Camera calibration for reproduction work

[MischaR]

Hi everyone,

I recently started using my H5D for reproduction work, and I'm trying to get my captures as close as possible to the specifications set by FADGI and Metamorfoze.
Phocus features a tool with which I can calibrate my Hasselblad, using a color target like the X-Rite Color Checker.

A great tool, but I'm not too sure about how I can make adjustments in the color calibration window to set the limit of the DeltaE value to something else than the default of 3.0.
I notice that this isn't a set limit, but that I should be able to select another value.
But when I do this (for instance select value 2.0) it automatically "jumps" back to value 3.0 when I click on the "calibrate"button.
At the DeltaE 3.0 setting I don't see any serious deviations from the desired color rendition, but setting the slider at 2.0 there are four warning signs, and at a DeltaE setting of 1.0 thirteen(!) patches show the orange triangle.

So what's the correct procedure to calibrate for another DeltaE limit? 

Who, with a little more experience with this tool, can help me out?
Thanks!

Cheers,
MischaR

[Alex]

Hi MischaR

The colour calibration tool does not have any tuning facility;
What it does is create a custom profile that changes the colour and tonal reproduction of the camera profile based on the reference information it has of the target patches versus the target patch values within the photograph. The changes made have an effect throughout the whole of the profile space (squeezing 1 part will cause a stretch in another part), rather than selectable focused areas.

Due to this the correction cannot always bring all of the calibration points/patches within tolerance (<1DeltaE), and so one may have patches that still lie far out from their desired values.
Some of these points/patches aren't of importance in terms of the subject matter's colour range and so can be discounted. The point of the slider is to show you which colours are within tolerance which ones you need to watch out for in terms of the amount they deviate from their reference (these would need individual attention in post later on).

So if you want to decrease your deltaE values, your best bet is primarily to address the quality and positioning of the illuminant as this will have the greatest effect. Secondly get your target measured and update the reference file with it - this can really make a difference. Lastly, remember that the colour correcting is quite basic (it's based on Lab* values rather than Spectral), and it's not in the same league as Basiccolor INput or Lumariver Profile Designer; certainly i have seen times where it has help make more of a mess of the colour than make it better, so I wouldn't see it as an intermediate or halfway solution for correcting Max DeltaE's larger than 2.

[MischaR]

Hi Alex,

Thanks for your reply.

This answers my direct question about the functionality of the colour calibration tool.
But as you correctly assumed, this is only the first of a series of questions I have to find answers to, to be able to make the necessary adjustments during my workflow and arrive at a reproduction that conforms to the specifications of the mentioned institutions.
So I expect to encounter more smaller or bigger "bumps" to overcome during my learning process :-)

The interesting thing is that it's not really the color patches that are out of tolerance, but the grey ones.
In ISA Golden Thread the tonescale looks like a perfect "wave"for surf enthousiasts, but it's really out of bounds in the low as well as in the high densities.
I'm not too sure how correct this.
The curves tool in Phocus doesn't really help as both extremities are fixed and you only can manipulate what's "in between".
I tried to use the curves tool in Photoshop CS6, and the result was a small improvement, but the changes I have to make are too subtle, so this isn't the solution either.

Now you mentioned Basiccolor INput as a means to create a profile to solve these problems.
Searching on the Internet for more information on calibrating the camera I came upon the "Fine Art Reproduction Cofiguration Guide" by Scott Geffert from ImagingEtc.
As it adresses calibrating Hasselblad camera's in Phocus it seemed to be exactly what I needed to create the perfect profile for this purpose.
It dates from 2012, and some screen shots show input panes that I don't recognize from my version of Phocus.
But I guess the process he describes in the manual is basically the same.

What threw me off, hesitating to continue following the procedure described herein, is the price of this piece of software.
I'm not too thrifty to spend some money on a good application if this is a good investment which will help me to reach my goal.
But if there are other ways to bring this about, at least I'd like to compare the different options.

