Imaging Working Group

Hi IWG, 

 As part of building our Community, we will be asking a conservation imaging question every month!

So here's January's question: What can conservators do to make their multiband imaging more useful for material analysis (identify pigments, material types, etc.) that can sometimes be missed? 

I look forward to your responses!

Cheers, 

 Wendy R.

One thing that I have noticed as a new MBI user is that a lot of the literature references use kind of vague color terms to describe their imaging results?  For example, something appears "orange" under IRRFC or "dark green" under UVRFC.  Unfortunately, these words cover a lot of ground, and it makes interpretation and comparing my results to the literature very difficult.

I'm not sure what the solution here is--if there is one-- as I'm looking at dyes and there are a lot of variables in how they respond to imaging (dye concentration, mordant, substrate, etc), but I think it's something worth discussing.

Callie Jerman

The Textile Museum

Hi IWG, 

 As part of building our Community, we will be asking a conservation imaging question every month!

So here's January's question: What can conservators do to make their multiband imaging more useful for material analysis (identify pigments, material types, etc.) that can sometimes be missed? 

I look forward to your responses!

Cheers, 

 Wendy R.

Hi Callista, 

  Great point! Thanks for raising this. Is anyone aware of an attempt to standardize these response descriptors?

Cheers, 

 Wendy

One thing that I have noticed as a new MBI user is that a lot of the literature references use kind of vague color terms to describe their imaging results?  For example, something appears "orange" under IRRFC or "dark green" under UVRFC.  Unfortunately, these words cover a lot of ground, and it makes interpretation and comparing my results to the literature very difficult.

I'm not sure what the solution here is--if there is one-- as I'm looking at dyes and there are a lot of variables in how they respond to imaging (dye concentration, mordant, substrate, etc), but I think it's something worth discussing.

Callie Jerman

The Textile Museum

Hi IWG, 

 As part of building our Community, we will be asking a conservation imaging question every month!

So here's January's question: What can conservators do to make their multiband imaging more useful for material analysis (identify pigments, material types, etc.) that can sometimes be missed? 

I look forward to your responses!

Cheers, 

 Wendy R.

I'll open by saying I am not a scientist or a photographer, however, I tend to think of FC/false color images as a qualitative tool where each image is a bit of a closed system because they are very difficult to reproduce and standardize. I recently had the experience of regenerating a false color image after a work had been cleaned of its synthetic varnish (not a big color change, and the input IR image is the same). In this case, the color images were taken in the same studio, same equipment and same photographer, the overall color difference was pretty subtle, honestly, and yet I would have to do some adjusting to get the new false color to look like the original. More succinctly, what seemed like a subtle difference in the rgb images made a bigger difference in the false color. So I'd love to hear more from my more experienced colleagues here, but I find the issue with comparing false colors across artworks and systems to be in the capture and generation of the images themselves rather than in their description. You could photograph the same object with your system, but get a different result (instead of orange, it may look more red, more salmon, etc.) It's difficult to talk about this in the abstract, but there are so many variables in the captures, from the camera sensors, lights and their placement, filters used, things like varnish and dirt layers which may influence the color on either your capture or the one to which you want to compare it, even the color calibration of the screen (especially if comparing it to a printed reproduction, which may have had its own adjustments to "read better") and even when using standard protocols and spectralon targets in the images, I find it extremely challenging to get truly comparable results (where one pigment looks distinctly the same shade of orange in every image, for example). I know our wiki talks a lot about these variables and about including the metadata and conditions with false color images for reproducibility, but even with that information, if you're not working with exactly the same system, or if one of the works is varnished or dirty, can you get truly comparable results? 

Hi Callista, 

  Great point! Thanks for raising this. Is anyone aware of an attempt to standardize these response descriptors?

Cheers, 

 Wendy

One thing that I have noticed as a new MBI user is that a lot of the literature references use kind of vague color terms to describe their imaging results?  For example, something appears "orange" under IRRFC or "dark green" under UVRFC.  Unfortunately, these words cover a lot of ground, and it makes interpretation and comparing my results to the literature very difficult.

