OctaneRender® 1.024 beta 2.47 (lin/mac/win) [OBSOLETE]

[FamilyGuy]

I did my own test with GTX 460 2GB, and divided them into three groups:

1. pmc 1000x500pix
2. pmc 3000x1500pix, noise reduction, 3x downscale
3. pathtracing

The light source:
- 50 downlights simulating IES (12 triangles each) very deadly for Unbiased rendering
- 6 big lamps with simple poly



This is what I get (you have to resize to full screen to see the difference)



And this is what I received in IndigoRender MLT after 4h (core i7-860 @ 3.9GHz)



My impression:

- in this specific rendering OCTAN is at least 8x faster than IndigoRender!!! (in my opinion Indigo have a very good interior engine)
- supersamling technics does not give any significant effect
- fantastic progress in interior rendering

[Refracty]

I haven't tested it, but I don't think you can get sun light through specular materials. You still have to use the architectural glass trick.

Ups, I thought this is an major reason for 2.5

[funk]

Well done guys! This is looking great!

I really like the fact that I can work in other applications while rendering!!!!! Most of us have only one card and path tracing makes it impossible to multitask while rendering (too sluggish)

[bepeg4d]

thanks octane team for this test version it's very promising
here is my mac test with 1.24.7 with a gtx 285.
it's an open scene and there are two big one poly plane emitters and a very little sphere over the door + an hdri at low intensity for some reflections on the metal, both are rendered to 16000
@ FamilyGuy: very interesting test comparison, thanks
in my opinion supersamling technics can be a good solution for speed up the things even with pmc
ciao beppe

[kubo]

well, now that the zombies have been fed, why not put them to use?
Same scene, 2M tris, one emmiter (duh), and hdri sky for the mood. Zombies are glass with a little roughness and 0.012 disperssion, no hotpixel reduction (=1)
PT was faster 16 min for 750 samples, PMC, althou it took 22 minutes, it resolved it much better.
Testing this is sooooo much fun.

[vinz]

Well, here some fast tests: i used my ocedar in a cube. used one IES light Plane. rendered with 1 gtx 460

there are two kind of comparison one by time anoher by sampling level.

nice work guys ! thank you so much for yur hard work

[vipvip]

" I've notice that when i use pmc i can use my pc normally, it's not slow, instead when i use pathracing i can't use my pc for other works "
-> +++

[vipvip]

just few tests with new PMC engine:
i know it's just the beta of the future beta but i obtain regulary this problem: a splitted screen with 2 diffrents results, fotr the same rendering:
( see attached ) .
Perhaps it's because i use 2 card simult. for rendering ( multigpu with 2 gtx470 ) ?

precision:
the picture is issued of the same calculation, with whql latest nvidia 275 drivers

[face]

Perhaps it's because i use 2 card simult. for rendering ( multigpu with 2 gtx470 ) ?

Looks like as two different cards.
The overall samples are 256, the left card make 192 and the right card 64 samples...

face

[pixelrush]

2.47 seems to work fine with Quadro driver 275.36...
this one is Cuda 4.1 according to the Nvidia notes but I think thats a typo.