Octane Render vs Luxrender

[Welti]

I did some test with Octane render and Luxrender. The Octane picture rendered nearly 9h, the luxrender image rendered nearly 10h (sorry i can't control it while sleeping). I don't see much of a difference between those two pictures, regardless to the colors. Both are using a Depth of 16. So i don't see the "1000%-5000% (10X to 50X) speed increase" like it is described at the homepage of refractive software.

LuxRender 0.9dev from 04.03.2012, rendered with a Core 2 Quad Q6600 (4 x 2,4 GHz) and Metropolis Light Transport with a Eye and Light Depth of 16
Octane Render 1.025 beta2.57, rendered with a GTX470 and PMC with a depth of 16

[gabrielefx]

if I render one teapot with iray, arion, vrayrt and octane the render time will be the same.

Try to render a complex scene filling 3gb of vram.

[Welti]

if I render one teapot with iray, arion, vrayrt and octane the render time will be the same.

You are talking about renders which are using the GPU, LuxRender uses only the CPU so it is a "typical un-biased, CPU based renderer." I am a bit disappointed now based on the render power of Octane...
A typical unbiased CPU renderer can achieve the same render result in the same time as a GPU unbiased renderer.

[matej]

This test is too synthetic (I could render such a picture with Yafaray photonmapping in 20min and have better quality).

To have a true comparison, it would be better and more interesting to use a real archviz project, as Gabriele said. The heavier on complex materials & geometry, the better.

[Welti]

I tried but it doesn't look that realistic...

[t_3]

I did some test with Octane render and Luxrender. The Octane picture rendered nearly 9h, the luxrender image rendered nearly 10h (sorry i can't control it while sleeping). I don't see much of a difference between those two pictures, regardless to the colors. Both are using a Depth of 16. So i don't see the "1000%-5000% (10X to 50X) speed increase" like it is described at the homepage of refractive software.

LuxRender 0.9dev from 04.03.2012, rendered with a Core 2 Quad Q6600 (4 x 2,4 GHz) and Metropolis Light Transport with a Eye and Light Depth of 16
Octane Render 1.025 beta2.57, rendered with a GTX470 and PMC with a depth of 16

if i interpret your words right: if you let them both render over night, what difference would you expect, regarding time or result?
i'd guess the octanerender image was already after 2hrs looking like it also looked after 9hrs - depending on your graphics card.

when i switched from lux to octane, it did replace me 4 i7 2600k @ 4ghz with one gtx 590 to achieve the same render time for the same quality with the same scenes...

[Welti]

what difference would you expect, regarding time or result?

I expected that the Octane Render picture would be clean and not so noisy. I though octane would be faster then a cpu based unbiased render..

[t_3]

in fact, it is; at least, if you compare a single q6600 with a single gtx 470.

[Welti]

My test shows that isn't right . Octane is faster at simple light situation but if there is much indirect lightning octane isn't really faster then a typical CPU unbiased renderer. Maybe you have to improve the algorithms for casted rays, don't know...I am just disappointed that octane is not 10x faster then a typical cpu unbiased renderer, maybe 1.5x faster...

[t_3]

My test shows that isn't right . Octane is faster at simple light situation but if there is much indirect lightning octane isn't really faster then a typical CPU unbiased renderer. Maybe you have to improve the algorithms for casted rays, don't know...I am just disappointed that octane is not 10x faster then a typical cpu unbiased renderer, maybe 1.5x faster...

don't see how to judge that from the images above. the luxrender image is dull, has no defined caustics, ... if the output would be the same, it would be possible to compare render times, but this way?

and you will of course be able to create specific situations that favors one engine above the other, but what gives? use it in a real world situation and see what it gets (for you ). this test imo doesn't prove that there is a general factor 1, 1.5, 10 or whatever

ps: i'd consider the scene above already a very simple light situation; maybe it's just the other way round? the more complex a scene, the more advantage for octane?