New color chart for high quality profiling in captures

[MHMG]

Mark, for the sake of other forums members like myself who are curious and eager to learn, please share your insights!

Sam, no, please don't go there with me! I truly believe there are limits to what we can accomplish in a forum discussion, and I feel expending further intellectual and emotional capital on a new product I've never seen nor used is not fair to me nor to the OP who is clearly proud of his new product.

I will leave everyone here with one parting comment, however, and it comes from my experiences in one of my former lives, namely as a co-owner of Old Town Editions, Inc back in the 1990s where we did many many fine art repros of original artwork on an Iris 3047 printer (notice how I avoided using the term "giclee" ). The challenge wasn't mapping in-gamut source color to in-gamut destination color. The challenge was to map out of gamut colors (ones that failed to reproduce realistically in the camera capture and even less so in the print output, for example, extremely vivid fluorescent dyes as are used commonly in the textile industry, and so forth, in the original art). When faced with that challenge, it came down to one's photoshop and printmaking skills, and working closely with the client, to resolve the problems in a subjectively pleasing, albeit, colorimetrically inadequate way. I got lot's of in-the-trenches experience that's hard to describe objectively. So, for example, in the sample Hugo posted earlier in this thread of Capture 1 versus ACR initial "default" camera output, I can work with either one of those images with equal success. The ACR rendition, IMHO, is closer to reality colorimetrically, whereas, the Capture 1 version is more "pleasing" to folks loving more "pop" in their images.  Yet either one is an acceptable starting point to a personally envisioned end result, IMHO, so I don't get hung up at the camera capture step. I spend my time on the workflow from edited source original to final output printed image.

That's just me. If you can find a camera capture technique that nails it right in the first step of the source file to final output workflow or even gets us just a little closer to it more often than not, then go for it.

[DP]

ICC profiles take place on rendered data that might be either, NOT the raw data.

in real life they do operate how software developer wants, not how ICC (organization) says they should be used... ICC in that case is just a container... that was noted here on this same forum many times.

[DP]

It's 'finger-printing' as best it can, the rendered (output referred) image.

you don't need a rendered image to create camera profiles in icc/icm containers for raw converters like C1... they can be created directly from raw data (raw files) using tools like rawdigger and dcamprof w/o using C1 itself at all ... this was again noted many many times in this forum

[DP]

whereas, the Capture 1 version is more "pleasing" to folks loving more "pop" in their images. 

just because manufacturer decided so by tuning camera profiles to their vision of how rendering shall be - it is not an genetic feature of C1 somehow...

[DP]

The SG had a wider gamut using glossy patches but it was a bitch and a half to shoot (no reflections please). Then they came up with a newer target without the glossy patches but smaller color gamut.

you mean DC chart (the one that was with "glossy" patches --> http://www.chromaxion.com/information/colorchecker_dc.html ) actually and then they come up with with SG ... getting old, Andrew, forgetting things 

[digitaldog]

you don't need a rendered image to create camera profiles in icc/icm containers for raw converters like C1... they can be created directly from raw data (raw files) using tools like rawdigger and dcamprof w/o using C1 itself at all ... this was again noted many many times in this forum

Yes, you're correct in the case of the two products you speak of. Now how raw that data is, I can't say. The majority of other products no.

[digitaldog]

you mean DC chart (the one that was with "glossy" patches --> http://www.chromaxion.com/information/colorchecker_dc.html ) actually and then they come up with with SG ... getting old, Andrew, forgetting things 

I sit corrected     but indeed, there are two such targets, one with glossy patches and a wider gamut, difficult to capture. There's one without, easier to capture. I have both of course and at the time, spent a lot of time working directly with GMB (still have some hand made targets). Haven't used them in like 10 years: no reason to do so.
Getting old, yes. Beats the alternative. 

[Doug Gray]

I've used both the SG CC and the regular CC. The SG has significantly larger gamut but one has to go to lengths to avoid reflections. But it can be done with care and understanding in a studio. Not trivial though.

Of course with the SG one won't get linear mapping because the spectral diversity doesn't nicely map into a camera's RGB response . With a printed chart with a large number of patches, one can get quite low dEs. Under 1.0 ave is fairly easy. But it's because the inks don't have much spectral diversity. So it's a fairly simple job to get pretty much any camera's RGB RAW capture to map colorimetrically. But it's an illusion. Capturing real life images will see metameric error.  OTOH, capturing images of printed fine art can be quite accurate as the CMY spectral characteristics tend to be similar form printer to printer.

So it all depends on what you are trying to replicate. Printed materials tend to be more accurately rendered with profiles made from printed charts. OTOH, oils, will not be captured as accurately.

