SSS Test Scene

Marcofly · Post by **Marcofly** » Wed May 23, 2007 2:32 am

stur, the test is very interesting, at least for me..

keep on testing, please! if you upload the .igs i can also render it for you..

Good work!

Stur · Post by **Stur** » Wed May 23, 2007 3:10 am

Cool it pleases some people.

Ok I've finally made a small & simple prog to generate test scenes. I'll improve it a bit to make it configurable with a textfile and I'll give it to the community.
Soon.

ps : it will be a console application for Windows, running with dot.net framework, sorry for linux users.
ps2 : maybe dot.net framework exists for linux though ...

zsouthboy · Post by **zsouthboy** » Wed May 23, 2007 8:24 am

Stur wrote:Cool it pleases some people.

Ok I've finally made a small & simple prog to generate test scenes. I'll improve it a bit to make it configurable with a textfile and I'll give it to the community.
Soon.

ps : it will be a console application for Windows, running with dot.net framework, sorry for linux users.
ps2 : maybe dot.net framework exists for linux though ...

As long as it's < .Net 3.0, Linux will be fine - Mono is an open-source implementation of .Net that is surprisingly high quality and mature.

And what part of the framework will you be using? Especially for a console application - just #include stdio.h and you can read and write files just fine.

Stur · Post by **Stur** » Wed May 23, 2007 7:28 pm

I coded it with VS Express C# 2005, so, it's supposed to run under .net 2.0, but I think it could run under .net 1.1 too.

I use Sytem, System.IO, System.XML and System.Globalization namespaces. I think it should run with Mono so, it's a good test anyway

Now to get back to the topic, I've launched a higher render yesterday evening, but it was still too noised to be shown.

Stur · Post by **Stur** » Thu May 24, 2007 8:02 am

I continue my quest to understand SSS. Here is a new test, still noised, but it's enough to be instructive.
I modified Suzanne a bit because I realized the mesh was not a closed volume. The eyes are now separated from the head mesh, and just diffuse white.

The absorbption RGB is on X axis, and the SSS RGB on Y one. Suzanne is about 13x9x8cm.

Interesting things seem to happen between 0 and 200, so, I'll make another test with a better scale.

DaveC · Post by **DaveC** » Thu May 24, 2007 8:23 am

Now THAT is right. That looks like a VERY useful chart.

Kram1032 · Post by **Kram1032** » Thu May 24, 2007 8:30 am

the best chat, I've seen, yet

0XX means, yes, red, no not green, no not blue?
and the Y is same as X?

Sebastian · Post by **Sebastian** » Thu May 24, 2007 8:34 am

Nice !! ^^

MegaResolution, Plz !!!! [/quote]

Stur · Post by **Stur** » Thu May 24, 2007 8:37 am

Thx guys

You need to replace the X and Y by the values on the array. So, Suzanne's instance on (x=101, y=201) has a material with (R=0, G=101, B=101) as absorpbtion value, and (R=201, G=201, B=0) as SSS value.

Edit : here is the medium of Suzanne's instance at (101, 201) :

Code: Select all

	<medium>
		<name>MyMedium</name>
		<precedence>10</precedence>
		<basic>
			<ior>1.3</ior>
			<cauchy_b_coeff>0</cauchy_b_coeff>
			<absorption_coefficient_spectrum>
				<rgb>
					<rgb>0 101 101</rgb>
				</rgb>
			</absorption_coefficient_spectrum>
			<subsurface_scattering>
				<scattering_coefficient_spectrum>
					<rgb>
						<rgb>201 201 0</rgb>
					</rgb>
				</scattering_coefficient_spectrum>
				<phase_function>
					<uniform/>
				</phase_function>
			</subsurface_scattering>
		</basic>
	</medium>

Kram1032 · Post by **Kram1032** » Thu May 24, 2007 8:39 am

oh, ok.

so, it is, as I said^^
I thought that last part...

I guess, I asked kinda confusing, as all my words get confusing, fast

Stur · Post by **Stur** » Thu May 24, 2007 9:08 am

@Kram : lol

@Sebastian : I wish I could render a big picture, but it would take ages on my computer and I want to go further with my tests instead of waiting days to have one good big render.
Marcofly proposed his help, so maybe ... ?

Anyway, here is the .igs with materials and mesh files, in case someone would like to render a big picture. Don't forget to edit the .igs as it is still at 800x600.

Marcofly · Post by **Marcofly** » Thu May 24, 2007 10:06 am

will try to do it tomorrow..

tinuvielle · Post by **tinuvielle** » Thu May 24, 2007 8:16 pm

Wow, that's a very useful chart indeed, and your script will allow each of us to test our settings.
Congratulations, and thank you for sharing your work with the community

Marcofly · Post by **Marcofly** » Thu May 24, 2007 10:55 pm

Hey stur, i'm rendering at 1024x768.. not a big pic , but it's cooking a bit more than yours. i am at 600 samples/pixel.. it shows already some more subtle details.

can you please setup a scene that includes only the 1-101 range and post the .igs? don't have time to learn today how the script works

thanks, bye!

alex22 · Post by **alex22** » Thu May 24, 2007 11:31 pm

You should do the table logarithmic, since there are big differences between 1-101 and very little between 801-901.

SSS Test Scene

Who is online