measuring color

Hi,
In order to define colors for radiance models I'm looking for an instrument to measure the colors of walls and of mural paintings of archaeological sites. It should be portable, working with batteries and non destructive to the material. (max. price, $5000 US)
Any recommendation?

genevieve lucet

Hi Genevieve,

There are a couple of options that I can think of off hand:

   1. Color Checker and Digital Camera - using the techniques that are
      outlined in Rendering with Radiance, you can use the Color Checker
      Chart from Gretag Macbeth and a digital camera to sample color and
      reflectance. The color checker is an accurate set of color samples
      with known measured reflectances and chromaticity coordinates. It
      can be used to calibrate color/reflectance information that is
      acquired through a device such as a scanner, digital camera, or
      even a regular old film camera. I think used correctly the results
      can be within +or- 10%. We have used this method with reasonable
      success on a variety of materials. This is a very economically
      method (~$100 for color checker plus cost of sampling device)
   2. Handheld Spectrophotometer - there are a variety of options out
      there for hand held spectrophotometers. I believe that Gretag
      Macbeth has several. These will likely be more accurate than the
      above and cost a lot more. I think when I last looked into this (5
      or more years ago) such a device cost ~$5,000 and did not include
      the software necessary for operation. I have not had the
      opportunity to use one yet.

Regards,

-Jack de Valpine

genevieve lucet wrote:

···

Hi,
In order to define colors for radiance models I'm looking for an instrument to measure the colors of walls and of mural paintings of archaeological sites. It should be portable, working with batteries and non destructive to the material. (max. price, $5000 US)
Any recommendation?

genevieve lucet

_______________________________________________
Radiance-general mailing list
Radiance-general@radiance-online.org
http://www.radiance-online.org/mailman/listinfo/radiance-general

--
# John E. de Valpine
# president
#
# visarc incorporated
# http://www.visarc.com
#
# channeling technology for superior design and construction

Here are some colorimeters that may be usefull for you and, hopefully, OK with your budget:
http://www.konicaminoltaeurope.com/products/industrial/colorimeter.shtml

···

At 12.05.2004 08:19, you wrote:

Hi,
In order to define colors for radiance models I'm looking for an instrument to measure the colors of walls and of mural paintings of archaeological sites. It should be portable, working with batteries and non destructive to the material. (max. price, $5000 US)
Any recommendation?

genevieve lucet

_______________________________________________
Radiance-general mailing list
Radiance-general@radiance-online.org
http://www.radiance-online.org/mailman/listinfo/radiance-general

Hi Genevieve,

I definitely would recommend a more sophisticated method than using the macbethcolor checker card. Yes, it is true that the chromaticities are known for those colours, BUT this is under the assumption of a certain illumination source. Will you have that same illumination source? Probably not.

If you are after accuracy (rather than 'good looks')you might want to buy a spectroradiometer and a white reflectance standard, too. After all, what you measure with the spectro looking onto your wall is the colour signal, a combination of the surface reflectance and the illumination at hand (and some geometry). In order to model the material properties in RADIANCE, you will be interested in the surface reflectance and not in the overall colour signal. To retrieve the surface reflectance, you measure the spectrum of the white reflectance standard at the very same position where you measured the colour signal from the wall, then divide the colour signal (spectrum) by the spectrum from the white reflectance standard and you have your surface reflectance (this assumes a Lambertian surface). The spectrum from the white reflectance standard can be used to get the colour of the illumination source.
This approach will yield something with a clear tint, if the illumination source wasn't a natural light. If you want to have a good-looks picture then don't bother with the white reflectance standard
and model the light source as white (as described and recommended in the RADIANCE book).

