Program of Color Science/Munsell Color Science Laboratory

The Munsell Color Company was founded by Professor Albert H. Munsell, the creator of the Munsell Color Order System and the Munsell Book of Color. The directors of the Munsell Color Company eventually sold the company's assets and created the Munsell Color Foundation. The Foundation was charged with furthering the scientific and practical advancement of color knowledge.

In 1983, the Foundation trustees voted to dissolve the foundation, and donate the proceeds to an academic institution for the creation and endowment of the Munsell Color Science Laboratory. RIT was selected as the recipient of this donation, and MCSL was born. (Note: There is no formal relationship between the RIT Munsell Color Science Laboratory and the commercial Munsell Color products sold by X-Rite.)

The creation of MCSL was in large part due to the efforts of Franc Grum, the first MCSL Director (currently Prof. Mark Fairchild) and R.S. Hunter Professor of Color Science, Appearance, and Technology (currently Prof. Roy Berns). Prior to founding MCSL and RIT’s Department of Color Science, Professor Grum was a member of the Munsell Color Foundation, long-standing friend of RIT as an employee of the Eastman Kodak Research Laboratories, and an Adjunct Professor in the Photographic Science Department.

Initially, MCSL and the Color Science Department were part or RIT's College of Graphics Arts and Photography. RIT later created the Chester F. Carlson Center for Imaging Science, and MCSL became a research laboratory within that Center, currently housed in RIT's College of Science. In 1989, MCSL and the Center for Imaging Science moved to a new facility with approximately 6,500 square feet of space dedicated to color science research and education.

In the spring of 2003, space opened up in a nearby building. After extensive renovations, MCSL and the Color Science program moved into what is now formally called the Color Science Hall. The collocation of all our offices and laboratory space has fostered an amazing collaborative spirit in what was already an exciting, cooperative organization.

After careful strategic review of color science at RIT, the Program of Color Science (PoCS) and Munsell Color Science Laboratory (MCSL) were more closely joined as a stand-alone multidisciplinary graduate program in the College of Science. This new chapter in the storied history of these programs began in 2013, 30 years after the lab’s founding, with an aim of further extending the reach of color science across more scientific disciplines and into even more application areas.

"Nulli Secundus"

The Franc Grum Memorial Scholarship was established after his untimely death in 1985. It is intended to support scholarship in optical radiation measurements and color science. The funds for this award were made possible by gifts from the friends and family of Franc, as well as from industry.

Professor Munsell was born in Boston, Massachusetts on January 6th, 1858 and died June 28th, 1918 at age 60. Author of A Color Notation (1905) and the Atlas of the Munsell Color System (1915). Both an artist of distinction and a gifted teacher of art, he developed the first widely-accepted color order system to make the description of color accurate and convenient and to aid in the teaching of color. The Munsell color order system has gained international acceptance and has served as the foundation for other color order systems.

Albert Munsell founded the Munsell Color Company in 1917. Later, in 1942, the Munsell Color Foundation was formed by the company to promote the advancement of the science of color. Ultimately, the Munsell Color Foundation led to the founding of this laboratory, the Munsell Color Science Laboratory, in 1983, at the Rochester Institute of Technology.

Related Scholarly Research and Imagery

Unfortunately, no high-quality photo of Professor Munsell is available. In 1998, Munsell graduate student Doug Corbin restored the photo from Munsell's book, A Color Notation. The details of the procedure appeared in the February 1999 issue of Color Research and Application. Doug has provided the following information pertaining to this restoration:

The full text of the CR&A submission can be found here.
The LARGE uncorrected image can be found here. (21MB TIFF file)
The scientific diaries of Albert Munsell are available here.

For some other insights into the work of Professor Munsell, here is a self-published work form 1907 entitled, "Color and an Eye to Discern it." This was provided by Rolf Kuehni, who originally retrieved it from The Hagley Museum and Library.

These files are available for download. All come "as is." We have found them useful, and done our best to ensure their accuracy. If you find out otherwise, please let us know.

Instrument Evaluation Database
This is a database of instrument evaluation files. These data accompany a pair of 2007 articles in Color Research and Application. These files are available for download. Most are in Microsoft Excel format.

In all files, instruments are identified only by letter. The classes of instruments are:

• Handheld hemispherical instruments: A,B,C,D
• Benchtop hemispherical instruments: E,F,G,H
• Bidirectional instruments: I,J,K,L
• Additionally, there are two pairs of identical models: A and B, and K and L.
• All colorimetric data are calculated using D65 and the 10° standard observer

Short-Term Repeatability Data

• 50 repeated measurements without replacement
• Pressed PTFE (spectral and colorimetric data)
• Glossy White Tile (spectral and colorimetric data)
• Individual instrument files (PTFE and glossy tile): A, B, C, D, E, F, G, H, I, J, K, L

Medium-Term Repeatability Data

• 10 measurements per hour for eight hours
• Pressed PTFE (colorimetric data)

Long-Term Repeatability Data

• 2 measurements per day for five weeks
• samples include: glossy and diffuse whites, cyan BCRA tile, cyan printed ink
• Pressed PTFE (spectral data) Note: worksheets are labeled by instrument number. An X indicates incomplete data.

MetaCow Spectral Image Database
METACOW: A Public-Domain, HighResolution, Fully-Digital, Noise-Free, Metameric, Extended-Dynamic-Range, Spectral Test Target for Imaging System Analysis and Simulation

Standard, easily accessible, test targets have long served the field of color imaging as a foundation for comparison of the performance of various imaging systems and algorithms and the open and meaningful exchange of research results. This website details the creation and application of a new digital color test target useful for research and development of color imaging systems. The target has several advantages over previous types of targets that include spatial resolution, dynamic range, spectral resolution, metameric properties, lack of noise, and continuous tonal variations. All these features can be important for visual assessment, computational analysis, and colorimetric evaluation. This target, known as METACOW, is freely available to all performing research in color imaging.

METACOW is basically a very large (4200 x 6000 pixel) full-spectral image. It is rendered at 5nm increments, between 380 to 760...and is thus about 3gigs in size. It is designed such that both halves of each "cow" appear to match when illuminated with CIE D65, and viewed with the CIE 1931 Standard Observer. This is the example shown above. Each half of the cow is actually maximally metameric with itself. The left half of each cow has the spectral reflectance of the GretagMacbeth Color Checker, while the right half is a Metameric black that has been specially calculated to maximize color difference under Illuminant A. This type of target is designed to test both illumination and spectral responsivities for digital imaging solutions. Examples of illuminant (left) and observer (right) metamerism are shown below.

