Re: Spikes in the reflected spectrum

Erik Nikkanen wrote:
> There has been talk of spikes in the light source.  What I would like to
> know is if there are spikes in the reflected spectrums of pigments or
> materials that are illuminated with smooth light sources.
>
> Do the spikes tend to be mostly illumination related phenomena?
>
> Erik

This is an excellent question.

There are available on the internet a number of collections of object 
reflectances. My favorite is the Michael Vrhel data:
Use an ftp program or your regular browser to visit
ftp.eos.ncsu.edu    , then look in /pub/spectra  .

The same data are copied on other web sites, but the files at North 
Carolina State University are the originals.

Before using these reflectance data in my research, I wanted to check 
that they all looked reasonable. That is I did not expect spikes or 
jumps really, and I expected that objects called yellow, for instance, 
actually reflected more at long wavelengths and reflected little in the 
blue end of the spectrum. I wrote a program specifically to step through 
the files one at a time and display a graph on the screen. One rainy 
afternoon I went through all the graphs. None of them showed spikes, 
they all looked smooth as expected.

You can also go to this URL:
http://www.cns.nyu.edu/ftp/ltm/SSR/   , which must be associated with 
Larry Maloney at New York University. I do not see a readme file there 
now, but perhaps the explanation is someplace in Maloney's web space, or 
I can email the former readme file. In any case, from that point there 
is a directory called Kodak that has (guess what?) the Vrhel data. In 
the directory Munsell are the famous reflectances of Munsell colors 
collected by Dorothy Nickerson in the early 1940s, analyzed by Jozef 
Cohen in his famous 1964 article, etc.

Another rainy afternoon I went through the Nickerson Munsell data, and I 
DID find spikes. I think in every case it was reasonable to believe that 
a transcription error had caused the spike. I was able to get rid of the 
spikes by changing a single digit of a single number, such as the first 
digit after the decimal point. So I don't think that those data have 
spikes either.

There is a new technology in our lives, and that is longer-lasting 
fluorescent pigments. If you live in the USA, you may have seen 
fluorescent yellow-green pedestrian crossing signs. In my understanding, 
the fluorescence is stimulated by blue light, and there is an 
ultraviolet absorbing layer to slow sunlight damage to the pigment. I 
don't know much detail, but some fluorescent pigments may radiate in 
"spikes." I could only find one graphic about this on the web:
http://www.psicorp.com/mazel/fluor/fluor.html   .

So... I think that basically, reflectances do not have spikes, but the 
exceptions are interesting and new exceptions could show up.

Oh, yes, check this link:
http://www.cs.utah.edu/~bes/graphics/spectra/      and this one:
http://www.it.lut.fi/ip/research/color/lutcs_database.html   .

I can supply the Vrhel data in separate files with internal labels, but 
do NOT have them on my web site:
http://www.jimworthey.com   .  Email me if you need this kind of help. 
On my web site, especially in the second article, "Render Calc," I talk 
about a "linear model" expressing the smoothness of object reflectances. 
That whole article depends on the idea that most reflectances are 
spectrally smooth.