Regards,
alexa

···

Hi,
In order to define colors for radiance models I'm looking for an instrument to measure the colors of walls and of mural paintings of archaeological sites. It should be portable, working with batteries and non destructive to the material. (max. price, $5000 US)
Any recommendation?

genevieve lucet

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dr. Alexa I. Ruppertsberg
Department of Optometry
University of Bradford
Bradford
BD7 1DP
UK
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Hi Alexa,

Just a minor comment here. I have no doubt that the using a more sophisticated measuring device will yield more accurate results. However, I believe the method of using the machbethcolor checker card as described in Rendering with Radiance is designed to calibrate to the current lighting conditions, that is, assuming the use of a digital camera, if we want to sample a particular material in a give lighting condition, then we also sample the color checker chart under the same conditions with the same sampling device (digital camera) and then perform the calibration with macbethcal and pcomb. I do not know the underlying method that Greg put together to do the calibration, so I am not sure how he has accounted for varying light conditions vis a vis the internal dataset that Radiance uses as part of the calibration mechanism.

-Jack de Valpine

Alexa I. Ruppertsberg wrote:

···

Hi Genevieve,

I definitely would recommend a more sophisticated method than using the macbethcolor checker card. Yes, it is true that the chromaticities are known for those colours, BUT this is under the assumption of a certain illumination source. Will you have that same illumination source? Probably not.

If you are after accuracy (rather than 'good looks')you might want to buy a spectroradiometer and a white reflectance standard, too. After all, what you measure with the spectro looking onto your wall is the colour signal, a combination of the surface reflectance and the illumination at hand (and some geometry). In order to model the material properties in RADIANCE, you will be interested in the surface reflectance and not in the overall colour signal. To retrieve the surface reflectance, you measure the spectrum of the white reflectance standard at the very same position where you measured the colour signal from the wall, then divide the colour signal (spectrum) by the spectrum from the white reflectance standard and you have your surface reflectance (this assumes a Lambertian surface). The spectrum from the white reflectance standard can be used to get the colour of the illumination source.
This approach will yield something with a clear tint, if the illumination source wasn't a natural light. If you want to have a good-looks picture then don't bother with the white reflectance standard
and model the light source as white (as described and recommended in the RADIANCE book).

Regards,
alexa

Hi,
In order to define colors for radiance models I'm looking for an instrument to measure the colors of walls and of mural paintings of archaeological sites. It should be portable, working with batteries and non destructive to the material. (max. price, $5000 US)
Any recommendation?

genevieve lucet

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dr. Alexa I. Ruppertsberg
Department of Optometry
University of Bradford
Bradford
BD7 1DP
UK
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

_______________________________________________
Radiance-general mailing list
Radiance-general@radiance-online.org
http://www.radiance-online.org/mailman/listinfo/radiance-general

--
# John E. de Valpine
# president
#
# visarc incorporated
# http://www.visarc.com
#
# channeling technology for superior design and construction

Hi Jack,

I have not (yet) embarked on the use of a digital camera for calibration purposes, but I imagine that you also have to figure out the sensitivity of your sensors in the camera, as they will determine your 'signal', i.e. picture, let alone that exposure etc. will also affect it. Actually, it sounds quite complicated in contrast to using a spectro.

I agree that as long as you measure a macbethcolour checker card and your data sample under the same illumination, you could go ahead.The routine described in the book relies heavily on macbethcal (about neither you nor me have an idea how it works) and also points out its limitation (the word 'approximation' is used). Also, nobody could explain to me what magic things 'pcond' does, so I rather do not use it (the word 'mimick' appears rather often in the man pages). Human visual perception cannot be accounted for by a magic RADIANCE function. If it would, then why are several hundred scientists world-wide still studying it? I agree, that the results of RADIANCE are visually appealing and that the physics are simulated quite well, but only if you input accurate data your results will be accurate.

