Flat Art Reproduction Black and White Points?

[Craig Stocks]

I have a question for the folks who do art reproduction photography for other artists. How do you handle black and white points? Color management is pretty straightforward (I use Lumariver custom profiles) and can get a very good match to Color Checker values. Even matching the luminosity of the grayscale tones is pretty easy.

The problem that sometimes comes up is that the art has very flat tones. I match the Color Checker values but the art looks flat and dull. It's not unusual to photograph a work that only has a 4 or 5 stop range of luminosity. If my goal is to make a print then I just make the print match the original. But if they just want digital images for their records or to jury for exhibitions, what do you do? Do you stretch the tones to get closer to true white and black points, or do you let it look flat?

Note I'm not talking about creating digital records for a museum, simply recording artwork for artists for their own use.

The example here is fabricated to illustrate the point.

 

[Shashin]

I prefer the first image. In a reproduction, the point is to replicate the original rather than to optimize a histogram. Since I assume you are giving digital files to the artist, then they can modify the file if they wish. The artists I have worked with are more concerned about replicating their work than modifying it into something else.

Your example is a little odd as the original image file would be better than its reproduction, unless printing and paper choice is a significant part of the process, which does not seem the case here. I think if you had used a painting, you would would have preferred the technically accurate image.

 

[Craig Stocks]

The example is just one of my photos manipulated to illustrate the point. I didn’t want to use someone elses artwork for the example.

What I tend to see is that a painting that looks great in person looks flat as a digital file. Many of the artists are using the images for submissions to a jurried show so it needs to be presented in its best light but still true to the original. The painting typically has something of a white point point and a black point, it’s just that the white isn’t very bright. When it’s viewed on a computer monitor it comes out very dull since the monitor’s white is much brighter than the painting’s.

For instance, numerically the painting’s white point ends up around 220 and black around 30 so it’s numerical correct but visually wrong. If I stretch it so the painting’s white is 250 and black is 10 then it’s visually more like looking at the original but it’s numerically wrong.

 

[Boinger]

You could give the artist both versions. And simply call one for screen and the other is 100% accurate reproduction.

Let them choose what they want to use?

 

[dougpeterson]

I have a question for the folks who do art reproduction photography for other artists. How do you handle black and white points? Color management is pretty straightforward (I use Lumariver custom profiles) and can get a very good match to Color Checker values. Even matching the luminosity of the grayscale tones is pretty easy.

The problem that sometimes comes up is that the art has very flat tones. I match the Color Checker values but the art looks flat and dull. It's not unusual to photograph a work that only has a 4 or 5 stop range of luminosity. If my goal is to make a print then I just make the print match the original. But if they just want digital images for their records or to jury for exhibitions, what do you do? Do you stretch the tones to get closer to true white and black points, or do you let it look flat?

Note I'm not talking about creating digital records for a museum, simply recording artwork for artists for their own use.

The example here is fabricated to illustrate the point.

I don't know why you consider this separate than what museums/archives/libraries/galleries do. Your goals and clients are the same as a museum. Museums/archives/libraries/galleries went through the same existential questions you are asking. They just went through them in the late 90's and early 2000's. Their conclusion is resounding and nearly unanimous: a Preservation Digital Object (aka "linear reproduction" or "master file"; terms vary by source and authority) should always be the primary deliverable; all other deliverables should be marked as and treated as derivatives thereof.

If the artist wishes to make a more contrasty derivative (or if you wish to provide the creation of such additional derivatives as a value-added service) that is more appropriate for stand-alone use on the web (an environment in which a linear reproduction will be perceived by many lay people as too "dull") that is a fine idea that will increase the value and immediacy-of-use-across-different-use-cases of your deliverables.

Consider, for example, the use case of the client wanting to take 10 images you've delivered them, and adding contrast to be more suitable to web presentation. If you've custom-adjusted each of those 10 images then the suitable amount of contrast they should add or remove will vary per image. If you've made a Preservation Digital Object for each of those 10 images they can often make the same change to all 10 as a batch. Moreover, if they want to make a print reproduction of the original they will simply be able to print (within a high quality color managed environment) directly from the files you've provided them, rather than requiring custom adjustment to back out the contrast you've added.

