I wanted to share mitsuba result. 12 minutes with Photon Mapping and IC.
12 minutes! This is great! How was your experience with mitsuba?
I think there could be some problem with walls bump mapping (black dots).
Thanks!
Salvador Ureña
Sorry theoldghost, I didn’t notice your post.
Animation is an entire different world. It would be great to prepare an scene to test flicker, motion blur, and such.
By now I’m preparing a cornell box really hard to render engine, but I has not much time this week. I have test it and is almost impossible to get clean results without MLT or Bidir integrator. I want also prepare another for caustics but animation is also interesting.
Regards!
Salvador Ureña
Asticles, imo mitsuba is fastest and easiest to use from all free rendering engines. Eg for simple PTracing all you do is increase number of samples untill image is clear. No need to tweak number of sliders and waste your time to find optimal settings. Also somehow mitsuba gives nice photorealistic results out of the box. I wish I could copare it to arnold one day. Cons are: no realtime preview and material nodes yet.
Hi, I saw the picture is missing in my post about Octane render 9 Minutes, post it here again:
http://www.pasteall.org/pic/show.php?id=72629
Asticles, what do you like to have with the render_engine.blend, it is a very bad file for a raytracer. 
It need about 20 minutes to get it ~noise free.
Cheers, mib
EDIT: Cant add pictures, hm.
[ATTACH=CONFIG]314894[/ATTACH]
Thanks mib2berlin!
As the original scene is very easy to render to every engine, is not valid as a test. So I did a difficult cornell style scene to test convergence speed. Is interesting that only Thea did the render in unbiased method with almost no splotches. Well I don’t have vray. I want to close this post and start a new one.
Salvador Ureña