Cheers,
Alexa

Jack de Valpine wrote:

···

Hi Alexa,

Just a minor comment here. I have no doubt that the using a more sophisticated measuring device will yield more accurate results. However, I believe the method of using the machbethcolor checker card as described in Rendering with Radiance is designed to calibrate to the current lighting conditions, that is, assuming the use of a digital camera, if we want to sample a particular material in a give lighting condition, then we also sample the color checker chart under the same conditions with the same sampling device (digital camera) and then perform the calibration with macbethcal and pcomb. I do not know the underlying method that Greg put together to do the calibration, so I am not sure how he has accounted for varying light conditions vis a vis the internal dataset that Radiance uses as part of the calibration mechanism.

-Jack de Valpine

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dr. Alexa I. Ruppertsberg
Department of Optometry
University of Bradford
Bradford
BD7 1DP
UK
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Hi Alexa,

I think that we are agreed that using a spectrophotometer is a more accurate solution depending on the users needs, objectives etc... ;->

One clarification though. As far as I know "pcond" (the radiance program for physically based simulation of a variety of dynamic range and contrast sensitivity functions as based on research by Jim Ferwerda, Holly Rushmeier, Greg Ward and others) is not used with the macbethcal callibration process. The program I was refering to is "pcomb" which can be used for a variety of image process operations on radiance image including applying a calibration function as output from "macbethcal".

-Jack

Alexa I. Ruppertsberg wrote:

···

Hi Jack,

I have not (yet) embarked on the use of a digital camera for calibration purposes, but I imagine that you also have to figure out the sensitivity of your sensors in the camera, as they will determine your 'signal', i.e. picture, let alone that exposure etc. will also affect it. Actually, it sounds quite complicated in contrast to using a spectro.

I agree that as long as you measure a macbethcolour checker card and your data sample under the same illumination, you could go ahead.The routine described in the book relies heavily on macbethcal (about neither you nor me have an idea how it works) and also points out its limitation (the word 'approximation' is used). Also, nobody could explain to me what magic things 'pcond' does, so I rather do not use it (the word 'mimick' appears rather often in the man pages). Human visual perception cannot be accounted for by a magic RADIANCE function. If it would, then why are several hundred scientists world-wide still studying it? I agree, that the results of RADIANCE are visually appealing and that the physics are simulated quite well, but only if you input accurate data your results will be accurate.

Cheers,
Alexa

Jack de Valpine wrote:

Hi Alexa,

Just a minor comment here. I have no doubt that the using a more sophisticated measuring device will yield more accurate results. However, I believe the method of using the machbethcolor checker card as described in Rendering with Radiance is designed to calibrate to the current lighting conditions, that is, assuming the use of a digital camera, if we want to sample a particular material in a give lighting condition, then we also sample the color checker chart under the same conditions with the same sampling device (digital camera) and then perform the calibration with macbethcal and pcomb. I do not know the underlying method that Greg put together to do the calibration, so I am not sure how he has accounted for varying light conditions vis a vis the internal dataset that Radiance uses as part of the calibration mechanism.

-Jack de Valpine

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dr. Alexa I. Ruppertsberg
Department of Optometry
University of Bradford
Bradford
BD7 1DP
UK
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

_______________________________________________
Radiance-general mailing list
Radiance-general@radiance-online.org
http://www.radiance-online.org/mailman/listinfo/radiance-general

--
# John E. de Valpine
# president
#
# visarc incorporated
# http://www.visarc.com
#
# channeling technology for superior design and construction

This is an interesting thread. I just wanted to add a couple of notes.

First, you should check out the products available from X-rite for spectral measurement:

  http://www.xrite.com/product_overview.aspx?Line=24

I have used their spectrophotometers in the past, to good effect. I think they're prices are competitive, but not what I'd call "cheap."

Second, a quick explanation of what macbethcal does. Macbethcal takes a captured image (scanned, photographed, etc.) of a Macbeth ColorChecker chart and attempts to derive a transformation that takes this capture to the correct values known for the Macbeth chart. It first computes an overall response function for each channel using the 6 gray patches at the bottom of the chart. Then, it computes a 3x3 color matrix based on a least squares minimization of error on the unsaturated colors. Saturated colors are allowed to go "out of gamut" as needed to find a reasonable solution.