The image on the top was rendered under Illuminant A with the CIE 2 degree Observer, and then converted to display sRGB using the CIECAT02 chromatic adaptation transform. The image on the bottom was rendered under D65 with actual digital camera spectral sensitivities. The metameric nature of the METACOW target should be readily apparent from these examples.

The METACOW test target available for download, for use in imaging system design and evaluation. We ask only that you credit the Munsell Color Science Laboratory in any publications. Note that it is a large image; make sure you are on a decent internet connection for the download. Included in the files are: many GB of fullsize METACOW Glory, Smaller and Easier to Manage MINIMETACOW, Matlab Source Code for Reading and Rendering METACOW, and Much More. Or you can download a smaller version (420x600x77).
Small METACOW Images
Large METACOW Images (500MB file!)

Lippman2000 Spectral Imaging Database
This project is named Lippmann2000 in honor of Gabriel Lippmann who in 1891 devised a method to perfectly reconstruct the spectral content of real world scenes. In spite of Lippmann's invention, a more primitive three-channel model, first demonstrated by James Clerk Maxwell 30 years prior, has dominated the color imaging field. The Maxwellian model, universal in today's color image capture systems, relies on the metameric properties of the human visual system to simulate the appearance of an original color. The capture of full spectral data, while holding advantage over traditional three-channel methods offers new challenges at every point in the imaging chain.

PLEASE NOTE: THESE IMAGES ARE MADE AVAILABLE FOR RESEARCH PURPOSES ONLY. All other uses are prohibited. Copyright remains with PoCS / MCSL.

• Real World Spectral Images
• Synthesized Spectral Images
• Technical Papers on Spectral Imaging
• Programs useful for spectral imaging research
• Spectral Measurement Database

Overview and History

In the 1940's the color science community recognized that the most visually-uniform color space to date, the Munsell Color Order System, had inconsistencies that required examination and remedy. Towards this goal, a large-scale visual experiment was taken with many observers across several continents. The results amounted to an adjustment of the target color coordinates for the Munsell colors. The files here reflect that correction.

There are three files available for download. All are of the same format: six columns of Munsell hue, Munsell value, Munsell chroma, CIE x, y, and Y. The chromaticity coordinates were calculated using illuminant C and the CIE 1931 2 degree observer.In a sense, all three files represent the same set of data, in that all depend on the scaling experiments of the late 1930's.

A report entitled "One Set of Munsell Re-renotations," by Deane B. Judd and Dorothy Nickerson was issues by the National Bureau of Standards (now NIST) in 1967. As the title implies, they proposed an alternative to the original renotation scheme. As far as we know, these did not receive much attention, and their utility is uncertain. The report and associated data table have been scanned. If you use this please let us know! We would be interested in any useful application of the report of data.

One important note is that these data are taken from Wyszechi & Stiles 2nd Ed (1982) and the value scale is base on the original fifth order polynomial (relating Y/Ymgo to V). Modern instruments rererence the perfecting reflecting diffuser, so you may have better results if your multiple the Y values in these tables by 0.975, which is Ymgo, the Y value for the smoked magnesium dioxide reference white.

NONE OF THESE DATA SHOULD BE CONFUSED WITH ACTUAL
MEASUREMENTS FROM A MUNSELL BOOK OF COLOR!

all.dat: real and unreal

File download: all.dat

These are all the Munsell data, including the extrapolated colors. Note that extrapolated colors are in some cases unreal. That is, some lie outsize the Macadam limits.

This file should be used for those performing multidimensional interpolation to/from Munsell data. You will need the unreal colors in order to completely encompass the real colors, which is required to do the interpolation when near the Macadam limits.

real.dat: by the book

File download: real.dat

These are real colors only, "real" being those lying inside the Macadam limits. Specifically, these are those colors listed the original 1943 renotation article (Newhall, Judd, and Nickerson, JOSA, 1943).

This file should be used for a complete mapping between the Munsell system and its CIE equivalents. Note, however, that many of these colors were not used in the original scaling experiments, and are therefore extrapolated or at best interpolated from the test colors used.

Flash! Here are sRGB values and CIELAB for most of the colors in the real.dat file. There are some important notes regarding these data in the spreadsheet.

1929.dat: back to the source

File download: 1929.dat

These are only those colors physically appearing in the 1929 Munsell Book of Color. These data might be of useful for those interested in the input colors used for the scaling experiments leading to the 1943 renotation. Remember though, these are renotation colors of those original patches, not necessarily the colors of the input data used in the visual experiment.

We are pleased to make these diaries available online. The links below are individual PDFs, each in the range of 0.5 to 1MB. They represent all of volumes A and B in approximately twenty page increments. The index pages list the names of people mentioned in the diaries. Sorry, but there is no subject index. If you would like to create one, we will gladly publish it here!

Note that some PDFs are actually more than 20 pages. We grouped the tiles by the page number in the typed copies. Many pages were inserted with letter notation (4a, 4b, 4c, etc). Also, some of the handwritten pages are unnumbered. You may want to download the document before or after to make sure you get the desired pages.

ORIGINAL COVER SHEET
Below is the text from the cover sheet in the diary binders as received. It was slightly edited for typographical errors. 

The diary hereby made available is one kept by A. H. Munsell during the years in which he was developing both the Munsell color system and apparatus and charts by which to explain it.

A typewritten copy was made at the Munsell Color Company in the years 1920-23 from 6 volumes of a handwritten diary kept by Professor Munsell. Drawings and sketches were all hand-traced, and handwriting was inserted where corrections or additions were made in the original. Volume A covers the period 1899 through May 1908; volume B, May 1908 through 1918.

In 1939 the Inter-Society Color council, with permission from the Munsell family and company, deposited a bibliofilm negative of this typed material with the American Documentation Institute, their Document No. 1307. Early documents of this Institute now filed with the photo-Duplication Division of the Library of Congress, Washington, D.C., from which photoprints of microfilm positives may be ordered.

