Photographing Artwork For Reproduction

[Garnick]

Hello all,

Somewhat OT here I suppose, but I'll give it a go regardless. I'm sure some of you good folks are reproducing original pieces for artists. Over the past few years this has become a rather large part of my business as well, but I do have a question concerning shooting original art. For the most part I am scanning art, anything up to 20x24 and a couple of 20 x30 pieces as well. I'm using an old Epson Expression 1680 and it's doing an excellent job. Of course I do have to scan sections and then reassemble in PS, but that's seldom a problem. Occasionally I am presented with a piece that simply will not scan properly due to the medium and its reflective properties. The only draw back to scanning in this manner is the fact that the scanner captures EVERYTHING. Now you might say, well, that's a good thing, but not always. And I'm sure some of you folks know very well what I mean. When I have an original that presents this sort of problem I have a friend shoot it on camera in his studio. Now of course that would be great if he had a 4x5 camera back and a load of resolution, but that's not the case. 14MP is the best he can do and quite frankly it just isn't enough. However, I've been thinking that perhaps we could devise a setup where he could shoot sections, as I do with the scanner, and capture a higher resolution per section. Then I can stitch them together as usual and have more res and hence more detail. Could also control the lighting better to facilitate the capture of brushstrokes etc as with a Cruse Scanner. I suppose in essence I am trying to construct a poor man's version of a Cruse Scanner, but I'd like to get some feedback from the group if any of you are approaching this task in such a manner. Perhaps some sort of description of your setup etc. Any info would be greatly appreciated.

Thank you,
Gary 

[fetish]

If his 14MP is a digital back, simply use the same 4x5 and shift the rear standard around the image circle. the images will stitch flawlessly.
If not, invest in a Gigapan robot and use a telephoto lens.
lighting wise, you need polarized high CRI continuous lights. Proofing light tubes, HMIs and kinoflos work well.
http://teddillard.wordpress.com/2011/10/24/on-fine-art-reproduction-or-youre-doing-it-wrong-part-1/
http://www.guildsourcebooks.com/pdf/sourcebook/howto_photograph_artwork.pdf

[epatsellis]

Your best bet is a Betterlight back, or find a used Dicomed Field Pro if money is an issue. Probably the easiest and best way to consistently achieve color accuracy, high resolution and repeatability.

[John Nollendorfs]

I've been doing exactly this for over 10 years--even before stitching software became available. While I've read about many paradigms in doing this, including mounting the camera body on a 4x5 and using lens shifts, I prefer my simple method of using a camera stand in the studio, and using cross-polarized strobe lighting. I use the same camera to do studio shots from portraits to commercial product shots.

I shoot everything in RAW, and set up custom presets in CS 5 Bridge. then have CS5 stitch them together. I overlap shots by about 25%.

[epatsellis]

I've been doing exactly this for over 10 years--even before stitching software became available. While I've read about many paradigms in doing this, including mounting the camera body on a 4x5 and using lens shifts, I prefer my simple method of using a camera stand in the studio, and using cross-polarized strobe lighting. I use the same camera to do studio shots from portraits to commercial product shots.

I shoot everything in RAW, and set up custom presets in CS 5 Bridge. then have CS5 stitch them together. I overlap shots by about 25%.

John,
Your approach is certainly valid, as evidenced by your using it for the last decade. I used to do the same, until I spent some time with my Dicomed back, the color and detail differences that a tri-linear array capture are drastically different than a Bayer array sensor. There's several good white papers and a comparison between the Betterlight, a DSLR and film on the Betterlight website.

[John Nollendorfs]

There is no doubt, that the Dicomed will deliver excellent results. Not much argument there, but spending $10k for just the camera back is a healthy investment. A camera that only works well for copy shots. Although still supported, this is really old technology, and requires quite bright continuous light source for exposure--many use HDMI light sources, again quite expensive. To be adequately set up, requires easily $15k--you better be doing a lot of copying to justify that!

