Subsurface scattering: my first work

ave · Post by **ave** » Wed Jul 16, 2008 10:08 pm

Hi! I am a PhD student in Physics, and I work in the field of optics. So guess how I was excited by the physically accuracy of INDIGO! This is my first serious render, to test how the absorption and scattering properties of the materials are rendered.
I am a noob, but this is fantastic, to me.
The liquids are skimmed milk, chardonnay and merlot from Jensen paper. Note that the milk is really tricky to simulate, since a single ray can be scattered thousands of time in the material.
24 h of rendering, (still a bit of noise).
In the next months I am thinking to use INDIGO to do real research on a material we are developing here. More details in a future post.
To me it is amazing that you can plug in real quantites (spectral intensities, scattering parameters, absorption...) from measurements. It is really more than CG: it is physics!

BbB · Post by **BbB** » Wed Jul 16, 2008 10:45 pm

Nice. I'm not sure the milk works for me though. First the volume is facetted, then the SSS is hardly visible. You sure your scene scale is ok?

Borgleader · Post by **Borgleader** » Thu Jul 17, 2008 12:05 am

Middle and right ones are nice, but the milk has facets as BbB mentioned.

ave · Post by **ave** » Thu Jul 17, 2008 1:10 am

I was expecting this kind of comments

... the milk is faceted because of a fault of mine: I did not subdivide my model enough, and when I realized it it was already a few hours of rendering. Anyway, it can be easiliy fixed.
About the milk appearance: i am pretty sure about scale and everything. Actually, to me it looks ok. Bear in mind that milk is a very particular scattering material, since the scattering is dominated by multiple scattering: a single ray can be scattered thousands of times inside the material.

Interestigly, ALL the white you see from milk is due to scattering by its fats and proteins particles! But the scattering is so much that it looks like an opaque material (it is the same for clouds, btw). Without SSS milk would appear like a reddish water.

Also, comparing visually with the results published by Jensen in:
http://graphics.ucsd.edu/~henrik/papers ... ie_theory/
to me it seems good.
Maybe the lighting conditions are not ideal: I am using just a single light from above and no env maps.
Thanks for the comments!

zsouthboy · Post by **zsouthboy** » Thu Jul 17, 2008 5:01 am

Looking good - can't wait to see what you post about your research.

Kram1032 · Post by **Kram1032** » Thu Jul 17, 2008 6:08 am

you don't need to subdivide it, you rather can smooth the normals....
To get save, do both together though^^

I think, I see some scattering...
Maybe a direct comparison to a diffuse white material (roughly of milk colour) could show the difference better

joegiampaoli · Post by **joegiampaoli** » Thu Jul 17, 2008 6:14 am

Yay!
Indigo goes scientific!
Cooooooooool....................

ave · Post by **ave** » Thu Jul 17, 2008 8:37 am

[quote="Kram1032"]you don't need to subdivide it, you rather can smooth the normals....
To get save, do both together though^^

Smooth the normals ?? What does it mean?

zsouthboy · Post by **zsouthboy** » Thu Jul 17, 2008 9:07 am

Turn on normal smoothing for the milk mesh is what he meant.

What 3d program are you working in?

ave · Post by **ave** » Thu Jul 17, 2008 10:02 am

This evening I did a couple of experiments at home with REAL milk to better understand how its SSS work.
Am I a physicist or not, after all ?

I tried to put on a similar lighting condition as in the render. What I understood (maybe it is obvious to you) is that the SSS depends critically on the kind of surfaces which surrounds the sample. In this case -as light is coming from above- the floor properties are critical.

Indeed, a great part of the light travels unperturbed in the milk and reaches the floor, where it is reflected (diffused) or absorbed in different proportions depending on the material.

The reflected part bounces back in the milk, where it is scattered again. It is like having two light sources: one from above and a weaker one from below.

You can see it clearly in the two pictures: in the first the floor was white paper. You see that milk looks almost opaque (as in the render).

In the second the floor is black (a bit reflecting) paper. You see that in this case only direct light from above is scattered, and the bottom part of the milk is almost black. Here SSS is much clearer!

I would try this next in INDIGO...

Of course this concept applies whenever you have SSS.

For the "smoothing normals" story: I am using CINEMA4d. Thanks for help.

zsouthboy · Post by **zsouthboy** » Thu Jul 17, 2008 10:41 am

No experience with C4D, sorry.

I like where you're going with this though

Sayris · Post by **Sayris** » Thu Jul 17, 2008 10:59 am

did you tried to use the phong-tag?

nice "experiment"

Borgleader · Post by **Borgleader** » Thu Jul 17, 2008 11:19 am

The new ones look very promising

nice stuff

Cire · Post by **Cire** » Thu Jul 17, 2008 11:49 am

Borgleader wrote:The new ones look very promising nice stuff

The "new ones" are photos, not renders.

Borgleader · Post by **Borgleader** » Thu Jul 17, 2008 11:52 am

Cire wrote:
Borgleader wrote:The new ones look very promising nice stuff
The "new ones" are photos, not renders.

lool whoops....i guess i should stop being a child and read the posts and not just check out the pictures

Subsurface scattering: my first work

Subsurface scattering: my first work

milk experiments!

Who is online