You mentioned Lumariver Profile Designer which is available for less than half the price of Basiccolor.
I can download trial versions of both.
And then there is "BabelColor".
But it would be of great benefit if I could make a decision also based on advice from someone who has worked with this kind of software.

Best,
MischaR

[Alex]

The interesting thing is that it's not really the color patches that are out of tolerance, but the grey ones.
...it's really out of bounds in the low as well as in the high densities.
I'm not too sure how correct this.

Lighting uniformity is what is going to correct this - Essentially you need to be testing your lighting area with a uniform subject matter - like a white wall/board/background paper, that'll help direct you in the positioning of the light sources to give an even distribution, then you want to use a dark surface with a slightly satin finish to show up any spectral light source reflections coming from the angle of the light source/camera (which may require further re-positioning before going back to the white surface.

Ideally you don't want to be using the profile to be correcting tonal issues as this is meaning that the colour patches are not receiving optimal lighting  they need in the first place.

Searching on the Internet for more information on calibrating the camera I came upon the "Fine Art Reproduction Cofiguration Guide" by Scott Geffert from ImagingEtc.
As it adresses calibrating Hasselblad camera's in Phocus it seemed to be exactly what I needed to create the perfect profile for this purpose.
It dates from 2012, and some screen shots show input panes that I don't recognize from my version of Phocus.
But I guess the process he describes in the manual is basically the same.

Yep Scott's methodology is the tried and tested one that works - as it was written in 2012 Hasselblad hadn't yet released Low Gain Reproduction mode, which as the most linear tone curve you can set in Phocus is the most ideal curve to set the images at.

I can calibrate my Hasselblad, using a color target like the X-Rite Color Checker.

If you're trying to hit Metamorphoze image requirements - you want to be a least using an XRite SG Colorchecker Device Level Target (you can pick them up with individual readings from ISA (http://www.imagescienceassociates.com/mm5/merchant.mvc?Screen=PROD&Store_Code=ISA001&Product_Code=XRITECC&Category_Code=TARGETS))

What threw me off, hesitating to continue following the procedure described herein, is the price of this piece of software.
I'm not too thrifty to spend some money on a good application if this is a good investment which will help me to reach my goal.
But if there are other ways to bring this about, at least I'd like to compare the different options.

Yep Colour management costs money - why do you think Phase One cost what they cost - they're actively involved in the getting in creating solutions for the Museum/Archival/Cultural Heritage world. Check out the DT Photon XL reproduction lights - they cost a good deal. However if you want top quality accuracy you do have to pay - INput is relatively the most expensive option out there but it really does a great job with ICC profiling and does it so easily. LRPD (reproduction vers), cost less, does more but also requires a bit more study and tweaking. I forgot that there's also  a free option out there - which many people bypass: Delt.ae  deltae.picturae.com: as well as checking your targets for compliance they also create colour profiles. Unfortunately you do have to manually enter in all the Lab* values, which can be mind numbing for the 140 patches of the SG target!

[MischaR]

Hi Alex,

Thank you for your comprehensive explanation, and taking the time to answer all my questions.
This is a good starting point for me to delve a little deeper into the world of cultural heritage digitization.
I'll have a look on the Delt.ae website as well.

Best,
MischaR

[MischaR]

Hello Alex,

I have one more question about using the reproduction tool in Phocus.
I hope you won't mind taking some time to send me your reaction.

I'm running some tests, using the reproduction tool, and slowly but surely I'm getting there.
The test results have improved dramatically, and I'm getting very close to the FADGI 4-star and the Metamorfoze specs.
Calibrating the camera, and subsequently using the profile when capturing the test targets, definitely improves the tonal reproduction.

But there's still one aspect unclear to me concerning the use of the reproduction tool.
Neither the Phocus Manual, nor the Hasselblad tutorial video clips on YouTube by Karl Taylor are very clear on this, at least in my opinion.

