curve dialog to actually plot the shape of the embedded transfer function, so you can see for sure that it's broken.
Edit: as shown below not all profile makers need the transfer function (X-Rite can do without). I was assuming that everyone did as the information has always been there. If the profile maker doesn't make use of the transfer function it must in some way estimate it, and it can do this using the target's neutral patches from black to white. I think that process is unnecessary when the information is there and is exact, but if Phase One is slow in fixing this bug I may need to reconsider...
It seems to be broken again with C1 20. Does others have this issue too?
-
It seems to be broken again with C1 20. Does others have this issue too?
No, I just tested it, and it works without issues. Tested both with a standard and a repro profile
-
Thanks, therefore I will do a complete new install as the support suggested.
-
No, I just tested it, and it works without issues. Tested both with a standard and a repro profile
Would you (or someone else) please do me a favour and add a layer to the raw file before export?
The layer does not need any adjustments. Just paint a mask on a new layer and export to tiff.
It seems to me that the issue (in lumariver) only occurs when there was a visible layer while exporting.
-
Would you (or someone else) please do me a favour and add a layer to the raw file before export?
The layer does not need any adjustments. Just paint a mask on a new layer and export to tiff.
It seems to me that the issue (in lumariver) only occurs when there was a visible layer while exporting.
I haven't had the chance and may try this later. In any case, why would you want to use a file with a layer for creating a profile?
As far as I know, the idea is to have a completely unmodified raw, except for white balance
-
I've been testing Lumariver for repro profiles made from C1 20 exports.
No edits to the file except white balance/No Color Correction/Linear, and Embed Camera Profile on export. That's it.
Working well for me.
D.
-
I haven't had the chance and may try this later. In any case, why would you want to use a file with a layer for creating a profile?
As far as I know, the idea is to have a completely unmodified raw, except for white balance
Well creating a tiff is never ever the same as "having a completely unmodified raw". ;)
Just kidding.
I'm doing some more steps after importing the shot of the target:
base: "no color correction" profile, "linear scientific" curve, no shaperning, no noise reduction, no chromatic abberration, default adjustments for the rest
1. add a LCC to correct light falloff and color cast.
2. Shoot another picture with a white plane in front of the target. With this shot I create a layer to equalize the brightness for the whole area of the target (set the dark and white point on the levels tool as close as possible --> adjust with the brush on the layer). Of course I did everything before to get an "as perfect as possible" lumination of the target and the layer has only a slight brightness change (<0.1) with a very subtle mask. This layer is copied to the target-shot
3. adjust the brightness of the target shot to get the right lumination at 65% (a grey card with sRGB:165). This adjustment is also below 0.1. Otherwise I reshoot it.
4. Crop the image to the frame of the target.
Imo this give me better results than using the flatfield correction (for 1 and 2) in Lumariver and customized tone curves have the same brightness base (thanks to 3). As all this is possible with C1 and thanks to tethered shooting it's also not a big thing to get a good shot/check quickly, why not using these adjustments?
-
Could you tell me your LAB or RGB values when you shoot the card and when you export. Usually i shoot and set to Auto and i get White 240 240 240 and middle gray 130 130 130. This is when my curve it's on Auto when i set to Linear Response i get -2 EV at least. And i have to increase exposure or compensate with pushing up +2EV in C1.
Could you give me an advice?
Thank You
-
After import I'll check the white patches of the CCSG-shot. E5 should have a mean sRGB value of 242-243 (with a linear scientific curve). When I export the file, I'll check lab L-value of 65 for G5 (or similar patches on other targets).
Your correct with the +2EV. But of course you have to increase the EV while doing the shot and not later in C1.
-
Then you adjust your exposure in Linear Capture, not in AUTO. Am i right?
After import I'll check the white patches of the CCSG-shot. E5 should have a mean sRGB value of 242-243 (with a linear scientific curve). When I export the file, I'll check lab L-value of 65 for G5 (or similar patches on other targets).
Your correct with the +2EV. But of course you have to increase the EV while doing the shot and not later in C1.
-
Then you adjust your exposure in Linear Capture, not in AUTO. Am i right?
Yes, that's correct. My curve in lumariver does add contrast with an s-curve form. It is quite similar to the auto-curve.
Sorry for the late reply. I'm not very often in the forum.