Photographing degraded 11/12C frescos

[OldRoy]

Some time ago I photographed some church frescos dating from the late 11th to the early 12th century for an academic friend who has a specialised interest in them. They are severely degraded, having been painted originally in red/yellow ochre, lime white and carbon black which has both faded and been damaged by later repairs.

By a bit of judicious tweaking of saturation, levels etc, I was able to reveal quite a lot more of the original forms. To the non-photographer recipient of these modified images they looked quite magically revealing; it's certainly possible to discern more of the detail although of course much of the damage is equally "enhanced".

I've been asked to do some more similar work - there are four locations deriving from the same group of artists. I was wondering if there's a viable way of deriving even more detail - "viable" in this case meaning inexpensive as there's no funding available. Would illumination by wavelengths outside the visible spectrum, eg IR, be likely to reveal much? I'd considered converting an unused D200 to IR as it's not doing anything else and this would give me an excuse. Alternatively given that the subject's static maybe I could get a cheaper result by using an IR-pass filter?

Any suggestions would be appreciated. I can post a sample image if it would help (not currently available from this PC.)
Roy

[t6b9p]

Convert the D200 to wide spectrum UV through IR. That gives you the option to vary the capture wavelengths in the IR as well as shooting reflected UV. In addition, consider fluorescence photography. You will of course need some kind of budget for filters. Some lenses will transmit sufficient UV for reflected UV work. Some flashes are relatively easily modified for UV work.

PM me if you have any questions.

[Doug Peterson]

Definitely try the IR route.

Infrared Cultural Heritage Imaging can be very effective for some types of images.

[OldRoy]

Definitely try the IR route.

Infrared Cultural Heritage Imaging can be very effective for some types of images.

Thanks Doug.
I'm not surprised to find that there are dedicated solutions available however something tells me that this hardware isn't inexpensive! At present the project I've been asked to participate in has no budget whatsoever other than what I'm prepared to spend. Even converting the D200 (about £250 here) is something I'd have to think hard about.
With a bit of care it would seem feasible to accomplish a multiple exposure (visible, UV, IR) image and manually combine them. I hope...
Roy

[JayWPage]

What are your lighting options?

[OldRoy]

What are your lighting options?

At present completely open. I have a couple of flash units - 1 Nikon and 1 old Vivitar. The constraints are more about lack of budget unless someone volunteers to donate one (there may be a small academic source but I'm not optimistic). Any suggestion goes into the pot for consideration. Right now I'm just waiting for more information about what's required. I originally supplied pix gratis (it was not out of my way to shoot them) and now I've been asked if I would consider doing something more. Having said "yes" and made a couple of suggestions I'm awaiting a response. There's a possibility of a small commercial spinoff for myself.
Roy

[joneil]

 I had one of my DX, DSLRs convered to pure IF .  780 to 850 nm.    Sometimes on old paintings you see nothing new, and sometimes you go "wow."

 for your kind of work, I would suggest, if you are going that route, a clear filter conversion over the sensor.   That is, full spectrum from UV to deep IR.   Then you photograph the exact same fresco or painting, etc, using several differnet filters.  One filter at a time of course.

 Good luck, sounds like the kind of thing I would love to be doing.


   

[OldRoy]

There's a company called Protech quite near me. I've used them previously for other work. They offer a "full spectrum" conversion (£250 for crop sensors) which they advertise as for astro photography. My assumption is that I'll have to buy a couple of filters to differentiate between the two ends of the spectrum. What results can be expected if one shoots without anything on the lens?
Roy

[Jim Kasson]

I've been asked to do some more similar work - there are four locations deriving from the same group of artists. I was wondering if there's a viable way of deriving even more detail - "viable" in this case meaning inexpensive as there's no funding available. Would illumination by wavelengths outside the visible spectrum, eg IR, be likely to reveal much? I'd considered converting an unused D200 to IR as it's not doing anything else and this would give me an excuse. Alternatively given that the subject's static maybe I could get a cheaper result by using an IR-pass filter?

Any suggestions would be appreciated. I can post a sample image if it would help (not currently available from this PC.)
Roy

I'd suggest a "full-spectrum" conversion, which will allow you to change filtration by adding filters in front of the lens. If you use a mirrorless camera, you'll be able to see the effects in the finder. Unfortunately, the camera's color filter array will modify the spectral characteristics of the sensor and the filter. You could try UV-only filters, IR-only filters with various cutoff wavelengths (the R72 probably being a good place to start).

The high road is a monochromatic sensor, and all color separation done with filters in front of the lens and combined in post, but that sounds like it's beyond your budget. Maybe you can get a university interested in your work?

Jim

[Jim Kasson]

There's a company called Protech quite near me. I've used them previously for other work. They offer a "full spectrum" conversion (£250 for crop sensors) which they advertise as for astro photography. My assumption is that I'll have to buy a couple of filters to differentiate between the two ends of the spectrum. What results can be expected if one shoots without anything on the lens?