In contrast, you can pick up the D800e for $3300 and a couple of strobes with polarizing filters for another $500-1k--all of which can be used in other photography work.

The real determining factor is how good your prints look compared to the original, not pixel peeping dicomed vs bayer files on a screen.

[dgberg]

Gary,
No intent on hijacking your thread. Just a couple of notes relating to my experience.
 As John notes the D800 models get you into this for considerably less then a MF rig.
Just finished shooting a half dozen watercolors with D800E and Elinchrome light system.(BXri 500's)
Lights set at 45 degrees take very few adjustments to get right.
Shot a grey card to set a custom preset white balance and the raws look like you would expect on screen.
After some minor adjustments I printed several proofs and the colors are all more saturated then the origonial.
Probably spent an hour fooling with the color on each one and got them close but not perfect.
Plan on ordering the Color Checker Passport to see if that helps.
D800e's files are razor sharp and cannot imagine needing any more resolution then what this camera puts out.
As soon as I solve the color issues should be good to go.

[fetish]

Dan, the color passport does help a lot in profiling the camera's sensor. I have 2, 1 in my studio and 1 travels everywhere with my main camera system.
It integrates painlessly into lightroom and ACR and cuts the pain in color artwork reproduction by a lot. =)

[Craig Murphy]

Aside from having a proper camera creating a custom profile for the specific lighting, camera, lens combo I think is the most important.   You can really see the subtle color changes on the checker when grabbing the profile in Lightroom.  Do you think its necessary to shoot a grey card and do a white balance ahead of time?  I'm not seeing that as needed with the profile production as long as your not shooting totally out of color temp range.  I grab the profile and then white balance off the second lightest gray patch.   

[Garnick]

Gary,
No intent on hijacking your thread. Just a couple of notes relating to my experience.
 As John notes the D800 models get you into this for considerably less then a MF rig.
Just finished shooting a half dozen watercolors with D800E and Elinchrome light system.(BXri 500's)
Lights set at 45 degrees take very few adjustments to get right.
Shot a grey card to set a custom preset white balance and the raws look like you would expect on screen.
After some minor adjustments I printed several proofs and the colors are all more saturated then the origonial.
Probably spent an hour fooling with the color on each one and got them close but not perfect.
Plan on ordering the Color Checker Passport to see if that helps.
D800e's files are razor sharp and cannot imagine needing any more resolution then what this camera puts out.
As soon as I solve the color issues should be good to go.

Hi Dan,

Thank you for this reply.  Sounds like the path I might be traveling as well, since I cannot rationalize a digital back now.  From what I have read so far it seems that the D800E is winning over the D800.  Perhaps you could expand on that a bit if you don't mind.  I haven't had an opportunity yet to really discern the main differences in the two models, but if I do decide to go this route I want to be sure I'm making the right choice. 

Sorry, it's a busy day and I'm a couple of hours behind already, so I'll have to get back to work and stop 'playing' in LULA for now.  Actually I look at LULA and other related forums as being an integral part of my work in most cases, but not at the moment.  The work has to come first and then the research, as we all know.

Thanks again and I'll be watching.
Gary

[epatsellis]

There is no doubt, that the Dicomed will deliver excellent results. Not much argument there, but spending $10k for just the camera back is a healthy investment. A camera that only works well for copy shots. Although still supported, this is really old technology, and requires quite bright continuous light source for exposure--many use HDMI light sources, again quite expensive. To be adequately set up, requires easily $15k--you better be doing a lot of copying to justify that!

In contrast, you can pick up the D800e for $3300 and a couple of strobes with polarizing filters for another $500-1k--all of which can be used in other photography work.

The real determining factor is how good your prints look compared to the original, not pixel peeping dicomed vs bayer files on a screen.