This copy of the Munsell Diary is one of three photocopied by Hunter Associates Laboratory in the spring of 1973 from the original typed copy loaned to Dorothy Nickerson by the Munsell Color Company. The three copies are filed with Miss Nickerson, the Rochester Institute of Technology through Milton Person (these copies are now at the Munsell Color Science Laboratory), and the Hunterlab library.

Reports of the development and application of the Munsell Color Systems since 1918 are available in the two historical papers by D. Nickerson: 1940, Journal of the Optical Society of America, 30, 575-586; 1969, Color Engineering, 7, 42-51.

Hunter Associates Laboratory, Inc.
Fairfax, VA. 22030
June, 1973

These files are available for download. All come "as is." We have found them useful, and done our best to ensure their accuracy. If you find out otherwise, please let us know.

CIE Standard Colorimetric Observer Data

• 1931 2° CIE Standard Colorimetric Observer Data
• 1964 10 °CIE Standard Colorimetric Observer Data
• Excel spreadsheet
   

CIE Standard Illuminant Data

• CIE illuminants A and D65, 300nm to 830nm, sampled at 5nm (Excel)
• Excel Daylight Series Calculator: use the CIE method to calculate relative spectral power distributions for your choice of Correlated Color Temperature
• CIE Fluorescent series: F1 to F12, 380nm to 730nm, sampled at 5nm (Excel)
   

Full set of 1nm data, including all of the following:

• Illuminant A
• Illuminant D65
• VM(λ) 1988 Spectral Luminous Efficiency Function for photopic vision
• V'(λ) Spectral Luminous Efficiency Function for scotopic vision
• 1931 2° CIE Standard Colorimetric Observer Data
• 1964 10 °CIE Standard Colorimetric Observer Data
• Excel with all of the above
   

Spectral Data for Commonly Used Color Products

• These are NOT standard CIE data, but they are useful for many applications in color and spectral imaging. Additionally, these data are not standardized for the repective product. These data are actual measurements of products in the Munsell Lab or related organizations.
• Macbeth Colorchecker (Excel Spreadsheet)
• CERAM Series II tiles (Excel Spreadsheet)
   

Code for CIE Color Difference formula CIEDE2000

Note: The above code was debugged to match the output of the Witt spreadsheet.

• These are NOT official CIE code, and are not sanctioned or guaranteed in any way.
• Excel Spreadsheet, originally provided by Dr. Klaus Witt, BAM, Germany
• Matlab code
• IDL code
• Visual Basic code, intended to be used as a macro within Excel.
   

Pointer Data Set

• This spreadsheet is available by permission from Dr Michael Pointer. These data are those on which the often cited paper was based. The original article is "The Gamut of Real Surface Colors", M.R.Pointer, Color Research and Application 5 (1980).
• Excel Spreadsheet, as provided by Dr. Pointer
   

CIE 170-1:2006 Cone Fundamentals for Various Field Sizes and Observer Ages

• Excel Spreadsheet to Compute Cone Fundamentals (Color Matching Functions) in Terms of Energy for 5nm Increments
• Matlab code matching the above spreadsheet. (Required files: Alms.txt, RelativeMacularDensity.txt, docul.txt)

Planck's Blackbody Equation Enumerated

• Excel Spreadsheet to Compute Blackbody Spectral Curves
• This spreadsheet is provided as a courtesy. It was derived form wikipedia.com articles in March 2010. The plot makes a nice picture. If you plan to do real science, we suggest you verify the calculations yourself.

RIT’s Program of Color Science and Munsell Color Science Laboratory have a rich history of education and research dating back over 30 years to the early 1980s. Well over 100 graduate-program alumni have helped produce about 1000 journal papers and conference proceedings in areas such as spectroscopic and colorimetric instrumentation, color modeling and formulation, chromatic adaptation and color appearance, color matching functions and metamerism, color difference perception and prediction, image perception and reproduction, measurement and conservation of cultural heritage, and many more.

Going forward, these research initiatives will be expanded in an even more multidisciplinary fashion. The program and lab will work closely with researchers in the fundamental scientific disciplines of biology (human and animal vision), chemistry (colorants and formulation), physics (optical spectroscopy and illumination sources), mathematics (modeling of systems and observers), and psychology (understanding color perception). In addition, and in step with RIT’s strong history in applied research, we will pursue the applications of color science to a wide variety of technology areas including photography, imaging, textiles, computing, materials, photonics, lighting, sustainability, systems engineering, architecture, packaging, printing, gaming, cinema, and design.

Please see the pages linked here for some archives of information from past student and faculty projects and visit the individual faculty web pages for more detailed research information and opportunities.

Each calendar year, the faculty, staff, and students of PoCS/MCSL gather some highlights of the past year into an annual report for our friends, supporters, alumni, and others. These reports also include a record of the journal publications, conference presentations, and other accomplishments of the group. They are also good references for lists of then-current students and a running list of MCSL alumni. Please take a moment to download and review our latest annual report (or historical reports dating back to the lab's founding).

Annual Report 2018
Annual Report 2017
Annual Report 2016
Annual Report 2015
Annual Report 2014
Biennial Report 2012-13
Annual Report 2011
Annual Report 2010
Annual Report 2009
Annual Report 2008
Annual Report 2007
Annual Report 2006
Annual Report 2005
Annual Report 2004
Annual Report 2003
Annual Report 2002
Annual Report 2001
Annual Report 2000
Annual Report 1999
Annual Report 1998
Annual Report 1997
Annual Report 1996
Annual Report 1995
Annual Report 1994
Annual Report 1993
Annual Report 1992
Annual Report 1991
Annual Report 1990
Annual Report 1989
Annual Report 1988
Annual Report 1987
Annual Report 1985-86
Annual Report 1984-85

This is a collection of past research results, often in the form of student theses or unpublished technical reports.