Then you'll get the filtration of the CFA plus the lens/sensor. A bigger problem is loss of sharpness, since few lenses are designed to bring IR, visible, and UV light to a focus at the same point. One lens that is, the Coastal Optical 60mm f/4 UV-VIS-IR Apo Macro, is very expensive. You'll be better off combining pictures made with various filtration in register in post. You'll need live view to focus the non-visible light.

Jim

[sniper]

Thanks Doug.
I'm not surprised to find that there are dedicated solutions available however something tells me that this hardware isn't inexpensive! At present the project I've been asked to participate in has no budget whatsoever other than what I'm prepared to spend. Even converting the D200 (about £250 here) is something I'd have to think hard about.
With a bit of care it would seem feasible to accomplish a multiple exposure (visible, UV, IR) image and manually combine them. I hope...
Roy

If I remember rightly the old D70 was pretty senstive to IR without modification, certainly the one I used was.  you could pick up one of those for a lot less than £250.

[Bart_van_der_Wolf]

Some time ago I photographed some church frescos dating from the late 11th to the early 12th century for an academic friend who has a specialised interest in them. They are severely degraded, having been painted originally in red/yellow ochre, lime white and carbon black which has both faded and been damaged by later repairs.

By a bit of judicious tweaking of saturation, levels etc, I was able to reveal quite a lot more of the original forms. To the non-photographer recipient of these modified images they looked quite magically revealing; it's certainly possible to discern more of the detail although of course much of the damage is equally "enhanced".

Hi Roy,

I'm wondering if you really need any of the UV/IR suggestions to improve the image. Sure, it may be possible to improve the visibility of things that e.g. are relatively transparent (or opaque) to certain wavelengths, but only you can determine that. On a low budget, your options are going to be limited. It's also not necessarily going to help with restoring/reviving colors, if that's your goal.

What you could do, is use your unmodified camera with an IR transmission filter and use a long exposure. That will already tell you if anything becomes better visible. If not, you can spare the expense of having that camera type modified. UV light will probably be limited, unless one uses specialized lenses that have higher transmission for UV than regular glass. Besides, silicon based sensors are not very sensitive to UV anyway, so you'd have more luck if there is a form of fluorescence with visible wavelengths, when excited by UV light.

I can post a sample image if it would help (not currently available from this PC.)

It's not clear to me if (revived) color reproduction is the goal, or making things visible that the naked eye cannot easily discern. Maybe a before/after example is indeed helpful.

If the goal is to restore color, then you may be able to get results faster by using a good color conversion process. You could look into techniques that use 3D Color LUT grading with Hald lookup tables. RawTherapee supports them, and RawTherapee is free to use (which would agree with your budget). One typically treats the base Hald table the same as the improved image, and then reapplies that modified look-up table on all subsequent images to get consistent results. The LUT conversion will use interpolation, so you can also use lower or higher resolution LUTs that are hand tweaked for very specific colors (although consistent exposure levels become increasingly important if you get too specific).

Cheers,
Bart

P.S. Here is some more background on the use of HALD CLUTs in RawTherapee.

[OldRoy]

Thanks for all the suggestions so far.
At present I'm still waiting for some decision about what/when/where/how much... We're dealing with the timescales of academia here. However I thought that I might, as suggested, try a few long exposures with a D700 and an IR filter on the lens. To that end would someone like to suggest the best wavelength option as there seem to be several available; as usual there are some low-cost examples on Amazon which might suffice for an experiment.

As to the objectives, it would be a matter of revealing additional detail in the frescos. Tweaking in LR/PS has already helped quite a bit. The colour is almost irrelevant since even when these were painted there were limited pigments available. I've seen some successful examples showing fossils.

I'm curious about the focusing issue. If I use the D700 (or, possibly, my EM5) I have live view available. How big is the focus shift at each end of the spectrum in comparison with the visible? In other words, given static subjects and no restriction on exposure times, would there be sufficient DOF if stopped down to (say) F8 on a normal or wide lens?
Roy
EDIT. Unsurprisingly, subsequent to posting the above I find that Wikipedia has very detailed information on these issues.

[Jim Kasson]

I'm curious about the focusing issue. If I use the D700 (or, possibly, my EM5) I have live view available. How big is the focus shift at each end of the spectrum in comparison with the visible? In other words, given static subjects and no restriction on exposure times, would there be sufficient DOF if stopped down to (say) F8 on a normal or wide lens?

If you're using live view and narrow filters, you don't have to worry about focus shift, because you're looking at the camera's response to light falling on the sensor.

With wide filters, or combining visible abnd UV or IR light, you now have to consider depth of focus (as opposed to depth of field, assuming you're photographing something flat). LV will give you a compromise. MTF testing will tell you how far down you have to stop.

http://blog.kasson.com/?p=8543

Jim

[t6b9p]

However I thought that I might, as suggested, try a few long exposures with a D700 and an IR filter on the lens.