The OP's approach should work quite well capturing printed materials. But how well it captures real life or painted materials is sort of a crap shoot. OTOH, there is no way to do that really well without multiple exposures using color filters that are then processed. There are academic works that delve into this. Especially from RIT.

[Czornyj]

OK, let me know what you already predicted and concluded just by looking at the samples (I guess you also got the reference spectral file and did some calculations and research...).

1) Why CC24 colors are distorted on "calibrated" sample image?


2) Why target producers like X-Rite ettc. are so stupid that they don't print targets like you when it's 100x cheaper?

3) If your idea is so clever, why not just use a target displayed on a reference color critical wide gamut monitor without the whole printing hassle?

4) All joking aside - can you provide spectral measurements of your target?

[hurodal]

Hi andrew,

Couple points:
1. This comment about linear or non linear is moot. ICC profiles take place on rendered data that might be either, NOT the raw data.

I'm sorry but I don't get the point. Of course ICC profiles take place after RAW data has been put into a standard file, and those data can be either.
What I said is that I'm not that sure that camera response is absolutely linear when it comes to color render.

2. There is no such thing as ‘no setting’ in ACR/LR. You might not like the defaults (dirt rendering?) but they can easily be changed and since these products CANNOT use ICC camera profiles, for lots of good reasons, that comment is moot too.

What I meant with no setting is that there was no change in the default settings. It might be something related to the language, as in my native language (castilian spanish) it's usual to say that.
So 'no setting' = 'default, unchanged setting'.
Of course if I don't like the defaults I can change that, but what conclusion do you get when comparing two products and one offers a much better result with default setting? 'you can easily change the settings'?
I'd get another conclusion, but it's up to everyone...

Yes, ACR cannot use ICC camera profiles, but I really doubt there are lots of good reasons. Just look at the facts: Capture one gets WAY better color using the same 'old' icc profiles than ACR with the new DCP profiles.
In my country you'll never see any product, fashion, ecommerce pro photographer developing with other than C1. Why?

3. Your site shows a before image with a profile. Is this a custom profile you built from a ‘standard’ available target and if so, which target using what software? IF it’s C1’s canned profile, your examples are rather worthless sorry. You can’t compare a canned profile (built using any target) to your custom profile with your target. So hopefully you’ll tell us that the before is a custom profile in C1 using a differing (and inferior) target.

It's very easy Andrew: it's the generic embedded profile into C1 (or ACR in that case). It's the profile that C1 automatically uses when you open your raw files. Canned profile? I've never heard that, don't know what you mean.
How do Phase one's engineers built that profile is something I don't know, but what do I know is that they aim to 'pleasant' colors, not faithful.

I hope I clarified the questions.

Best regards,

[hurodal]

1) Why CC24 colors are distorted on "calibrated" sample image?

Distorted? Did you notice that image was published in Adobe RGB space, so it should be view in a capable monitor?

The only 'different' thing in that image is that it's slightly underexposed, due to the necessity of phosphor colours that reflect more visible light they receive (-> causing color clipping).
Take a look at this closer crop, just with the babelcolor AVG, with a small downscaling of the tonal range (with simple white output level) to match the white: (see attached images)

2) Why target producers like X-Rite ettc. are so stupid that they don't print targets like you when it's 100x cheaper?

I haven't said they're stupid. Probably because they stick to the custom pigments because they might offer a slight advantage, but this is clearly overcome with the high resolution and the perceptually uniform distribution. Probably the custom pigments also allow them to make more money. But I believe that nowadays these is not an issue anymore.
However, 15-20 years ago it was.

3) If your idea is so clever, why not just use a target displayed on a reference color critical wide gamut monitor without the whole printing hassle?

Because photographing a monitor is a nightmare, and I know it well, trust me. At least from LCD monitors. I haven't tried from OLED monitor yet. I did careful and extensive testing on that and could eliminate 100% the moire in the very capture (no software) but there are lots of problems yet to solve, and they aren't small. It's way easier from a print.

4) All joking aside - can you provide spectral measurements of your target?

Yes, of course. I'm just starting to make the first sample and still don't have the final chart and its file, but this is one of the latest, just without some last-minute changes on skin tones and without the perimetral patches.

[digitaldog]

Hi andrew,

I'm sorry but I don't get the point. Of course ICC profiles take place after RAW data has been put into a standard file, and those data can be either.
What I said is that I'm not that sure that camera response is absolutely linear when it comes to color render.

I don't believe it matters considering where the profile is taking effect (the vast majority of raw converters doing so on the rendered data just to access the target for profile creation).