Observer Function Database
This is a database of observer functions (color matching functions) that were derived in a PhD thesis work of Yuta Asano. The database includes individual colorimetric observer model, categorical observers, and estimated CMFs for 151 color-normal human observers. (more info)

Introducing (pronounced Waypoint) Wpt
A normalization methodology has been developed that linearly transforms sensor values / cone excitations (or linear transforms of sensor excitations) into a material color equivalency representation that can be used as a waypoint for defining Material Adjustment Transforms. The normalization process adjusts for the white point and independently preserves the perceptive aspects of lightness, chroma, and hue resulting in an opponent like coordinate system designated by the axes W, p, and t. (more info)

Introducing WLab
A set of invertible non-linear transforms was derived that adjusts Wpt (Waypoint) coordinates to and from a more perceptually uniform coordinate system (WLab or Waypoint-Lab) that allows for the advantageous features of Wpt to be directly applied to situations where other standard color spaces are typically used. (more info)

Euclidean Color Spaces
A movie [Euclid2CIEDE2K (3).mov] of how the CIEDE2000 system is embedded into an Euclidean space.

iCAM06: HDR Rendering
The latest High Dynamic Range rendering model.

Advanced Image Quality Studies of LCTVs
Detailed image quality analyses of liquid crystal televisions.

Helmholtz-Kohlrausch Effect
The Perceptual Amplification of Color for a Common Computer Monitor: Helmholtz-Kohlrausch at Work on the Desktop Computer

Fluorescence Measurement
Evaluation of Bispectral Spectrophotometry for Accurate Colorimetry of Printing Materials

Hue correction Look Up Tables
These tables (LUTs) are used to transform CIELAB coordinates to and from Hung & Berns hue-corrected space. This information pertains to Appendix H of the Ph.D. dissertation A Paradigm for Color Gamut Mapping of Pictorial Images, by Gustav J. Braun,RIT, 1999. See LSO P.Hung and R.Berns, "Determination of Constant Hue Loci for a CRT Gamut and Their Predictions Using Color Appearance Spaces," Color Res Appl 20, 285-295, 1995.)

There are two plain text, tab-delimited ASCII files: forward transform and inverse transform. The data from these files can be used to estimate the destination hue for any given input color, specified by its [Cab*,hab] coordinates, using bilinear interpolation.

Evaluating Color Matching Functions
Research on evaluating the 1931 CIE color matching functions.

Spectral Sensitivities
Evaluation and Optimal Design of Spectral Sensitivities for Digital Color Imaging.

Paint Research
Developing a Spectral and Colorimetric Spectral and Colorimetric Database of Artist Paint Materials.

Visit the Art-Spectral Imaging website.

Bechmarking Art Image Interchange Cycles

This is a collection of past research results, often in the form of student theses or unpublished technical reports.

Many cultural heritage institutions are currently spending significant resources photographing their works of art for a variety of applications with distinctly different requirements. To create reproductions of their artwork, cultural heritage institutions employ a range of technology and a variety of workflows. A similar variety is used to publish these images in a number of output media. This project was undertaken to explore these workflows, their requirements, and the resulting image quality of the reproductions produced.

The main goals of this project were to: (1) determine the image quality inherent in the art image interchange cycles in use today, (2) understand the image quality expectations of the users, and (3) develop the capability to tie the two together. The three-year project started in April 2008 with financial support from The Andrew W. Mellon Foundation. (download final report 32MB PDF)

A three-year grant from the Andrew W. Mellon Foundation has established the Studio for Scientific Imaging and Archiving of Cultural Heritage at RIT, a research, outreach, and service facility under the leadership of Dr. Roy S. Berns, R. S. Hunter Professor in Color Science, Appearance, and Technology within the Program of Color Science and Munsell Color Science Laboratory. The vision statement for the Studio is to promulgate scientific imaging practices within cultural heritage institutions. We firmly believe that the best way to effect lasting changes in how institutions digitize their cultural assets is to lead by example. The proposed Studio will contain aspects of both imaging services and conservation science departments found at museums, libraries, and archives. The Studio is a natural extension of previous research sponsored by the Andrew W. Mellon Foundation with Dr. Berns as principal investigator.

Outreach
Outreach is a key component of the Studio, addressing the education barrier to incorporating scientific imaging within imaging services.

We will continue to submit papers for conferences and refereed journals, in similar fashion to past projects. There will be more emphasis on conservation than in the past, such as the American Institute of Conservation, the CIC Image Archiving conference, and the Museum Computer Network.

A short course will be developed to teach color science to artists, conservators, and curators. The will improve the general knowledge of museum personnel and provide important background that will aid in knowledge transfer of the results from the proposed project. It will also be a vehicle to demonstrate applications of spectral and four-light imaging systems. For example, a painting imaged with the four-light system was input to Maya, a computer-graphics rendering software package used by animators. A virtual museum gallery was produced and the painting rendered for spot illumination, shown in the figure above. Such images and walk-through animations will be included in the short course notes.

Workshops will be developed that are hands-on, in similar fashion to the workshop given during the previous Mellon-funded project, “Improving Artwork Reproduction Through 3D-Spectral Capture and Computer Graphics Rendering – Phase 2.” These will be presented both at RIT and museums.

Roy Berns will visit three institutions per year, on average, and demonstrate the techniques developed and tested at the Studio. Each visit will last approximately one week. Equipment will be brought as needed. During the visit, a workshop and short course will be delivered. Such visits will be extremely valuable as a source of knowledge transfer for both imaging services and Berns.

Another type of outreach is software. Any software developed during the proposed project will be downloadable for free from www.art-si.org. We are using the Matlab programming language and have purchased a commercial license during 2012 using Berns' discretionary funds. This enables executable software to be written and distributed, either as a commercial product or for free at our discretion. Funds from the proposed project will be used to maintain the commercial license during the course of the project. As an example, all the four-light imaging software developed during the last Mellon-funded project is available for downloading. The software is executable and includes a graphical interface to run the software. A screenshot from ArtViewer is shown in Figure 13, a program that renders images for user-selectable lighting geometry. The specific software to be written is listed in the expected outcomes.

Matlab was selected because we have over a decade of experience with this language and an extensive subroutine library of color-science tools.

The purpose of the Studio is to lead by example and prove that an imaging studio using commercial equipment can provide both images for science and scholarly communication. This approach will clarify the two imaging stages of analysis and synthesis. The archival images can be used directly by both conservators and conservation scientists. Following rendering (synthesis), images are used for printing and web display..

Four imaging systems are envisioned.