We cover, in great detail, exact methodology, theory, and quality control for the creation of a Preservation Digital Object using Capture One here:
Fine Art Reproduction and Linear Reproduction in Capture One (the Reflective guide would be the one salient to your current work)

 

[Jack]

I first asked if they wanted as exact a digital copy as possible, or did they want it digitally "cleaned up" to look more like it might have as an original. A good example is an old yellowed B&W print or newspaper article. My point was if they were having me clean up dust spots or tears in the original, I might as well make it look something like what it did when new...

From there I had the direction to proceed. Me personally with the latter, I always set my white point to just under paper white at around 248-252 depending. I'd then set my Black to whatever the paper could hold, usually was around 004 with my photo papers and Epson ink combo at the time. In the end with these parameters, I never had anything banded, blown or blocked up. YMMV

 

[dougpeterson]

I first asked if they wanted as exact a digital copy as possible, or did they want it digitally "cleaned up" to look more like it might have as an original. A good example is an old yellowed B&W print or newspaper article. My point was if they were having me clean up dust spots or tears in the original, I might as well make it look something like what it did when new...

From there I had the direction to proceed. Me personally with the latter, I always set my white point to just under paper white at around 248-252 depending. I'd then set my Black to whatever the paper could hold, usually was around 004 with my photo papers and Epson ink combo at the time. In the end with these parameters, I never had anything banded, blown or blocked up. YMMV

Restoration is a great value added service to provide for digitization/reproduction.

But in my opinion it should be delivered as a derivative of the Preservation Digital Object. Having both the stark difference between the two provides a powerful example of the value you brought to the table. If people only see the "after" photo of a kitchen renovation project they are unlikely to appreciate how much work went into it. As a separate matter, delivering the Preservation Digital Object also means that 40 years from now they (or a descendant) have an unadulterated starting point if they want to do their own restoration work (maybe by then AI and Deep Learning will mean a single-click can outperform the quality of the restoration you performed by significant manual effort).

 

[Craig Stocks]

Doug - Thanks for the link and the thoughts. I've downloaded the guide and am reading through it.

Doug, Boinger & Jack - Thanks for the ideas. I'll start offering both versions as an enhanced service.

 

[Shashin]

The example is just one of my photos manipulated to illustrate the point. I didn’t want to use someone elses artwork for the example.

What I tend to see is that a painting that looks great in person looks flat as a digital file. Many of the artists are using the images for submissions to a jurried show so it needs to be presented in its best light but still true to the original. The painting typically has something of a white point point and a black point, it’s just that the white isn’t very bright. When it’s viewed on a computer monitor it comes out very dull since the monitor’s white is much brighter than the painting’s.

For instance, numerically the painting’s white point ends up around 220 and black around 30 so it’s numerical correct but visually wrong. If I stretch it so the painting’s white is 250 and black is 10 then it’s visually more like looking at the original but it’s numerically wrong.

I have found the quality and intensity of the lighting is important. But it also depends on the original. For example, if you use large soft boxes with oil paintings, the image does not contain the texture of the original and looks flat. A harsher light can be much better. Paper works made of several pieces need a top light to create a shadow to give dimension to the elements, but needs to have the entire surface evenly illuminated so not to show too much fall off from top to bottom. What was a pain was when you had multiple material types, paper, metal, and transparent plastic. Lighting choices were hard to optimize. I was always lucky to have artist with me so we could work on how to interpret the work and discuss how the light changes the emphasis on the material in the image. Most artists I work with are trying to reproduce the work "faithfully."

 

[yaya]

Craig remember that "the look of the original" is largely dependant on the lighting conditions in which it is being viewed, and the lighting used during digitisation. If you and/ or the artist are viewing it under e.g. room lights or modelling lights, and you then use flash or consumer-grade LED (which will normally "spike" in the blue channel) to capture it, chances of experiencing the same "look" on scree and on the wall are very, very small.

Print reproduction is a whole different story as that will involve converting data from RGB to CMYK with different gamuts and colour spaces, using UCR/ GCR to optimise the amount of "K" being used and so on...

As other suggested, Creating a flat looking, linear digital image is the better option and then you can offer "good looking" variants or let the client find the look they prefer.

BR

Yair