John, the Dicomed Field Pro that I have is the predecessor to the Betterlight, top of the line '90s technology. Mine cost me all of $750 with a Macbook Wallstreet. Even factoring in self built HMI (really high CRI MH) or high output florescent fixtures, it's still well under the cost of the D800E body alone. Mine gets used either with a Sinar or my RB67, for 6k x 6k captures. Not singleshot and requires some skill and experience, but a high resolution, high accuracy digital capture for under $1k deserves mention.

[John Nollendorfs]

John, the Dicomed Field Pro that I have is the predecessor to the Betterlight, top of the line '90s technology. Mine cost me all of $750 with a Macbook Wallstreet. Even factoring in self built HMI (really high CRI MH) or high output florescent fixtures, it's still well under the cost of the D800E body alone. Mine gets used either with a Sinar or my RB67, for 6k x 6k captures. Not singleshot and requires some skill and experience, but a high resolution, high accuracy digital capture for under $1k deserves mention.

I just checked the Better Light site, and they only have two models "not sold out". those are listed for $15k and $18k. Even checked "fleabay" and they had nothing. It sure looks to me like Better Light is near the end of it's life. I can't see how it can really compete with the new cheaper camera  models such as the D800E, and the great stitching software in  PS. 

Yes, in theory, Betterlight is a solid product. But it's so one dimensional. I would bet that even Joseph Holmes has quite using it for his landscapes, because of it's limitations in the field.

[teddillard]

Thanks for the link, sir!  

IMHO, and due respect...  I think you guys are missing the point.  It's about the lighting.  

I've written a lot about it on the thread linked so I won't repeat myself, but here's a hint: a painting is a 3-dimensional object.  Standard copy-lighting with polarized sources etc doesn't nearly touch it.  

The camera is almost immaterial.  I've shot artwork and compared files from a lowly Nikon D5000 (albeit with the legendary 55mm Micro-Nikkor) that were stitched to the same painting shot with the Cruse, and the Nikon files are far better, both from a color and resolution perspective.  You want more detail?  Shoot closer, and more samples.  (I was a Cruse operator for over 2 years, I know the camera... and I've worked with Betterlight, Phase, Hasselblad, Leaf and even Imacon and MegaVision for over a decade.)  The secret to the color management is the Passport profiling - can't rave about that thing enough!

But if the lighting doesn't reproduce the viewing light of the painting, then you're missing the target entirely.  I can almost guarantee one thing...  there is no artist in the world working with polarized lights at 45º positioned on either side of their painting.  They never saw their work lit like that, they didn't paint it with that intention.  Why do we shoot it like that?  (If you're talking posters or print reproduction, OK, that's fine.  But anything that's painted, and has any texture, even just the tooth of a paper needs something other than copy-board lighting.)

I've developed an easel that is still in it's gestation period, but it's an x-y easel.  It can hold the art, and move it up, down and sideways relative to the camera, just like a CNC router.  The frame stays stationary, and is about 16 x 20", so that's the only area you have to light.  Because of this, it's far FAR more powerful than the Cruse, or any flatbed scanning method, and a lot easier than any method of moving the camera.  

Here's a bit about that:
http://teddillard.wordpress.com/2012/05/04/the-secret-of-the-16x20-frame-dslr-fine-art-reproduction/

[John Nollendorfs]

Thanks for the link, sir!  

IMHO, and due respect...  I think you guys are missing the point.  It's about the lighting.  

I've written a lot about it on the thread linked so I won't repeat myself, but here's a hint: a painting is a 3-dimensional object.  Standard copy-lighting with polarized sources etc doesn't nearly touch it.  
SNIP

But if the lighting doesn't reproduce the viewing light of the painting, then you're missing the target entirely.  I can almost guarantee one thing...  there is no artist in the world working with polarized lights at 45º positioned on either side of their painting.  They never saw their work lit like that, they didn't paint it with that intention.  Why do w e shoot it like that?  (If you're talking posters or print reproduction, OK, that's fine.  But anything that's painted, and has any texture, even just the tooth of a paper needs something other than copy-board lighting.)