Of course if I don't like the defaults I can change that, but what conclusion do you get when comparing two products and one offers a much better result with default setting? 'you can easily change the settings'?
I'd get another conclusion, but it's up to everyone...

The settings and the profile are completely different. So it's moot if you don't like them. Plus ACR/LR don't use ICC camera profiles so moot again.

Yes, ACR cannot use ICC camera profiles, but I really doubt there are lots of good reasons.

You need to have a conversation with a fellow named Thomas Knoll; heard of him?

Just look at the facts: Capture one gets WAY better color using the same 'old' icc profiles than ACR with the new DCP profiles.

You are entitled to your opinion, you're not entitled to your made up facts. That statement is simply nothing more than your opinion of a subjective rendering and, defaults in ACR by your own admission you have to alter.

In my country you'll never see any product, fashion, ecommerce pro photographer developing with other than C1. Why?

First, please provide actual facts with metrics of how many such photographers work in your country, then the same with how many use C1 vs. anything else. Of course you've got data that examines how every such image was rendered in what raw processor? Please, if you go down such a path with such text, you'll soon see no one here is going to take you seriously.

It's very easy Andrew: it's the generic embedded profile into C1 (or ACR in that case). It's the profile that C1 automatically uses when you open your raw files. Canned profile?

Then your before and after examples are bogus (not acceptable, not good science). I told you WHY.
Are you trying to show us the effect of your target in the creation of ICC camera profiles, yes or no? IF yes, you need an apples to apples comparison and your site is far, far from doing that. Now if, big if, you really want to show the qualities of your new target, make a CUSTOM profile for C1 with any other target, then one with your target and show the differences with the SAME software and settings used to build both. Otherwise, as you have defined above how you've produced the before and after shots on your site, those comparisons are not to be taken at all seriously.

[digitaldog]

1) Why CC24 colors are distorted on "calibrated" sample image?

Why, with the exception of the shorts having far more vibrancelSaturation, does everything else in the image look (subjectively) WORSE? 

[Guillermo Luijk]

Following the exercise done here, the only drawback I see in measuring glossy patches is in the darkest ones. I have a feeling it's very difficult or even impossible to properly model the transformation between very dark patches and their expected output values because any residual reflection due to the nature of the printing paper's surface (not its real colour nor brightness in terms of printing density), even unnoticeable to the naked eye, could fool the model (profile) in those darkest patches. A null output (L=0) should only be achieved for a null input (R=G=B=0 in the RAW data), but the fact is that the input (camera) vs expected output (spectrophotometer measurements) curve seems to produce a non-null black point clipping behaviour as seen in this input G vs output L curve. I wonder if a possible solution would be avoiding the darkest patches in the card and provide the model (profile) with hard-coded R=G=B=0 to L=0 correspondences. Otherwise what we are doing is correcting veiling glare type effects inherent to the material used to produce the card that do not reflect any sensor's genuine behaviour.

On the contrary, to achieve proper rendition on very saturated synthetic colours, which are much more important for the intended applications of this SuperChroma card, I don't see any reason why a chart like this one couldn't work better than general purpose matte cards like the CC, where extreme colours need strong extrapolation well beyond the colours seen during the creation of the profile.

Regards

[Daverich]

Yes, I have lots of examples that I'm preparing for an article to be published here in the main website, with the kind help ot Kevin, main editor of this site.

If you're referring to Kevin Raber, he hasn't been connected to this site for some time.

[MauriceRR]

Following the exercise done here, the only drawback I see in measuring glossy patches is in the darkest ones. I have a feeling it's very difficult or even impossible to properly model the transformation between very dark patches and their expected output values because any residual reflection due to the nature of the printing paper's surface (not its real colour nor brightness in terms of printing density), even unnoticeable to the naked eye, could fool the model (profile) in those darkest patches. A null output (L=0) should only be achieved for a null input (R=G=B=0 in the RAW data), but the fact is that the input (camera) vs expected output (spectrophotometer measurements) curve seems to produce a non-null black point clipping behaviour as seen in this input G vs output L curve. I wonder if a possible solution would be avoiding the darkest patches in the card and provide the model (profile) with hard-coded R=G=B=0 to L=0 correspondences. Otherwise what we are doing is correcting veiling glare type effects inherent to the material used to produce the card that do not reflect any sensor's genuine behaviour.