I'm not sure what the solution here is--if there is one-- as I'm looking at dyes and there are a lot of variables in how they respond to imaging (dye concentration, mordant, substrate, etc), but I think it's something worth discussing.

Callie Jerman

The Textile Museum

Hi IWG, 

 As part of building our Community, we will be asking a conservation imaging question every month!

So here's January's question: What can conservators do to make their multiband imaging more useful for material analysis (identify pigments, material types, etc.) that can sometimes be missed? 

I look forward to your responses!

Cheers, 

 Wendy R.

I'll open by saying I am not a scientist or a photographer, however, I tend to think of FC/false color images as a qualitative tool where each image is a bit of a closed system because they are very difficult to reproduce and standardize. I recently had the experience of regenerating a false color image after a work had been cleaned of its synthetic varnish (not a big color change, and the input IR image is the same). In this case, the color images were taken in the same studio, same equipment and same photographer, the overall color difference was pretty subtle, honestly, and yet I would have to do some adjusting to get the new false color to look like the original. More succinctly, what seemed like a subtle difference in the rgb images made a bigger difference in the false color. So I'd love to hear more from my more experienced colleagues here, but I find the issue with comparing false colors across artworks and systems to be in the capture and generation of the images themselves rather than in their description. You could photograph the same object with your system, but get a different result (instead of orange, it may look more red, more salmon, etc.) It's difficult to talk about this in the abstract, but there are so many variables in the captures, from the camera sensors, lights and their placement, filters used, things like varnish and dirt layers which may influence the color on either your capture or the one to which you want to compare it, even the color calibration of the screen (especially if comparing it to a printed reproduction, which may have had its own adjustments to "read better") and even when using standard protocols and spectralon targets in the images, I find it extremely challenging to get truly comparable results (where one pigment looks distinctly the same shade of orange in every image, for example). I know our wiki talks a lot about these variables and about including the metadata and conditions with false color images for reproducibility, but even with that information, if you're not working with exactly the same system, or if one of the works is varnished or dirty, can you get truly comparable results? 

Hi Callista, 

  Great point! Thanks for raising this. Is anyone aware of an attempt to standardize these response descriptors?

Cheers, 

 Wendy

One thing that I have noticed as a new MBI user is that a lot of the literature references use kind of vague color terms to describe their imaging results?  For example, something appears "orange" under IRRFC or "dark green" under UVRFC.  Unfortunately, these words cover a lot of ground, and it makes interpretation and comparing my results to the literature very difficult.

I'm not sure what the solution here is--if there is one-- as I'm looking at dyes and there are a lot of variables in how they respond to imaging (dye concentration, mordant, substrate, etc), but I think it's something worth discussing.

Callie Jerman

The Textile Museum

Hi IWG, 

 As part of building our Community, we will be asking a conservation imaging question every month!

So here's January's question: What can conservators do to make their multiband imaging more useful for material analysis (identify pigments, material types, etc.) that can sometimes be missed? 

I look forward to your responses!

Cheers, 

 Wendy R.

I'll open by saying I am not a scientist or a photographer, however, I tend to think of FC/false color images as a qualitative tool where each image is a bit of a closed system because they are very difficult to reproduce and standardize. I recently had the experience of regenerating a false color image after a work had been cleaned of its synthetic varnish (not a big color change, and the input IR image is the same). In this case, the color images were taken in the same studio, same equipment and same photographer, the overall color difference was pretty subtle, honestly, and yet I would have to do some adjusting to get the new false color to look like the original. More succinctly, what seemed like a subtle difference in the rgb images made a bigger difference in the false color. So I'd love to hear more from my more experienced colleagues here, but I find the issue with comparing false colors across artworks and systems to be in the capture and generation of the images themselves rather than in their description. You could photograph the same object with your system, but get a different result (instead of orange, it may look more red, more salmon, etc.) It's difficult to talk about this in the abstract, but there are so many variables in the captures, from the camera sensors, lights and their placement, filters used, things like varnish and dirt layers which may influence the color on either your capture or the one to which you want to compare it, even the color calibration of the screen (especially if comparing it to a printed reproduction, which may have had its own adjustments to "read better") and even when using standard protocols and spectralon targets in the images, I find it extremely challenging to get truly comparable results (where one pigment looks distinctly the same shade of orange in every image, for example). I know our wiki talks a lot about these variables and about including the metadata and conditions with false color images for reproducibility, but even with that information, if you're not working with exactly the same system, or if one of the works is varnished or dirty, can you get truly comparable results? 