As I understand the procedure it's as follows:
1. De-activate all adjustments in Phocus except the Reproduction Tool (both boxes checked) and use the default settings
2. Capture a test target (in my case a Color Checker Digital SG) after neutralizing on patch G5 and lighting and exposure set to L*on G5 = 66 and on E5 = 97
3. Open the 3F file and choose between Hasselblad RGB and LRGB, select "Edit" in "Color Calibration", perform the calibration and create the profile
4. Use this new camera profile to capture the Color Checker target
5. Export this 3F file to a  "Full Size" Tiff-16 file and select "Source" as "Output Profile". Run the FADGI or Metamorfoze tests on this file.

What is not clear to me is the function of the Reproduction Tool in this last stage.
Should it be activated while exporting the 3F to the Tiff-16, or not?
In other words; does the camera profile directly affect the RAW file that is produced at the moment of capture?
Or does it correct the RAW image information when it is exported to the final Tiff file?

I would appreciate your advice.

Thanks,
MischaR

[Alex]

Should it be activated while exporting the 3F to the Tiff-16, or not?

In other words; does the camera profile directly affect the RAW file that is produced at the moment of capture?
Or does it correct the RAW image information when it is exported to the final Tiff file?

We'll the custom made assigned profile is changing the quality of the image and apparently this is happening at the base level (before everything is else is applied

[Alex]

... whether this is before the scene calibration is applied, I'm not sure), however being assigned, it does not apply the profile until one exports it. So this means one can chip and change without worrying about damaging the original 3FR RAW - cool.

Does this help?

Alex

[MischaR]

Hi Alex,

I tried to find some more info and found some explanation of the process in which is mentioned that the camera profile, when applied, is imbedded in the raw fff-file.

I ran two tests exporting the fff to a Tiff-16 of a raw file (captured with repro profile), one with the repro tool checked and the other one unchecked.
It looks like it applies the profile twice when exporting with the tool being activated, with unwanted results.
So I guess that confirms your suggestion (and my reasoning).

I'm reasonably new to working with profiles in my photographic workflow and I can understand the theory of of it.
But I find that implementing this in my own practical workflow can sometimes be a little confusing, especially when at the same time I'm trying to familiarize myself with a new piece of software like Phocus.
But I'll get the hang of it sooner or later...

Thanks,
MischaR 

[Alex]

I ran two tests exporting the fff to a Tiff-16 of a raw file (captured with repro profile), one with the repro tool checked and the other one unchecked.

When you say repro tool do you mean the Reproduction module i.e. as in Exposure, White Balance?
When you say repro profile do mean the Reproduction response curve? or do you mean your custom Color Calibration?

It looks like it applies the profile twice when exporting with the tool being activated, with unwanted results.

It's always good to know what's really going on;

• Your safest route is to first of all do all your captures with the Reproduction module activated, your Working space set to Hasselblad L Star, the Response curve set at Reproduction Low Gain, & your export recipe/preset set to a Source RGB ICC (IOW Hasselblad L Star) profile.
• Set your Photoshop colour settings to the Hasselbald L Star RGB working space, conversions Absolute Colormetric (who know what intent Phocus uses for their conversions), and all warnings to On. In this way you'll know what you're getting and when you open files in Photoshop that aren't Hasselblad L Star RGB you won't have images being silently converted to Photoshop's working space profile without knowing about it.

Once you have finalised your edit you can export the finished image in whatever dodgy colour profile the world might want off of you (probably using a Relative Colormetric, intent would be the ideal in this case)

I'm reasonably new to working with profiles in my photographic workflow and I can understand the theory of of it.
But I find that implementing this in my own practical workflow can sometimes be a little confusing, especially when at the same time I'm trying to familiarize myself with a new piece of software like Phocus.

It's worth booking some time in with a reputable Color Management consultant (who'll educate rather than try and sell you things), it'll get you on the right track and if you're procative will hopefully get you to see the world, the equipment and processes required to achieve the desired result in the correct way.