The D700 has an internal IR LED that monitors the shutter. This can cause problems with long exposures, high ISO and particularly when shooting IR.

[kirkt]

Without a sample, it is difficult to imagine the level of detail that your current technique is capturing.  Do you supplement the lighting in the location at all, or shoot a relatively longish exposure on a tripod?  Does this approach give you uniform illumination across the field of view?

Because the subject you are shooting is static, you may also want to consider using image calibration and stacking/integration techniques (the kind used in astrophotography) to increase signal-to-noise in your image data.  You may also be able to employ drizzle integration to your data to squeeze out the most detail during conversion and integration of the raw images (you are shooting raw, right?).

These techniques will also work well with non-visible spectral data - again, similar to astrophotography.  Consider an expanded image processing workflow as you investigate image acquisition and hardware enhancements, so that you know what you need to acquire and how to acquire it in terms of post-processing.

Consider also that you are shooting a relatively planar object - your choice of lens is important interns of field curvature and maintaining good focus and contrast across the field of view.  You can also employ panoramic/mosaic techniques to capture significantly more pixel information as you image the subject.  

More money to spend.

Good luck, sounds like a cool project!

kirk

[OldRoy]

Thank you to all who have contributed advice so far. This has been very slow to develop but I've started some of the associated photography. I bought a pair of (Chinese) IR pass filters by way of inexpensive experiment @ 720 and 650 Nm. So far I'm at a bit of a loss as to even how to experiment with these. I take the points about lens characteristics and  I have a limited selection of manual primes available - Nikon AIS 20mm F3.5, F50mm 1.8, 200mm F4 - plus 50 1.8G and 85 1.8D (excluding FEs); ie nothing special.

I've tried the 720 on both the D700 and D200 using the 20 and 50mm lenses and just pointing a camera out of my window at the local Yacht club. I was under the impression that this filter pretty much excluded all visible light - but it sure doesn't look like it. It does however give everything, unsurprisingly, a deep red cast. With a couple of stops of +ve EC the D200 looks a bit more like what I'd expected - a bit - but excruciatingly noisy even at ISO 200. I've yet to try my EM5.

Could someone suggest a methodology for getting to grips with this technique? I'm prepared to negotiate the problems of field curvature, depth of focus etc once I've got some idea of the general attributes of IR photography. I'd possibly consider getting the D200 converted at some point as there's a company quite near me (Protech: http://www.protechrepairs.co.uk/) who I've spoken to and sound extremely helpful. But right now I'm not even on sq. 1.
Roy

[langier]

I'm in the same boat when I'm in Serbia and the Balkans shooting some pretty ancient frescoes, some quite marginal.

I'm now using a pair of D800 bodies and have changed my lighting technique from flash to using a small high-output LED spot, a Lenser AAA powered light. The modern icons, I'll light but the old ones need to be done carefully. This allows me to gently light the icon and paint larger panels as needed. It's mainly used to highlight and emphasize a particular writing.

When I can, I'm using a tripod, etc. but sometimes, I've got to wing it and crank up the ISO.

Getting a good exposure is key when you can. I am not using IR since I'm trying to document and slightly enhance the original work.

However, the magic happens in post processing. Don't be afraid to push the sliders in ACR. Sometimes that will pull out some of the hidden detail. Next is to do a b&w conversion using Nik Silver Efex. It seems weird to do this to enhance the color, but stay with me. Sometimes, you need to change the filtration at this point. Nik gives you a box of Wratten filters that allows you to simply click the color and push the slider. This may get you in the IR range without having a converted body.

My goal here is to get a b&w image that converts well and has some contrast and "pop". Mostly, a canned preset is all it takes me here. When it looks good, let Nik do its thing and you'll have a b&w layer over your color image. Next change the blending mode to "Luminance". The contrast and detail should really pop. However, sometimes the image seems a bit flat. To counteract this, I add another Vibrance layer with both the vibrance and saturation pumps up +20 to 40 on both sliders.

I've used this technique to push certain images that I could not easily go back to rephotographed and comparing the processed with the raw image is really something to see. The original raw captures simply seem lifeless and flat. Once processed, the detail comes out and is amazing. It may not work for all your frescoes, but it may be what may work.

I don't have any examples on this computer but I do have them in my studio. I'd be happy to show you off-list if this may help.

[Ajoy Roy]

Most of the lenses will not transmit UV, you need special lenses for that.

For IR, just set up the camera on a tripod, and shoot once without filters and once with each filter you have. You can also try out filters of various colours (used in B&W era). Each colour will give you a unique B&W image, which can later be enhanced to reveal what is not discernible in a normal colour image.

You can also experiment with various light sources - While LED, Flash and Fluorescent. Each combination of filter and light will give you different perspective.