Hi Callista, 

  Great point! Thanks for raising this. Is anyone aware of an attempt to standardize these response descriptors?

Cheers, 

 Wendy

One thing that I have noticed as a new MBI user is that a lot of the literature references use kind of vague color terms to describe their imaging results?  For example, something appears "orange" under IRRFC or "dark green" under UVRFC.  Unfortunately, these words cover a lot of ground, and it makes interpretation and comparing my results to the literature very difficult.

I'm not sure what the solution here is--if there is one-- as I'm looking at dyes and there are a lot of variables in how they respond to imaging (dye concentration, mordant, substrate, etc), but I think it's something worth discussing.

Callie Jerman

The Textile Museum

Hi IWG, 

 As part of building our Community, we will be asking a conservation imaging question every month!

So here's January's question: What can conservators do to make their multiband imaging more useful for material analysis (identify pigments, material types, etc.) that can sometimes be missed? 

I look forward to your responses!

Cheers, 

 Wendy R.

Hi Kelly,

I'm relatively new to this group, and not a conservator by any means.  I work in the area of Imaging Science.  But our group is heavily involved in multispectral and hyperspectral imaging for lots of cultural heritage applications, including art conservation.  Most of our work has historically been on manuscripts and maps, but we are starting to work more with the art conservation community and I am looking forward to learning more about your processes and challenges.

You're correct, false color images are good for qualitative analysis, but in most cases are not useful for quantitative analyses.  Unless your workflow involves calibration of the images to physical units like reflectance (not just color balancing with a color checker), I think what you're experiencing is likely due to (relatively) uncalibrated images.  You mention the use of spectralon in the images, but how are you doing the calibration?  As you say, for every image collected there are multiple factors at play including the illumination conditions, the camera capture parameters, the specifics of the filters used, etc.  All of these have impacts on what digital counts are actually recorded by the detector.  Radiometric calibration to physical units can, to an extent, remove all of these effects and allow for more direct comparison across artifacts, cameras, materials, etc.

Your comment about a small difference in the RGB making a big difference in the False Color depends on the material and the special response function of the camera.  There are many materials that "look" similar in the visible portion of the spectrum but have reflectance differences in the IR or other regions.  Is that what you are referring to?

I would be interested in learning what calibration procedures you are using to help make comparison across cameras, artifacts, materials, etc. more robust.  As I said, I'm new here so might not be understanding your question correctly, and if so I apologize.   

I'll open by saying I am not a scientist or a photographer, however, I tend to think of FC/false color images as a qualitative tool where each image is a bit of a closed system because they are very difficult to reproduce and standardize. I recently had the experience of regenerating a false color image after a work had been cleaned of its synthetic varnish (not a big color change, and the input IR image is the same). In this case, the color images were taken in the same studio, same equipment and same photographer, the overall color difference was pretty subtle, honestly, and yet I would have to do some adjusting to get the new false color to look like the original. More succinctly, what seemed like a subtle difference in the rgb images made a bigger difference in the false color. So I'd love to hear more from my more experienced colleagues here, but I find the issue with comparing false colors across artworks and systems to be in the capture and generation of the images themselves rather than in their description. You could photograph the same object with your system, but get a different result (instead of orange, it may look more red, more salmon, etc.) It's difficult to talk about this in the abstract, but there are so many variables in the captures, from the camera sensors, lights and their placement, filters used, things like varnish and dirt layers which may influence the color on either your capture or the one to which you want to compare it, even the color calibration of the screen (especially if comparing it to a printed reproduction, which may have had its own adjustments to "read better") and even when using standard protocols and spectralon targets in the images, I find it extremely challenging to get truly comparable results (where one pigment looks distinctly the same shade of orange in every image, for example). I know our wiki talks a lot about these variables and about including the metadata and conditions with false color images for reproducibility, but even with that information, if you're not working with exactly the same system, or if one of the works is varnished or dirty, can you get truly comparable results? 