The first is easel based, using a high-resolution medium-format commercial camera and Xenon strobes, typically found in museum imaging services. Our particular camera is a new Sinar Photography AG system, consisting of a 86H 48MP color sensor, rePro camera body, eShutter with HR100 lens, and the CTM attachment. “CTM” stands for “color to match” the product name for a filter slider housing a pair of filters designed by the Roy Berns. When combined with software written at RIT, this system is capable of significantly higher color accuracy without a loss in spatial image quality. This is a commercialization of our Dual-RGB approach to scientific imaging. This approach was tested in our first Mellon-funded project. One feature of this camera is that by using only one of the filters, it produces images that are equivalent to a typical high-quality commercial camera. This system will have the greatest leverage towards demonstrating the advantages of scientific imaging within an imaging services studio. This system will image flat artwork and produce images appropriate for scientists, conservators, and photographers working in imaging services.

The second system will use the same easel and camera stand as the first system. A Canon commercial DSLR camera will be used. The system will be initiated using a Canon 5D Mark I and upgraded during the project to a higher-resolution DSLR. This will facilitate capturing data for computer graphics rendering of artwork using four-light polarization enhanced photometric stereo imaging, abbreviated as four-light imaging. This technique was developed during our last two Mellon-funded projects. This system will be used primarily for research since one of the proposed research themes will improve this system.

The third imaging system will be a “rapid system” using a Canon 5D Mark II DSLR modified for use as a Dual-RGB or standard DSLR. Rapid systems are gaining popularity for artwork that does not require the highest resolution and spatial image quality. The camera will be mounted on a motorized copy stand. A pair of Xenon linear strobes and controlling power pack will complete the system. This system will be used for imaging drawings, watercolors, photographic prints, and similar media for both research and imaging services. The first and third systems will be used for the imaging services. As such, we will not be imaging sculpture and other three-dimensional works.

The fourth system will be a spectral system, which consists of a 1.4MP Lumenera LW165m monochrome sensor (1392x1040 pixels) coupled with a CRI liquid crystal tunable filter. This system will be attached to the camera stand used in the first and second systems. The spectral system will be used primarily for research. Specifically, it will provide “ground truth” for our proposed research in improving pigment mapping. The output of this system will be a complete spectral reflectance curve for each pixel.

Research Themes

Rendering Scientific Images for the Web and Print

Bernie Lehmann,Apres Auvers, oil on canvas. Scientific image is on the left and display-preferred rendering is on the right. (The tone-mapping changes were amplified to clarify the type of adjustments necessary when rendering scientific images.)

Rendering scientific images addresses the assertion that scientific images are not appropriate for rendering. We will incorporate published literature on color and tone mapping that accounts for dynamic range, color gamut, and image size differences. Software will be developed that automatically re-renders image data that is based on user-supplied input such as output media and size. We will verify the algorithms with visual experiments where subjects will compare different approaches to rendering with the actual artwork. As an example, a painting was imaged using a scientific approach, shown above on the left. Since most displays have ambient flare (light reflecting off the display screen) and we view displays in fully lit rooms, a change in tone mapping is required for images rendered for web sites, shown above on the right.

A second component of this research theme is improving the scientific quality of existing camera systems in museums, archives, and libraries. Although cameras producing more than the usual RGB data can achieve the highest accuracy, significant improvement can be made with RGB cameras using profiling software optimized for scientific imaging. This software was developed in the last Mellon-funded project and is a component of the four-light imaging system. This piece of the software will be extracted, improved, and made into a stand-alone program that any camera system can use.

Improving Spectral Estimation of Using Commercial Camera Systems

The second research theme will demonstrate that conservators and conservation scientists can use images produced in an imaging services studio. In the past, we have used science to improve the color accuracy of images produced in imaging services. In this research, we will develop techniques to expand the utility of scientific images.

For conservators and conservation scientists, spectral data are more useful when such data are applied to pigment mapping, pigment selection for inpainting losses, and simulating proposed treatments, among others. The spectral data can be obtained using a number of different approaches including hyper-spectral (dozens to hundreds of channels sampling the visible and NIR spectral regions), multi-spectral (five to a dozen channels using a monochrome camera and either interference filters or narrow-band colored LEDs), or Dual-RGB. Despite these many approaches to spectral imaging, there is a paucity of software to process the images that is useful for art conservation. In this research, improvements will be made to our previous research in pigment mapping. Two approaches will be implemented. The first uses a priori knowledge such as the artist’s palette or direct spectral measurements using a small aperture spectrophotometer, a device quite common in imaging services, surprisingly. The second will combine low-resolution hyper-spectral and high-resolution multi-spectral or Dual-RGB images yielding high-resolution hyper-spectral images. This produces images of much greater resolution than achievable by cameras designed for spectral imaging. As an illustrative example, a painting was imaged using a Dual-RGB camera system producing the scientific image shown above. Assuming a known palette of paints and their optical properties, the image was mapped to concentration, and then used to generate a high-resolution spectral image.

Implementing a Spectral Color Management Workflow

The third theme will incorporate spectral color management and workflow into the proposed Studio imaging services. This research theme will provide further evidence of the utility of scientific imaging within imaging services and demonstrate that the Studio is at the forefront of color reproduction. One of the PI’s part-time doctoral students is a working member of the International Color Consortium (ICC) involved in spectral color management workflows. (All imaging services studios use an ICC workflow through Adobe Photoshop.) We will incorporate ICC advances in our imaging practices. We have already made progress in this area. Our current multi-spectral or Dual-RGB image encoding is a multi-channel tiff file where the first three channels are rendered for ProPhotoRGB and the remaining channels are the multi-spectral or Dual-RGB data, useful for scientific analysis, spectral color management, and spectral estimation. A screen-shot from Photoshop is shown in the above image.

Participating in spectral-based ICC activities will insure that new imaging systems and future versions of Adobe Photoshop will have built-in features for our approaches to changing imaging services practices. It will demonstrate that the Studio is leading the museum imaging community.

Improving Four-light Imaging

​Diffuse image for a set of targets where the diffuse data were based on averaging four different lighting geometries.

The fourth theme addresses a lack of turnkey software and creates another scientific imaging service. In our last Mellon-funded project, we developed the four-light system that captures diffuse color and surface normal information. After using the system for the last six months, we have found that there are opportunities for improvement including spatial quality, incorporating conventional imaging, adding a metallic feature to the rendering software, and improving realism. For example, several panels from the Artist Materials Database were imaged using the four-light system, as seen above, Theoretically, there should not be any observable texture for these samples and the deployed software used the average of the four images to calculate the diffuse data. As seen below, the average method results in only a small amount of observable texture. However, because surface normal maps produce shading but not shadows, rendered images do not appear as realistic as conventional imaging. Furthermore, the images appear blurry because all the high-spatial-frequency specular data are filtered by cross polarization. By using the minimum of the four images rather than the average, the diffuse maps produce shadows, improving both image sharpness and realism. This is similar in concept to rendering the scientific images for display. An additional goal of this theme is increasing the utility of the image data, for example, documentation to quantify long-term color and surface changes.