In my experience, this process works well for flatbed scanners and some capture methods, but sometimes does poorly with certain digital cameras. I do not know why. As Alexa and Jack pointed out, macbethcal does not account for changes in the illumination source -- this is assumed to be part of the capture process (i.e., a scanner with a builtin source), and since the purpose is to derive a transform that recovers colors under this souce, this is appropriate.

If you wish to simulate general reflectance spectra under different illuminants, I suggest you go with a full spectral measurement and follow the procedure outlined in the following paper:

  http://www.anyhere.com/gward/papers/egwr02/

Third, I wanted to point Alexa and other interested readers to the following reference, which explains the parts of human vision pcond attempts to simulate and the algorithms behind it:

  http://radsite.lbl.gov/radiance/papers/lbnl39882/tonemap.pdf

I'll be the first to admit that it is not a perfect simulation of human vision, and this is an ongoing area of research. (Try a websearch on "color appearance models" to get an idea.) However, I think others will agree that pcond can be useful, and is an improvement over the standard "linear with clamping" method for image display.

-Greg

Hi Greg,

Second, a quick explanation of what macbethcal does. Macbethcal takes a

Thanks, that was very useful!

Third, I wanted to point Alexa and other interested readers to the following reference, which explains the parts of human vision pcond attempts to simulate and the algorithms behind it:

    http://radsite.lbl.gov/radiance/papers/lbnl39882/tonemap.pdf

I'll be the first to admit that it is not a perfect simulation of human vision, and this is an ongoing area of research. (Try a websearch on "color appearance models" to get an idea.) However, I think others will agree that pcond can be useful, and is an improvement over the standard "linear with clamping" method for image display.

I don't doubt that it is better for image display and I think the idea with these kind of mappings is to go down the route of 'perception' rather than 'physics', one reason being the current limitations of CRT's e.g.

My concern is physical accuracy. I want to display an image on a screen (eventually) and people should not be able to tell it apart from the real world scene. Of course, I can only simulate colours and luminances that my daft 8-bit monitor can actually render.

Talking about the limitations of CRT's. Has somebody seen DLP-technology displays in action and maybe even made some colour measurements? Apparently, the colour resolution can be cranked up to 15 bits per channel and the luminance range is also much bigger. I only wonder, if the 15-bit resolution per channel is actually really an improvement in terms of colour resolution for a given luminance level (in comparison to a traditional 8 bit resolution) or is basically necessary to cover the increased luminance range?

Is that the same technology used in last year's emerging technology SIGGRAPH session:
Seetzen, H., Stuerzlinger, W., Vorozcovs, A., Wilson, H.R., Ashdown, I.,
Ward, W. and Whitehead, L., 2003, High Dynamic Range Display System. SIGGRAPH 2003, Emerging Technologies Program.
http://www.siggraph.org/s2003/conference/etech/hdr.html

Cheers,
Alexa

Hi Alexa,

...
However, I think others will agree that pcond can be useful, and is an improvement over the standard "linear with clamping" method for image display.

I don't doubt that it is better for image display and I think the idea with these kind of mappings is to go down the route of 'perception' rather than 'physics', one reason being the current limitations of CRT's e.g.

My concern is physical accuracy. I want to display an image on a screen (eventually) and people should not be able to tell it apart from the real world scene. Of course, I can only simulate colours and luminances that my daft 8-bit monitor can actually render.

There are surround issues to consider as well, especially since the monitor only covers a small portion of your view. Pcond attempts to adapt the original scene levels to the current viewing environment -- something not even an HDR display can do. You need something like the wide-field HDR viewer I built to test out tone operators for that.