Hi Callista, 

  Great point! Thanks for raising this. Is anyone aware of an attempt to standardize these response descriptors?

Cheers, 

 Wendy

One thing that I have noticed as a new MBI user is that a lot of the literature references use kind of vague color terms to describe their imaging results?  For example, something appears "orange" under IRRFC or "dark green" under UVRFC.  Unfortunately, these words cover a lot of ground, and it makes interpretation and comparing my results to the literature very difficult.

I'm not sure what the solution here is--if there is one-- as I'm looking at dyes and there are a lot of variables in how they respond to imaging (dye concentration, mordant, substrate, etc), but I think it's something worth discussing.

Callie Jerman

The Textile Museum

Hi IWG, 

 As part of building our Community, we will be asking a conservation imaging question every month!

So here's January's question: What can conservators do to make their multiband imaging more useful for material analysis (identify pigments, material types, etc.) that can sometimes be missed? 

I look forward to your responses!

Cheers, 

 Wendy R.

Thanks for your insights, David. 

I didn't really have a question, per se, more a comment that I find trying to compare one's own false color results against something published is nearly impossible given the number of variables (and asterisks). Even if your own system is well calibrated, you generally do not have enough information or enough specifics simply from what's published to know how the other system was calibrated and whether your results can be comparable. This was all in response to the original question about whether we could standardize our language around the colors one sees in a false color so that comparisons are possible. My comment was just to say that the root issue is deeper than that, and often even with calibrating across the board it's nearly impossible to have truly comparable results (where the same pigment looks exactly the same in false color images for a number of objects). To answer your question, for FC with MBI, I've only done a few examples, but have used the levels outlined in Aldrovandi, et. al. from 2005 in relation to the spectralons. I should also say that while I take some images, I am not the museum's primary photographer, so many of the images we use for FC will come from our Imaging department who have their own calibration and processing standards. But the crux of my point was this: whenever one of the source images for a FC is the rgb, it makes sense that if the rgb changes after a treatment due to cleaning (which is itself indicative of color change on the actual physical object), the after treatment false color will also be different from the before treatment. Yet another variable making comparisons difficult. Perhaps, as Callista suggested, it is different for other media where there are fewer variables, but for paintings, I find FC a very useful tool, but extremely difficult to use in a comparative way. Radiometric calibration sounds like it might be helpful in this regard, and I don't think our Imaging Department does that level of calibration (I know my workflow does not include it), but without similar calibration in the images you wish to compare (in the original query, published images or published written accounts of those images), you run into the same issue. Thanks again, this discussion has given me a lot to think about!

Hi Kelly,

I'm relatively new to this group, and not a conservator by any means.  I work in the area of Imaging Science.  But our group is heavily involved in multispectral and hyperspectral imaging for lots of cultural heritage applications, including art conservation.  Most of our work has historically been on manuscripts and maps, but we are starting to work more with the art conservation community and I am looking forward to learning more about your processes and challenges.

You're correct, false color images are good for qualitative analysis, but in most cases are not useful for quantitative analyses.  Unless your workflow involves calibration of the images to physical units like reflectance (not just color balancing with a color checker), I think what you're experiencing is likely due to (relatively) uncalibrated images.  You mention the use of spectralon in the images, but how are you doing the calibration?  As you say, for every image collected there are multiple factors at play including the illumination conditions, the camera capture parameters, the specifics of the filters used, etc.  All of these have impacts on what digital counts are actually recorded by the detector.  Radiometric calibration to physical units can, to an extent, remove all of these effects and allow for more direct comparison across artifacts, cameras, materials, etc.