Diffuse Image Comparison​
Diffuse image comparison between average and minimum methods of combining four different lighting geometries.

Many past publications are available at the Art-Spectral Imaging site. Listed below are work for the current project, and moving forward from 2014.

Journal Articles:

GEOMETRY-INDEPENDENT TARGET-BASED CAMERA COLORIMETRIC CHARACTERIZATION
Farhad Moghareh Abed, Roy S. Berns, Kenichiro Masaoka

The Saunderson equations were used to compensate for differences in illuminant geometry between imaging systems and reference spectrophotometers. J. Imaging Science Technology 57, 050503-1 – 15 (2013). download article

WAVELENGTH-DEPENDENT SPATIAL CORRECTION AND SPECTRAL CALIBRATION OF A LIQUID CRYSTAL TUNABLE FILTER SYSTEM
Roy S. Berns, Brittany D. Cox, Farhad Moghareh Abed

A practical workflow was developed for the spectral calibration of a multispectral imaging system using liquid crystal tunable filter (LCTF) based . The workflow calibrated the system such the scale of spectral reflectance factor was transfered to the imaging system. The novel aspect of this research was a spatial correction accounting for the angular dependency of interference filters. Applied Optics, 54 3687-3693 (2015). download article

Conference Proceedings:

ARTIST PAINT SPECTRAL DATABASE
Roy S. Berns

Colorimetric and spectral gamuts are useful for the design of imaging systems such as display and print, to evaluate encoding errors for digital photography, for spectral reconstruction, and for lighting design. An artist paint spectral database was developed that includes spectra, colorimetry, eigenvectors, and optical data to produce additional spectra if desired. Nineteen acrylic-dispersion paints were sampled and Kubelka-Munk theory with the Saunderson correction was used to characterize each paint’s optical properties. Each chromatic paint or hue-adjacent paints were mixed computationally with white and with black at a range of concentrations producing approximately uniform sampling in CIELAB. In total, there are 23 hues and one gray scale with 770 unique spectra. IS&T Color and Imaging Conference, submitted (2016). download proceedings, download Excel spreadsheet

MODIFICATION OF CIEDE2000 FOR ASSESSING COLOR QUALITY OF IMAGE ARCHIVES
Roy S. Berns

Evaluating color accuracy of image archives is accomplished using test targets, colorimetric measurements, and total color difference calculations. Both CIELAB and CIEDE2000 are used with the later often preferred being an international standard and having a weighting function for chroma position. Its lightness weighting function is problematic when used for image archiving because it allows for greater lightness errors for dark colors, colors critical in defining image quality. A statistical analysis was performed on data used to derive CIEDE2000. The results do not support the weighting function for lightness. For imaging applications, it is recommended that the lightness weighting function be replaced with unity. IS&T Archiving Conference in press(2016). download proceedings

USING MAYA TO CREATE A VIRTUAL MUSEUM
Brittany D. Cox and Roy S. Berns

Surface normal and diffuse color maps were rendered for different lighting configurations using computer graphics software. The maps were measured using the four-light imaging system developed for the Studio for Scientific Imaging and Archiving of Cultural Heritage. IS&T Archiving Conference 51-55 (2015). download proceedings

IMAGING ARTWORK IN A STUDIO ENVIRONMENT FOR COMPUTER GRAPHICS RENDERING
Brittany D. Cox and Roy S. Berns

The four-light imaging system developed by T. Chen and Berns was improved using a heirarchical selection technique to remove specular highlights rather than cross-polarization. IS&T/SPIE Electronic Imaging 939803-939803 (2015). download proceedings

INCREASING THE VERSATILITY OF DIGITIZATIONS THROUGH POST-CAMERA FLAT-FIELDING
Joel Witwer and Roy S. Berns

When lighting to enhance surface texture, there is significant fall off. An image of this falloff using a white board can be used for archiving or not used to improve realism. IS&T Archiving Conference 110-113 (2015). download proceedings

ETRGB: AN ENCODING SPACE FOR ARTWORK IMAGING
Roy S. Berns and Maxim Derhak

A new encoding space, ETRGB (Extended Tristimulus RGB), was derived to facilitate accurate archiving of modern artist materials such as metallic flakes and interference pigments. IS&T Archiving Conference 74-77 (2015). download proceedings

Technical Reports
SPECTRAL IMAGING USING A LIQUID CRYSTAL TUNABLE FILTER – PART II
Farhad Moghareh Abed and Roy S. Berns

A practical workflow was developed for the spectral calibration of a multispectral imaging system using liquid crystal tunable filter (LCTF) based . The workflow calibrated the system such the scale of spectral reflectance factor was transfered to the imaging system. The novel aspect of this research was a spatial correction accounting for the angular dependency of interference filters. (Refereed article below) download report

ARTIST PAINT TARGET (APT): A TOOL FOR VERIFYING CAMERA PERFORMANCE
Roy S. Berns

A new 24-patch target was developed to verify colorimetric accuracy for artist paints. download report, auxiliary data )

SPECTRAL AND COLOR CHARACTERISTICS OF BRONCOLOR PULSO F4 STROBE WITH UVE PROTECTION DOME
Roy S. Berns

Data were compiled for a Broncolor F4 strobe with UVE protection dome and CIE daylight spectral power distributions with equivalent correlated color temperature, distribution temperature, and 5500K. download report, auxiliary data

CAMERA ENCODING EVALUATION FOR IMAGE ARCHIVING OF CULTURAL HERITAGE
Roy S. Berns

A large set of colors were compiled including eight fluorescent artist paints, a sampling of glossy high-chroma artist paints, computationally extended artist paints, the Pointer gamut, and a computationally extended Pointer gamut. Their colorimetric data were used to evaluate potential encoding errors for sRGB, AdobeRGB(1998), ProPhotoRGB, and ProStarRGB. download report