Ted, I won't try and sway your mind about photographing the painting to capture the "impasto". For some art works you want to try and capture that, but for others you don't. It depends on the artist you are doing the work for. I've photographed the works of two landscape artists, one that basically uses a dry brush technique, with little to no "impasto", and another that uses fairly heavy oils with a lot of build up. They both prefer the "no shadow" flat style of copying. Have another artist that does a lot of multi-media work, painting on board panels with lots of textural effects and objects attached. For him, I use a large softbox to set the lighting, and then fill with a large reflector. Every artist will be different, in how he wants the work photographed. You have to be mindful of that when they come to you, and ask the right questions.

[teddillard]

I totally agree.  

As my posts on the blog explain, the entire process was developed in response to a client's feeling that nobody was reproducing his work faithfully.  

Just to keep on topic, though, if the OP needs more resolution, he simply needs to move in, and shoot more samples with his DSLR.  If he's shooting with an X-Y easel or table, then he simply needs to light the frame he's shooting, which is stationary. 

[BobDavid]

Think Cruse scanner or a multi-shot MF camera back, so you won't be fiddling endlessly trying to match color. A 35mm dSLR simply doesn't have what it takes to efficiently handle fine art repro. 30 years in the industry confirms this.

[teddillard]

Think Cruse scanner or a multi-shot MF camera back, so you won't be fiddling endlessly trying to match color. A 35mm dSLR simply doesn't have what it takes to efficiently handle fine art repro.

...a very common misconception.  If you're interested, the Passport software is a free download that can be used with any Colorchecker.  Get it here:
http://xritephoto.com/ph_product_overview.aspx?id=1257&action=support

...and judge for yourself.  My experience, from -over- 30 years in the business (lol), is that, with good color management front to back, the DSLR workflow for the cameras I've printed with is more accurate - and printable - than a Cruse. But by all means, try the systems out yourself, by renting, begging, borrowing or hiring before dumping tens of thousands of $$ at it.  If you do rent, hire a good production guy with experience with the MF software so you know it's being done right, too, BTW.  

To that point, compared to a MF digital back, I'd have to say it depends on how well you understand the software.  Knowing Sinar, Phase, Leaf, Hasselblad and various others pretty intimately, my experience is if you apply the same level of pre-RAW processing to your color management I'd say you're going to be in the ballpark, but I've never had first-try color results like with the DSLR with a Passport-ACR workflow.  Even Betterlight, for all it's remarkable color management, didn't get there so easily.  

And yes, the Cruse employs pretty typical color management, but as with most tri-linear CCD technology, it's not applied in the pre-TIFF processing.  

As with film, back in the day, it's all about how you process the file.  I found it hard to believe, too.  I tested it because of a comment by a fairly casual observer, to tell the truth.  "The Cruse is pretty old, isn't it?  Isn't digital technology sort of like dog years?"  Not so long ago, tri-linear CCD capture was the absolute top of the heap, followed by mutli-shot MF.  Today the Bayer-array processing of even the most humble DSLR it's a different animal.  It's come a long, long way since I first started working with them in 1998.  

...and the lighting.     'Nuff said.  I'm not giving away any more of my secrets.   

[Ernst Dinkla]

My 2 cents.

Lighting is important. A continuous spectral distribution at 5000K or a bit lower gives predictable color with DSLR sensors + color calibration tools. I am not so sure that Color Passport has enough pigment variety to cover all the substrates, pigments and paint mediums one encounters in reproduction photography. Increasing the number of pigments in the calibration card is one thing but not the only route and at some point there is no more gain in the number of calibration pigments with one light source + RGB sensor filtering. The other route is multi spectral photography which also allows a better use of an increase in the calibration card pigments as it can make the better distinction of different pigments in the reproduced originals and the calibration card. A similar approach with more scans or shots using different lighting sources probably is as effective to define the used pigments. Both methods have to rely on good algorithms to create the best RGB image of the original. A cheap example of the last method is found in two HP flatbed scanners, the Scanjet G4010 and G4050. The scanners actually do not have continuous spectral lighting but two spectrally different ccfl lamps that are used one at a time in a two scans sequence. Smart algorithms do the rest and HP has done a lot of color research over the last decades. Image Engineering in German has a test of the scanner at its site. A cheap approach of multi spectral filtering was a Sony camera with a sensor that had the normal Bayer mosaïc filter improved by changing one of the greens to a more cyan filtering. There are more examples: http://en.wikipedia.org/wiki/Color_filter_array

