Indigo Renderer Forum

odd, i cant seem to find a free paper for the one in 2006.
is http://graphics.ucsd.edu/~henrik/papers/skin_bssrdf/ the one your refering to?

oodmb wrote:odd, i cant seem to find a free paper for the one in 2006.
is http://graphics.ucsd.edu/~henrik/papers/skin_bssrdf/ the one your refering to?

Check out this...

Anyone want to do the 'fun' job of tabulating the data in figure 1 of 'A Spectral BSSRDF for Shading Human Skin'.
I need the Deoxy-hemoglobin and Oxy-hemoglobin curves in particular, at better than 20 nm resolution

Logarithmic or linear values are fine.

*looks around hopefully*

hmm why not?

maybe i'll get some sort of school credit from my physics teachers

i'll have it to you in about maybe a week and a half at most

Edit: now that i realize that "figure 1" != "table 1", i know what your talking about. maybe more like 2-3 days.

Hehe.. yeah, it's a pretty easy job, just boring.
Better yet would be to find the actual source data

i hear there are some parapsychologist studing tesla's memory who have created a device that prints images directly from thaughts. i bet if you could find one of them you can think about the graph realy hard and it will print a table for you

or, i could just use maple

I think you're misunderstanding what i'm after... don't worry, i'll do it myself.

i am pretty sure i know what you mean,
but if its a boring task it can be made interesting by overcomplicating it and wrighting a program that does it.

what i misunderstand is not you, but maple (the version for windows 3). and apparently i am not alone in that.

being ocd as i am, turning that qualifies as fun for me and i would be more than happy to do it for you.

Ono, i know what you mean. Do you want the absorbance/transmittance of hemoglobin/oxyhemoglobin at different wavelengths (interval of 20nm)? I studied spectrometry pretty front top to bottom. I could do if if you havent already.

EDIT: The graph is in absorption coeff, but the task is pretty much the same lol.

What I wanted was the height of the Deoxy-hemoglobin and Oxy-hemoglobin curves, at sufficient x resolution so that i reconstruct them reasonably accurately. I've entered the data at 10nm intervals now, which should do.

ok, well i was about half way into entering the data at 10 nm using the graphing methods of photoshop when i had a bout between sleep and sleep prevailed.

Here's a test, just a single dermis medium on this one:



Note that all the colour comes from the absorption and scattering coefficients varying with wavelength

Very promissing, good progression, did you plan to model your shader as a multi layered system like other renderer (mentalray, renderman) since it's more practical than using layers of polygons ?

Thanks to pointing us to the latest jenssen research he never stop and the result is outstanding !!

Don

ono, dont use a model of the hand to test the SSS. from what i have read in virutaly every paper, the layer examples that they give for skin are based off of data taken from the face. the skin around hands has different controbution and thickness of layers. in the palm of the hand, there is often a coule of extra of dead skin because the hands are used so often and develop a calous like layer. the bumps in the skin on the hand follow very different patterns from the bumps in other parts of the body. (thumb print as compared to cell print). above the skin the ratio of water to oil is higher but rather than smoothing itself over the whole surface like it does on other parts of the body, it pools into the groovs on the hand creating a sparkly effect. below that there is a higher concentration of capillaries and they are larger.
the top layers on the back of the hand are much more like the ones on the face, but the veins and arteries pop and thus the thickness of the top layers vary in a pattern that is highly noticable.

here are some sites with head scans
http://www.farfieldtechnology.com/download/
http://www.cyberware.com/products/scanners/index.html

if you want, you could use the model i have posted here. its blender and has an ok topology, you might not have as many problems with it as you would have with a 3d scan.

Ok, thanks oodmb, I'll try with a head scan

domparis:
I'll definitely have a look at Jensen's multipole approximations etc..,
but right now I just want to get the full simulation working, speed be damned.