I hope its the right section to post this (or in the off-topic)
I just wanna speak a little bit about
For those you don’t know what’s Octane render, its a physically based renderer (like Maxwell render or Fryrender) but its running on the GPU (that mean its 10 to 20 time faster depending on the GPU you use) Im not gonna speak a lot about Octane render, visit the website if you need more informations
I will speak about Blender now
Octane render open .obj files, and Blender can export those files.
We got some good renders by Blender users already in the forum !
I will just post one example here, its made in Blender 2.49 and Octane 0.81a
Click for 1000*500 size
Image by Tommy5 from Octane forums / Image submitted for the Octane January License competition
All rights Tomislav Corak / Blender 2.49 + Octane 0.81a / 166’661 Triangles / 2332 Samples per pixel
52,44 min using 8600GT @ 256 MB / 1280640 original size / resized to 1000500 for forums
I hope to see more Blender users in Octane forums, and maybe someday an exporter for Blender
Nice pic. Octane render had neat diffuse/spec etc nodal system. That seemed really nice. We have to mention the competition thou ;D
LuxRender is also unbiased and uses realworld physic calculations to calculate lightning and reflection. It is also barking on the GPU render tree. Ontop of that they are throwing in the ability to do network renders w/ GPUs.
And worth mentioning LuxRender is doing it OpenCL so it’s platform independent. AMD/ATi/nVidia/Intel doesn’t matter.
while octane is written in CUDA and only compatible with nVIDIA hardware.
Octane gonna support OpenCL but currently OpenCL is not mature enough
That mean Refractive Software can’t code some features using only OpenCL
Luxrender guys should seriously consider CUDA as its far better for now
Also CUDA gonna evolve faster than OpenCL because its already more used
Nvidia buyed Mental Images (mentalray) in 2008, so CUDA is evolving really fast
Its like OpenGL vs DirectX
OpenGL is better but nobody use it
Luxrender on GPU is far from done, the current CPU is great but they need to hurry up
Octane is available for download right now for Windows and Linux, and beta coming this month
Ontop of that they are throwing in the ability to do network renders w/ GPUs
Same thing for Octane, network render on 1 frame by all GPU or 1 frame per workstation or 1 frame per GPU per workstation
This picture has been rendered with a 8600 GT wich is SLOW (only 32 core)
A GTX260 render the same picture in 6min instead of 52min
8600 GT he used
32 cuda core and 113 GFLOPS = x1 = 52 min render time
8800 GT
112 cuda cores and 504 GFLOPS = x4.46 = 11.65 min render time
GTX260
240 cuda cores and 874 GFLOPS = x7.73 = 6.72 min render time
Some back of the envelope calculations: In order to render this scene in .025 seconds would require 235 TFLOPS of power. The GTX 295 offers 1.788 TFLOPS right now. Applying a little bit of Moore’s Law says that some time in the 2020s we will be running games rendered like this at 40 FPS. With some optimization I bet it might start before the end of the decade!
Sheap excuse and proven wrong also by the well known LuxRender which uses OpenCl in it’s latest (beta). And you don’t even have to pay for LuxRender nor buy a NVidia card.
Sheap excuse and proven wrong also by the well known LuxRender which uses OpenCl in it’s latest (beta). And you don’t even have to pay for LuxRender nor buy a NVidia card
Im sorry but Mental Images choosed CUDA for iRay, and there’s a good answer to that
OpenCL is great but its not mature enough, and I don’t speak about simple things like pathtracing, but I speak about documented API, strong support from Nvidia etc…
All Images by Pedrojafet from Octane forums / Image submitted for the Octane January License competition / All rights Pedrojafet / Octane 0.81a / 567.767 Triangles / 4096 Samples per pixel / 744.4MB of Texture Maps / 23, 26, 18 min using GTX275 @ 1.7 GB
1280530 original size / resized to 1000414
OpenCL does not mean the end of CUDA. CUDA is a proprietary API that allows NVIDIA to update it quickly and frequently to add new GPU features without unveil the architecture of NVIDIA new GPUs. So CUDA will be always ahead of OpenCL in terms of features.
That’s one answer…
As I said on Octane forum, I felt in love with Octane !
I’ve just ordered the upcoming Beta version.
I’m already an Indigo licence owner, but I never really rendered something good. Even if the latest exproter version is more handy,and even with the recent standalone material editor, I’m still very clumsy with Indigo. In fact, I have never been able to do something good with an external render before !
On the opposite, I’ve been able to setup a scene in Octane in very few time, despite the bugs of the early demo release and my weak graphic card (Geforce 8800 GTS 512).
Some examples below (also some images of the living room rendered with indigo for comparison) :