A NEW ENCODING SYSTEM FOR IMAGE ARCHIVING OF CULTURAL HERITAGE: ETRGB
Roy S. Berns and Maxim Derhak

A new encoding space, ETRGB (Extended Tristimulus RGB), was developed to insure modern artist material such as those containing metals and interference pigments are encoded without errors caused by clipping or gamut mapping. (Proceeding below) download report, auxiliary data - ICC profile)

EVALUATING SOLID STATE AND TUNGSTEN-HALOGEN LIGHTING FOR IMAGING ARTWORK VIA COMPUTER SIMULATION
Roy S. Berns

Solid state lighting was evaluated for use as a taking illuminant in the imaging of artwork. Based on computer simulation of camera signals, a high color-rendering solid state light achieved the same accuracy as a Xenon strobe. download report

SPECTRAL SENSITIVITY AND TRANSMITTANCE MEASUREMENTS OF A SINAR 86H CTM DUAL-RGB DIGITAL CAMERA
Roy S. Berns and Yixuan Wang

The spectral sensitivity and spectral transmittance factor of the sensor and filters of a Sinar CTM (color to match) digital camera system were measured using a calibrated monochromatic source and spectrophotometer, respectively. download report, auxiliary data

GEOMETRIC DISTORTIONS IN A CANON EF 50MM F/2.5 MACRO LENS
Joel Witwer

The geometric distortion was evaluated for a Canon lens. download report

LIGHTING EVALUATION FOR AESTHETIC IMAGING
Joel Witwer

The lighting capabilities of the Studio for Scientific Imaging and Archiving of Cultural Heritage were compared with typical practices at the Metropolitan Museum of Art. download report

Other:

INPAINTING SPREADSHEET OF GAMBLIN CONSERVATION COLORS
Roy S. Berns

Two-constant Kubelka-Munk theory with the Saunderson correction for refractive index discontinuity at the paint surface can be used to select pigments for inpainting leading to minimal metamerism. This Excel spreadsheet implements both colorimetric and spectral matching using the Gamblin Conservation Colors (circa 2000) as the optical database. The spreadsheet also includes a sheet to produce new optical data for your own paints. This is based on research by Berns and Mohammadi published in Studies in Conservation. download spreadsheet, download Studies article

Database of Spectral Images

The spectral information of  a handful of paintings with known constituent pigments was extracted. Two major categories of pigments were selected: old master and moderns. Each category contains several paintings with different number of primary pigments indicated following table. A set of primary pigment that was used for creating the validation paintings is shown in the following image:

These set of pigments have been evaluated comperehensivley by Okumura. Full spectral and pigment information of his research can be downloaded here.

The spectral data has been extracted using a LCTF-based acquisition system explained here.

A number of the examined paintings are shown in the image below:

Spectral Images and metadata At this time spectral images are stored as matlab *.mat files. sRGB files are TIF. Click the thumbnails for larger versions. Absorption and scattering coefficients (.xlsx format) Dark current image (camera cap on) for noise characterization purposes. Flatfield image (from a uniform Lambertian flat surface) for noise characterization purposes.

Painting Code	Primary Pigments	Download Link Thumbnail
C_1_1_1	1-Red Oxide 2-Yellow Ochre 3-Ultramarine Blue	Spectral sRGB
C_1_1_2	1-Red Oxide 2-Yellow Ochre 3-Ultramarine Blue	Spectral sRGB
C_1_1_3	1-Red Oxide 2-Yellow Ochre 3-Ultramarine Blue	Spectral sRGB
C_1_2_1	1-Red Oxide  2-Raw Umber  3-Yellow Ochre  4-Cobalt Blue	Spectral sRGB
C_1_2_2	1-Red Oxide  2-Raw Umber  3-Yellow Ochre  4-Ultramarine Blue	Spectral sRGB
C_1_2_3	1-Red Oxide  2-Raw Umber  3-Yellow Ochre  4-Ultramarine Blue	Spectral sRGB
C_1_3_1	1-Carbon Black  2-Red Oxide  3-Raw Umber  4-Diarylide Yellow  5-Ultramarine Blue	Spectral sRGB
C_1_3_2	1-Carbon Black  2-Red Oxide  3-Raw Umber  4-Yellow Ochre  5-Ultramarine Blue	Spectral sRGB
C_1_3_3	1-Carbon Black  2-Red Oxide  3-Raw Umber  4-Yellow Ochre  5-Ultramarine Blue	Spectral sRGB
C_1_4_1	1-Chromium Oxide Green  2-Red Oxide  3-Raw Umber  4-Yellow Ochre  5-Diarylide Yellow  6-Ultramarine Blue	Spectral sRGB
C_1_4_2	1-Chromium Oxide Green  2-Red Oxide  3-Raw Umber  4-Yellow Ochre  5-Diarylide Yellow  6-Ultramarine Blue	Spectral sRGB
C_1_4_3	1-Chromium Oxide Green  2-Red Oxide  3-Raw Umber  4-Yellow Ochre  5-Diarylide Yellow  6-Ultramarine Blue	Spectral sRGB
C_1_5_1	1-Chromium Oxide Green  2-Red Oxide  3-Raw Umber  4-Yellow Ochre  5-Pyrrole Red  6-Diarylide Yellow  7-Ultramarine Blue	Spectral sRGB
C_1_5_2	1-Chromium Oxide Green  2-Red Oxide  3-Raw Umber  4-Yellow Ochre  5-Pyrrole Red  6-Diarylide Yellow  7-Ultramarine Blue	Spectral sRGB
C_1_5_3	1-Chromium Oxide Green  2-Raw Umber  3-Yellow Ochre  4-Pyrrole Red  5-Diarylide Yellow  6-Ultramarine Blue	Spectral sRGB
C_1_6_1	1-Chromium Oxide Green  2-Raw Umber  3-Yellow Ochre  4-Cobalt Blue  5-Pyrrole Red  6-Diarylide Yellow  7-Ultramarine Blue	Spectral sRGB
C_1_6_2	1-Chromium Oxide Green  2-Raw Umber  3-Yellow Ochre  4-Cobalt Blue  5-Pyrrole Red  6-Diarylide Yellow  7-Ultramarine Blue	Spectral sRGB
C_1_6_3	1-Chromium Oxide Green  2-Raw Umber  3-Yellow Ochre  4-Cobalt Blue  5-Pyrrole Red  6-Diarylide Yellow  7-Ultramarine Blue	Spectral sRGB
C_2_1_1	1-Cobalt Blue  2-Hansa Yellow  3-Ultramarine Blue  4-Quinacradone Magenta	Spectral sRGB
C_2_1_2	1-Cobalt Blue  2-Hansa Yellow  3-Ultramarine Blue  4-Quinacradone Magenta	Spectral sRGB
C_2_2_1	1-Cobalt Blue  2-Phthalo Blue  3-Pyrrole Red  4-Hansa Yellow  5-Ultramarine Blue  6-Quinacradone Magenta	Spectral sRGB
C_2_2_2	1-Cobalt Blue  2-Phthalo Blue  3-Pyrrole Red  4-Hansa Yellow  5-Ultramarine Blue  6-Quinacradone Magenta	Spectral sRGB
C_2_3_1	1-Phthalo Blue  2-Hansa Yellow  3-Quinacradone Magenta	Spectral sRGB
C_2_3_2	1-Phthalo Blue  2-Hansa Yellow  3-Quinacradone Magenta	Spectral sRGB
C_2_5_1	1-Phthalo Blue  2-Pyrrole Red  3-Hansa Yellow  4-Pyrrole Orange  5-Quinacradone Magenta	Spectral sRGB
C_2_5_2	1-Phthalo Blue  2-Pyrrole Red  3-Hansa Yellow  4-Pyrrole Orange  5-Quinacradone Magenta	Spectral sRGB
C_2_7_1	1-Carbon Black  2-Phthalo Blue  3-Phthalo Green  4-Pyrrole Red  5-Hansa Yellow  6-Quinacradone Magenta	Spectral sRGB
C_2_7_2	1-Carbon Black  2-Phthalo Blue  3-Phthalo Green  4-Pyrrole Red  5-Hansa Yellow  6-Quinacradone Magenta	Spectral sRGB
C_2_8_1	1-Carbon Black  2-Phthalo Blue  3-Phthalo Green  4-Pyrrole Red  5-Diarylide Yellow  6-Hansa Yellow  7-Quinacradone Magenta	Spectral sRGB
C_2_8_2	1-Carbon Black  2-Phthalo Blue  3-Phthalo Green  4-Pyrrole Red  5-Diarylide Yellow  6-Hansa Yellow  7-Quinacradone Magenta	Spectral sRGB
C_Palette_Exp	1-Phthalo Blue  2-Hansa Yellow  3-Quinacradone Magenta	Spectral sRGB
C_Palette_FourPigs	1-Cobalt Blue  2-Phthalo Blue  3-Pyrrole Red  4-Permanent Green	Spectral sRGB
C_Palette_Pal	1-Phthalo Blue  2-Hansa Yellow  3-Quinacradone Magenta	Spectral sRGB