Regards

Hi Guillermo,
I think I did the same obeservation too near black point.
Here is a plotted curve from an icc input profile (reproduction profile). I made this profile for C1 with a glossy chart from (very close to your), (many patches between L*7 until L*3.5 measured with i1pro2). I made this profile in the purpose to be as closest as possible by pushing limits of two softwares for testing (3dlutcreator+lumariverPD). My chart is shot in studio with a max controlled on glare/flare.

[Doug Gray]

Because photographing a monitor is a nightmare, and I know it well, trust me. At least from LCD monitors. I haven't tried from OLED monitor yet. I did careful and extensive testing on that and could eliminate 100% the moire in the very capture (no software) but there are lots of problems yet to solve, and they aren't small. It's way easier from a print.

Photographing a profiled monitor's image of a chart to make a profile works quite well, and isn't that difficult. But only if the monitor's white point is D50 and you want to use the profile to tag pictures the camera takes of the monitor. Otherwise it's useless.

A monitor has 3 primaries that are quite separated in wavelength. Especially wide gamut monitors. You will get large metameric error when applying these to real life camera images or even images of printed material. CYM prints at least spectrally gradually overlap and blend into each other.

[hurodal]

I don't believe it matters considering where the profile is taking effect (the vast majority of raw converters doing so on the rendered data just to access the target for profile creation).
The settings and the profile are completely different. So it's moot if you don't like them. Plus ACR/LR don't use ICC camera profiles so moot again.

This is a perfect example of someone talking about apples and another replying with pears.
Who said the setting and the profile were the same? Who said ACR/LR use ICC camera profiles?

You need to have a conversation with a fellow named Thomas Knoll; heard of him?

Yes, I know who is him, but haven’t had the pleasure to meet him.
But I guess the guy you suggest I talk to is (was) Bruce Fraser, not Thomas. Wasn’t him the guy responsible for the DCP profiles?
Ask yourself many the of most renowned professional product, fashion photographers don't use ACR/Lr.
Anyway, I’m not here to discuss that thing.

You are entitled to your opinion, you're not entitled to your made up facts. That statement is simply nothing more than your opinion of a subjective rendering and, defaults in ACR by your own admission you have to alter.

Of course that is an opinion. But it’s a widely accepted opinion.
Many people agree that when opening a photo in C1 and compare to ACR/Lr, the colors don’t look ‘washed out’, but more clean.
Have you ever used C1?

First, please provide actual facts with metrics of how many such photographers work in your country, then the same with how many use C1 vs. anything else. Of course you've got data that examines how every such image was rendered in what raw processor? Please, if you go down such a path with such text, you'll soon see no one here is going to take you seriously.

1-   Seems to me that you don’t know what were the thoughts about C1 of the person that founded this site. Take a look:
https://luminous-landscape.com/capture-one/
May I remember you that Michael was PODAS instructor?

2-   Metrics about photographers? Are you kidding me? Go outside, meet some pros (of product, fashion, e-commerce) and ask them what are they using, and you’ll see. I don’t have any data and I doubt there’s such data, but I can have a good idea based and many many photographers I know in my country. Of course I don’t know photographers in your country, but I bet it’s a very similar story.

Then your before and after examples are bogus (not acceptable, not good science). I told you WHY.
Are you trying to show us the effect of your target in the creation of ICC camera profiles, yes or no? IF yes, you need an apples to apples comparison and your site is far, far from doing that. Now if, big if, you really want to show the qualities of your new target, make a CUSTOM profile for C1 with any other target, then one with your target and show the differences with the SAME software and settings used to build both. Otherwise, as you have defined above how you've produced the before and after shots on your site, those comparisons are not to be taken at all seriously.

Ok, let me sum it up: so you’re telling me that just because I didn’t use the examples YOU want, then the examples are ‘bogus’?
What makes you think that the only valid opinion is your opinion?

I used those examples because my target customer is a pro photographer that usually don’t make any camera calibration or has done one with the CC24, but with unsatisfactory results, and usually uses ACR/Lr. That’s why I’m showing the default/custom profile.
But if you ask me for the results and comparison with custom profiles with other charts, yes, of course I’ve done them and I also did careful tests and measurements, which I already have explained in the article I’d like to publish here.

[digitaldog]

This is a perfect example of someone talking about apples and another replying with pears.
Who said the setting and the profile were the same? Who said ACR/LR use ICC camera profiles?

No one. I said your comments about ACR were moot, they are. Do examine who brought up ACR here: YOU did (Reply #14 on: July 09, 2019, 03:07:22 pm). After Mark simply stated he used that product.

The apples to pear comparison is all yours; on your site. The bogus before and after shots!

Yes, I know who is him, but haven’t had the pleasure to meet him.

Then you don't know him.