Your comment about a small difference in the RGB making a big difference in the False Color depends on the material and the special response function of the camera.  There are many materials that "look" similar in the visible portion of the spectrum but have reflectance differences in the IR or other regions.  Is that what you are referring to?

I would be interested in learning what calibration procedures you are using to help make comparison across cameras, artifacts, materials, etc. more robust.  As I said, I'm new here so might not be understanding your question correctly, and if so I apologize.   

I'll open by saying I am not a scientist or a photographer, however, I tend to think of FC/false color images as a qualitative tool where each image is a bit of a closed system because they are very difficult to reproduce and standardize. I recently had the experience of regenerating a false color image after a work had been cleaned of its synthetic varnish (not a big color change, and the input IR image is the same). In this case, the color images were taken in the same studio, same equipment and same photographer, the overall color difference was pretty subtle, honestly, and yet I would have to do some adjusting to get the new false color to look like the original. More succinctly, what seemed like a subtle difference in the rgb images made a bigger difference in the false color. So I'd love to hear more from my more experienced colleagues here, but I find the issue with comparing false colors across artworks and systems to be in the capture and generation of the images themselves rather than in their description. You could photograph the same object with your system, but get a different result (instead of orange, it may look more red, more salmon, etc.) It's difficult to talk about this in the abstract, but there are so many variables in the captures, from the camera sensors, lights and their placement, filters used, things like varnish and dirt layers which may influence the color on either your capture or the one to which you want to compare it, even the color calibration of the screen (especially if comparing it to a printed reproduction, which may have had its own adjustments to "read better") and even when using standard protocols and spectralon targets in the images, I find it extremely challenging to get truly comparable results (where one pigment looks distinctly the same shade of orange in every image, for example). I know our wiki talks a lot about these variables and about including the metadata and conditions with false color images for reproducibility, but even with that information, if you're not working with exactly the same system, or if one of the works is varnished or dirty, can you get truly comparable results? 

Hi Callista, 

  Great point! Thanks for raising this. Is anyone aware of an attempt to standardize these response descriptors?

Cheers, 

 Wendy

One thing that I have noticed as a new MBI user is that a lot of the literature references use kind of vague color terms to describe their imaging results?  For example, something appears "orange" under IRRFC or "dark green" under UVRFC.  Unfortunately, these words cover a lot of ground, and it makes interpretation and comparing my results to the literature very difficult.

I'm not sure what the solution here is--if there is one-- as I'm looking at dyes and there are a lot of variables in how they respond to imaging (dye concentration, mordant, substrate, etc), but I think it's something worth discussing.

Callie Jerman

The Textile Museum

Hi IWG, 

 As part of building our Community, we will be asking a conservation imaging question every month!

So here's January's question: What can conservators do to make their multiband imaging more useful for material analysis (identify pigments, material types, etc.) that can sometimes be missed? 

I look forward to your responses!

Cheers, 

 Wendy R.

Hi Callista, 

  Great point! Thanks for raising this. Is anyone aware of an attempt to standardize these response descriptors?

Cheers, 

 Wendy

One thing that I have noticed as a new MBI user is that a lot of the literature references use kind of vague color terms to describe their imaging results?  For example, something appears "orange" under IRRFC or "dark green" under UVRFC.  Unfortunately, these words cover a lot of ground, and it makes interpretation and comparing my results to the literature very difficult.

I'm not sure what the solution here is--if there is one-- as I'm looking at dyes and there are a lot of variables in how they respond to imaging (dye concentration, mordant, substrate, etc), but I think it's something worth discussing.

Callie Jerman

The Textile Museum

Hi IWG, 

 As part of building our Community, we will be asking a conservation imaging question every month!

So here's January's question: What can conservators do to make their multiband imaging more useful for material analysis (identify pigments, material types, etc.) that can sometimes be missed? 

I look forward to your responses!

Cheers, 

 Wendy R.