This software is available for noncommercial, educational, scholarly, and/or charitable purposes only

Software in support of this research program is posted here. Most of this will be executable MatlabTM, which requires Matlab's compiler runtime. The specific version required is either R2014b (version 8.4) or R2016a (version 9.0.1). The 2014b software may not run on El Capitan MacOS.

Link to Matlab

These applications work on Mac OS. Running these applications may result in an error because they are from an unknown developer.

To bypass this, right click the application and select 'Open', then enter an administrator login.

X-rite ColorChecker Digital SG (CCSG) Evaluation
The CCSG is often used to build and evaluate camera color profiles. This software evaluates the colorimetric accuracy of the CCSG contained in an image. The zip file contains executable Matlab, instructions, reflectance data of CCSG, test image, example output, and source code.

(2014b) Download zip file containing executable Matlab, instructions, reflectance data of CCSG, test image, example output, and source code

(2016a) Download zip file executable Matlab and source code

MultiSpectralTools – Dual-RGB Processor
Collecting a pair of RGB images using different colored filters or lights can be used for spectral estimation and improving color accuracy compared with an RGB camera. This approach has been patented by RIT and commercialized by Flux Data and Sinar. Over a number of years, we have been improving our software. This is the March 2016 version. It runs on 16-bit linear Tiff and Sinar DNG file formats.

(2014b) Download zip file containing executable Matlab, instructions, reflectance data of CCSG, test images, and source code

(2016a) Download zip file containing executable and source code

SpectralPicker
MultiSpectralTools has both colorimetric and spectral workflows. The file format is a nine-channel tiff, where the top RGB are color managed and the bottom six are the multi-spectral channels. These channels are used to display the average spectral reflectance factor of a user-defined circular aperture moved about the image. The colorimetric data, spectral data and a spectral reflectance factor plot can be saved if desired.

(2016a) Download zip file containing executable Matlab and source code

Four-Light Imaging Simple
Collecting images from four different lighting geometries enables calculating diffuse and surface normal maps. These maps can be used to render paintings and drawings for archiving and reprographics. When stored in .pfm format, our software ArtViewer can be used for real-time viewing. The program runs on several file formats including 16-bit linear Tiff and Sinar DNG.

(2014b) Download zip file containing executable Matlab, an instructional video, test images, and source code

ArtViewer
This program is similar to the RTI Viewer. It uses files generated using the Four-Light Imaging Simple software.

Downoad Zip File

Convert DNG to Tiff
This program converts DNG format to linear Tiff. It is based on DCRAW (https://www.cybercom.net/~dcoffin/dcraw/). The zip file contains executable Matlab and source code.

(2014b) Download zip file

(2016a) Download zip file

Bit Depth Evaluation and Assignment
Sometimes following a conversion of a raw format to unassigned tif, the image appears very dark, a result of the sensor having a lower bit depth than 16 bit. This program evaluates and reassigns bit depth.

(2014b) Download zipped program

Convert CR2 to Tiff
Many CR2 to Tiff converters add a tone curve. This program maintains linearity. It is based on DCRAW (https://www.cybercom.net/~dcoffin/dcraw/). The zip file contains executable Matlab and source code.

(2014b) Download zip file

(2016a) Download zip file

Convert Monochrome CR2 to Tiff
We had a Canon Mark III converted to monochrome. The zip file contains executable Matlab and source code. (We get better results using AccuRaw Monochrome.)

(2014b) Download zip file

(2016a) Download zip file