Indigo SSS material and some artifacts

[STS]

Hello Indigo users!

I'm quite new to Indigo renderer and I use Blender to model & export my scenes. I started a simple dice scene and thought to give those dices some SSS. I have Indigo 0.8 stable and I use Blender to Indigo exporter script. I had to edit the .igs file for the SSS effect. Here is the material for the glass of the dice:

Code: Select all

<medium> 
      <name>glass</name>         
      <basic> 
            <ior>1.5</ior> 
            <cauchy_b_coeff>0.0</cauchy_b_coeff> 
            <absorption_coefficient_spectrum> 
                  <rgb> 
                        <rgb>15 90 128</rgb> 
                  </rgb> 
            </absorption_coefficient_spectrum> 
            <subsurface_scattering> 
                  <scattering_coefficient_spectrum> 
                        <uniform> 
                              <value>15</value> 
                        </uniform> 
                  </scattering_coefficient_spectrum>          
                  <phase_function> 
                        <uniform/> 
                  </phase_function> 
            </subsurface_scattering> 
      </basic> 
</medium> 

<material>
      <name>Dice</name>
      <specular>
            <transparent>true</transparent>
            <internal_medium_name>glass</internal_medium_name>
      </specular>
</material>

I Find it strange that the render has really awful artifacts around the dice edges. Here is the result after over 12h rendering (on my AMD Athlon 64 3200+, 2GB Ram). Jaggedness of the edges haven't changed at all in this time.



This is the used dice mesh.


I would like to know that is it really taking so long to get nicer results or what is the problem? Also there seems to be no noticeable SSS effect. Is that material setting all wrong or what?

I have had successful renders with non transparent materials (this is my first try with that glass material).

Thanks..

[Kram1032]

definately looks like supersampling factor beyond 1
try supersampling factor 2 or 2
It's the antialiasing value.

[Stur]

Hi STS, and welcome !

It looks like if you had 2 surfaces overlaping each other no the first picture.
What did you hand edited exactly on the .igs file ? Maybe it's the key to what is wrong

[manitwo]

definately looks like supersampling factor beyond 1
try supersampling factor 2 or 2
It's the antialiasing value.

please tell me that was a joke

@STS: really strange - never saw such artefacts with indigo. does it render correctly if you use a diffuse for example? is the scale of the dice correct?

[OnoSendai]

Looks like some kind of displacement map error. Are you using a displacement map?

[Whaat]

I have had this problem, too. But, I am not sure of the cause. I don't believe it has anything to do with material settings or super sample settings.
This problem has arose for me when I was rendering an object using sunlight and the background is very bright. It might also be because the the scale of the model is too large. I think the problem is magnified when the model scale is huge.
Try rendering the same scene with meshlights and a different background to test my theory.

[Kram1032]

Or simply make the model smaller?
If it's

magnified when the model scale is huge.

It could be a floatingpoint error...

But I don't think, these cubes are, like, 10000 m (= 10000 BU) big, lol

Otherwise, the SS and the absorbtion both would be much too much...

Someone once posted simmilarish problems, where it was supersampling, if I remember correctly...

[psor]

C'mon guys, ... you're just testing the new realtime preview!


take care
psor

[Kram1032]

hum....
if that#S the preview, I'd actually like to have less biasedness instead of 20x20 pixels.

[STS]

Hmm.. actually those dice are quite small.. like 0.03 blender units (i think it translates to 3 centimeters, which is rather big for a dice thou). Also I noticed that without thinking I had subdivided those way too much (erhm.. actually those dice have 81000 faces!). I don't know if that causes those strange errors.

No I don't use displacement maps and increasing supersampling doesn't help.

I have uploaded the .igs file here if someone want's to look at it:
http://uploader.polorix.net//files/393/Jagged_dice.zip

I also tried other exporter for blender and didn't subdivide those dice so much.. now it's coming out nicely.. this is what I got only after 15 - 20 minutes:

[CTZn]

Beta adaptative sampling issue I mean have you never talked about something similar, Ono ? Or maybe that's me...

[STS]

Beta adaptative sampling issue I mean have you never talked about something similar, Ono ? Or maybe that's me...

What's that? *going to look it up*

hmmm, I did'nt know indigo ran on a commodore 64 ???

Maybe not on C64 but surely on new commodores..

[CTZn]

Beta adaptative sampling issue I mean have you never talked about something similar, Ono ? Or maybe that's me...

What's that? *going to look it up*

I think I've read Ono telling he ehanced the edge anti-aliasing code but I'm not certain of that...

[cpfresh]

I've had similar results when rendering a structure on a large terrain model. Once I remove the terrain mesh, the problem goes away. Now I haven't tested to see whether it is the texture I've applied to the terrain mesh or the mesh itself, but I assume it may be either one of those things causing the problem.

edit:
Well i did some more experimenting with removing all the textures from the large mesh (~17,000 polygons) and still got the strange artifacts in the rendering. Then I went ahead and removed the mesh all together but it was still producing the same effect. So now I don't know what could be wrong, I guess my file is goofed some how, or maybe my modeling software! (SU 6) Dunno.