But I guess the guy you suggest I talk to is (was) Bruce Fraser, not Thomas

First, Bruce passed away many years ago so no. 2nd, Bruce and I were partners in a software company and very, very good friends. So I know the history.

Wasn’t him the guy responsible for the DCP profiles?

No. It was Thomas who did all the coding. But this is OT, you're way out of the loop on this part of the subject.

Ask yourself many the of most renowned professional product, fashion photographers don't use ACR/Lr.

That's an impossible question to ask, you're suggesting someone poll all such photographers and get their actual response as to what raw converters they or more likely, their tech's use.
I can give you a list of very well known photographers such as Greg Gorman, Jay Maisel, Steven Wilkes, Jeff Schewe, Seth Resnick, Martin Evening, Andy Katz, Art Wolfe, JP Caponigro, to name just a few, use ACR/LR.

Anyway, I’m not here to discuss that thing.

You shouldn't try, you're not up to speed on that part of the discussion.

Of course that is an opinion. But it’s a widely accepted opinion.

Rubbish, prove it or move on.

Many people agree that when opening a photo in C1 and compare to ACR/Lr, the colors don’t look ‘washed out’, but more clean.
Have you ever used C1?

Yes I have and: "All generalizations are false, including this one".-Mark Twain

You came here for feedback sir, you got it. First and foremost, your examples are bogus! It's interesting you've decided not to take up that important part of this discussion too. You're trying to show how well your target preforms and you've stacked the deck unfairly by showing a canned camera profile produced with who knows what target and software against your target with who knows what software and expect us to say "Wow, look how much better your examples are" but of course, as seen here with the image of the shorts, it looks WORSE!
You have a lot of work to do sir, after you've corrected the bogus examples on your site with a fair comparison of JUST what your target can do, we can move on.

[hurodal]

Why, with the exception of the shorts having far more vibrancelSaturation, does everything else in the image look (subjectively) WORSE? 

Ok, I think you’re going too far. I replied and gave arguments, screenshots and answers to every single critic that you told me; all I see from you is a person that simply jumps to another new small detail to criticize when I reply to the last one. And in this your last post, you're laughing about the project I've been working slowly for 4 years.

Let me tell you something, Andrew: when I came here to share this my project, I expected to have a nice discussion with many experts here. A discussion on where some would like my chart and some don’t. I will totally respect their opinion, of course, but as well as I would never throw shit over other person’s project, I expect them to not do that to me. Because I know he may have spent hundreds, thousands of hours on it, that would be a really dirt thing from myself to do that. Because that hurts.

There always should be a respect for others work, because we can all discuss without going low.
Let's say you have written a book (I know you did) and come to a forum to share it. I’d first ask about several things. Never criticize first. Get all the important information, then I can discuss with arguments. In the end, if I don’t like it, I just don’t buy it! But I'd never start a ‘campaign’ to discredit, even less with childish arguments that take the tangent path from the important stuff.

Now I see you’re not that kind of person. I hope you don’t face with that in your own person from others.

So now you’re telling me that the everything else in the image looks 'WORSE'. And you laugh.
You might not seen the answer and images of the CC24 I've already posted, on where it can be seen that the profiled capture is almost spot on.

But after you said that, now I can see that you ignore how phosphor colors behave, which in the end tells me that you might know Ps very well, but regarding camera profiling you're lacking quite some knowledge.

Also, if you ever come to a Spanish forum on where I use to be, I will first welcome you (because that’s gentleman do when a new one comes and introduces himself) and then will start to discuss. But I won’t try to find out any mistake or misspelling in your book just to hit hard with my finger on it. In fact, if I find some I will email you privately just to help you correct it. That’s my DNA, and i.e. Guillermo knows that.

I don't know if you know Chris Murphy or if he is also here, but you can ask him how did I treated him when I managed to bring him to a big intl. photo show in Barcelona a few years ago. We both performed a conference about color management, together.

Would you like if I read your book and post here every single and small mistake I find out? And, trust me, I will find them. Even that it’s not in my mother language.

Now, that said, let me answer the main question: That 'strange', 'WORSE' appearance is totally normal when faithfully reproducing phosphor colors, because they reflect more visible light than they receive, thus need under-develop, to avoid color clipping. That makes the non-phosphor colors underexposed.
The CC4 was almost spot-on, except for the scaling down of the RGB values.

Now, let me tell you that I wasted enough time replying to this; I won’t be replying unless you keep a respectful position.

PS: Regarding your last post, If you ever asked me to add another examples to make the comparison more fair/complete/real/whatever, I’d kindly find some time to add them. Now you achieved the opposite from me.