Talking about the limitations of CRT's. Has somebody seen DLP-technology displays in action and maybe even made some colour measurements? Apparently, the colour resolution can be cranked up to 15 bits per channel and the luminance range is also much bigger. I only wonder, if the 15-bit resolution per channel is actually really an improvement in terms of colour resolution for a given luminance level (in comparison to a traditional 8 bit resolution) or is basically necessary to cover the increased luminance range?

If the 15 bits is in a linear range, and a DLP is a linear device, then you don't actually gain much in terms of dynamic range over an 8-bit/primary encoding using a standard 2.2 gamma. Beware of manufacturer's claims regarding contrast ratios and bit depths. There are no standards, and the marketing people have a field day with these specs. Bottom line: don't trust it unless you've measured it yourself.

Is that the same technology used in last year's emerging technology SIGGRAPH session:
Seetzen, H., Stuerzlinger, W., Vorozcovs, A., Wilson, H.R., Ashdown, I.,
Ward, W. and Whitehead, L., 2003, High Dynamic Range Display System. SIGGRAPH 2003, Emerging Technologies Program.
http://www.siggraph.org/s2003/conference/etech/hdr.html

We actually have a full-blown paper in this year's Siggraph, which is now online as well at Wolfgang Heidrich's website if you're interested:

  http://www.cs.ubc.ca/~heidrich/Papers/index.html#siggraph

This is a very cool device, but is only available in limited quantities as a prototype to select companies and individuals. Sunnybrook Technologies (http://www.sunnybrooktech.com/ Vancouver, BC) hopes to be manufacturing the real deal by next year. As I said, even an HDR monitor doesn't deal with surround issues, so it's really quite difficult to reproduce the visual appearance of the real world in all it's glory.

-Greg

Yes, it is true that the chromaticities are known for those colours, BUT this is under the assumption of a certain illumination source. Will you have that same illumination source? Probably not.

Hi!

The trick is to use the SAME illumination source :wink: I have done a number of such measurements in archeological context by simply taking a camera and the color chart to the site. The method that is described in Rendering with Radiance requires some installation (I think it was developed to measure material values in some kind of "studio"). I did it a bit simpler: one person holds the chart next to the material, I take a photo of the surface with the chart. As long as there is no variation of lighting over the area (no highlights etc) this seams to be a rather comfortable way, and I think for me it was accurate enough. Of course the chart should not cover the 5% of the area at the border - the chances that it is in a different lighting situation is to high than.

However, any expensive hardware is fun - and maybe you can even compare the results of the color chart way and the hardware way. I would be interested in that :wink:

CU Lars.

···

--
Lars O. Grobe
grobe@gmx.net

Talking about the limitations of CRT's. Has somebody seen
DLP-technology displays in action and maybe even made some colour
measurements? Apparently, the colour resolution can be cranked up to
15 bits per channel and the luminance range is also much bigger. I
only wonder, if the 15-bit resolution per channel is actually really
an improvement in terms of colour resolution for a given luminance
level (in comparison to a traditional 8 bit resolution) or is
basically necessary to cover the increased luminance range?

We have just finished a project calles SynthLight, which sets out to teach
lighting through 3D (well, it's only stereoscopic images and movies).
http://www.learn.londonmet.ac.uk/packages/synthlight/index.html

We experimented a lot with different display technologies for our 3D
stuff. While DLP seems to be the be all and end all of projector
technologies (high output, contrasts in the range of 1000 .. 5000), the
colour rendering is really what lets them down. There are at least 3
generations of DLP projectors out now (DLP is a technology developed and
licensed by Texas Instruments), all seem to have issues with colour
rendering. If you compare the colours on a computer screen (TFT or CRT)
and compare them to a DLP projector hooked up to the same computer, you
will find that the colour rendering of the projector is noticeably (very
much) poorer. For colour, LCD projectors seem to be much better, although
they have less of a contrast ratio (several 100:1, rather than several
1000:1).

This link


has a very nice comarison chart, although it doesn't mention colour as
such (you will have to trust me on this one...).

Cheers

Axel