About a year ago I was intrigued by a message of Klaus Karcher on the Apple colorsync list where he mentioned 60000 measurements done of pigments, paints etc  and a calibration target of 809 patches based on that research. Klaus is a Cruse expert. I expressed my curiosity how that could be used properly not thinking yet of a kind of multispectral scanning. Later I thought it could only work with some kind of multi spectral method as described above. I could be wrong and he actually  needed that research to overcome the spectral limitations of the fluorescent tubes on the Cruse but in that case I would think the HP scan method sketched above could be the next step to improve the color fidelity of the Cruse.

http://www.mailrepository.com/colorsync-users.lists.apple.com/msg/3863551/
http://lists.apple.com/archives/colorsync-users/2010/May/msg00071.html

--
Met vriendelijke groet, Ernst

340+ paper white spectral plots:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
update april 2012: Harman by Hahnemühle, Innova IFA45 and more

[teddillard]

Yes, Cruse put together an "upgrade" of their color management process last year, I think it was, in response to a few of their higher-profile clients' (read: major museums, some of whom I've had personal chats with about the issues) demands for more color accuracy and some specific and predictable color issues.  It's a hugely expensive additional package, and I did not have the opportunity to test it, nor have I followed up on it's performance.  I'm not sure if his work was part of that process or not, but the other thread you link actually backs up much of my opinion of the Cruse, and the performance of a tri-linear CCD compared to today's Bayer-array capture and RAW processing.  I honestly feel it's time to move past it, and that's based on what I've done with today's DSLRs.

For academic discussion, let's go back to the OP's idea of a "poor man's Cruse", or, a solution that doesn't cost eleventy million dollars.  I'm holding to my guns of the lighting, and even to the point of the wavelength of the light source.  If you light the piece with the same type (color, wavelength, and quality, too) of lights as were used in the creation of the piece, you're starting off a lot closer to getting the color reflected back to the camera, as was visible to the artist, no?  You're not introducing something that was not there at the time.   If the painting was made with 3200ºK lights, the artist would see a palette available under that source, not something, say, only visible under 5000ºK.  Though it's there, in the painting, and you're probably working more accurately, if it's going to be rejected by the artist when finally printed, why introduce the color in the first place?  

I'm having a hard time putting it into words, but simply, my experience has been that if I shoot the piece under the light it was painted, apply current RAW-level color management, print the piece using good printer profiles and a decent printer, then view the print under the source it was intended by the artist to be viewed under, my results have been remarkably predictable, and far beyond any other method my clients have experienced.

To try to put it in other words...  If you're trying to reproduce a piece that "lives" in a given "space" (let's say, 3200ºK for the sake of argument) why would you introduce elements (wavelength) that exist outside that space?  (The DSLR filtering and processing, in today's world, seems to be far better "tuned" to the same world that our eyes live in.)

In my experience, the Cruse does exactly that - introducing wavelengths I don't want, and the profiling that I've done with the system merely brings me to a point of predictable color correction.  I've seen far more problems with the pigment response issues you describe with the Cruse lights than with other light sources.  I'd love to try the Cruse with different types of lights, but it's impractical to bypass the standard Cruse system.  

need.   more...   coffee!  

For the record, as well, I'm not of the opinion that you can't do this stuff with a MFDB, or even the Betterlight.  ...just that you don't necessarily need it.  It's not the only solution anymore.  Using less expensive available DSLR technology will get you to the same place, if used properly.