Rock Crystal New scene

Kram1032 · Post by **Kram1032** » Sat Feb 10, 2007 6:29 am

I started rendering a rock crystal.
It's lit by indirect caustics from a double-sided mesh emitter.
Just like in this winosi scene, with one difference: the emitter isn't inside a tube, but inside a half sphere.
It looks quite nice, but I completely failed the material of the crystal:
IOR set to 3, which is much too high, I know
caughy_b_coeff set to 0.5 at the moment. Is this ok, or should it be changed too?
It rendered for 1:17h right now, which is pretty short for my pc...

c&c plz

Post by **fused** » Sat Feb 10, 2007 8:14 am

looks strange

i think cauchy_b might be too high. but i dont know. the value for glass is 0.003

Kram1032 · Post by **Kram1032** » Sat Feb 10, 2007 8:33 am

I'm rendering a new version with IOR set to
1.544
and cbc to
0.009
It looks completely different now, but it's not a bit faster...
I'll show the image after one hour rendering (currently it's ~12 min)

Kram1032 · Post by **Kram1032** » Sat Feb 10, 2007 9:35 am

My crystal right now:

This is after ~1h 11min...
6327000 Muts so far...

c&c plz

Post by **fused** » Sat Feb 10, 2007 9:40 am

could you post a screen of the setup of your scene?

looks really really strange to me

Kram1032 · Post by **Kram1032** » Sat Feb 10, 2007 9:43 am

Ok...
Edit: here is the screenshot

The lightsource are two very tiny plains, nearly stacked together...
You can't see them at this screenshot

Kram1032 · Post by **Kram1032** » Sat Feb 10, 2007 10:48 am

Now, I know, why it looks like this -.-
~2h render
625000 Pixel
~12600000 Muts
12600000/625=20 -> 20 Muts/Pixel
20/2 = 10 -> 10 Muts/pixel/hour
-> Completely Mutated Img every 6 min -.-

CTZn · Post by **CTZn** » Sat Feb 10, 2007 2:41 pm

The lightsource are two very tiny planes, nearly stacked together...

Did you try with a single shape (cube or sphere) instead ?

Did you set an absorption for the medium ? If yes try to divide it by 10 or 100 or more if that doesn't change anything. At first it would be better not to use absorption till you fixed the issue.

Tried the linear tonemapping ? Are the gain and temperatures of lights the same ? If not Reinhard tonemapping could give strange results I think.

Optionally, how smal is <large_mutation_prob> set ? default is 0.4 (if you don't know don't bother with that question

).

Kram1032 · Post by **Kram1032** » Sat Feb 10, 2007 10:41 pm

1) No.
2) already is 0
3a) I think, linear is standard... Correct me, if I'm wrong.
3b) The light sources are 2 planes but they are combined in a single mesh.
4) Hmmm... That may is the problem. I've set it very low...
Thank you, I'll start rendering a completely new version, this time with large mutation prob set to standard.
Now, after 14h 18min and 49sec and 157,95840 Muts/Pixel, it's much better, but also very incomplete...
I'll post it and then stop rendering for another try.

Kram1032 · Post by **Kram1032** » Sat Feb 10, 2007 11:02 pm

Ok, here is my "final" result!
I'll start rendering again, with an increased value of large_Mutation_prob

(In my new image, the camera will be set a little bit different as well)

c&c plz

Kram1032 · Post by **Kram1032** » Sun Feb 11, 2007 12:17 am

Ok, this is much better now, thanks!
Here is my lastest result (after just 34 min)

c&c plz

CTZn · Post by **CTZn** » Sun Feb 11, 2007 12:36 am

Oh, ok

I was talking about large_mutation_prob as a side note only, it has just an impact on the nature of the noise, thus on the convergence, but I never tested if one setting was faster than another. But I know that extreme values (0 or 1) are not good.

Lower values will make a more "painterly" effect, while higher values will introduce a simpler, pixel level noise. I don't know for instance if the pixel threshold is at 1 or 0.5... I think that depends on the scene, and that 0.4 is a good default

Maybe you will want to put more lights in the scene, so the cristal can refract more light too... it won't look that dark then.

In the new image, what is odd to me is that there is no dispersion in the top of the cristal, I wonder why. Maybe that's because the camera position has changed, but that's really a radical difference

Oh, btw, have you set normal smoothing to false for the cristal ?

Kram1032 · Post by **Kram1032** » Sun Feb 11, 2007 4:13 am

This last image with low values has rendered for over 14 hours
al the other ones at least for one, but the last crystal just rendered for 30 minutes.
Now, the rainbow caustics at the bottom are nearly noise free and the tip of the crystal has more contrast to the background... So, there is a very weak caustics effect.
You already can see the walls. The contrast is quite weak, right now, but they are getting sharper...
Normal smoothing is on, although, it may should be off...
4h are already gone.
I'll add some light trough...
Here is the pic:

General question: Does anyone have a problem with seeing this pictures?
c&c plz

Kram1032 · Post by **Kram1032** » Sun Feb 11, 2007 5:20 am

Now, I've turned off normal smoothing and additionally, I've added some similar lights used as spotlights. One light source per corner of the room, except the corner, where the camera is set.
The default light is hanging a little higher.
The render looks much better than those before, already after 14 min

I'll post it, when it has rendered 30 min, to check with previous version and then again, when it has rendered 1h

c&c, when they are there

Edit: Here is the newest Version after 34 min

Edit: And here it's after 1h 20min...

Kram1032 · Post by **Kram1032** » Sun Feb 11, 2007 9:47 am

Ok, I think, from now on, I'll just update it.
Rendered for nearly 9h now.

Just for fun: Notice the smiley shaped caustic inside the middle light circle. (@ right side; It's upside down)

c&c plz

Rock Crystal New scene

Rock Crystal New scene

Who is online