OTOY Forums

Hi,

this animation doesn't look photo real, but maybe the calculation is quick done for an interior, so it could be an option. Which kernel mode could be used here?
https://www.youtube.com/watch?v=r7coFhFxM6A

Ciao,
Micha

It may be done with the Direct Lighting kernel (which is also available from the OctaneRender for Rhino plugin).

Paul

Thank you Paul, got it. The trick is to set the right AO distance. I take one of my old tests scenes and it works fine, the adjustment of the AO is quite easy with RT feedback. Quick rendered interior.

For the GI AO kernel I miss to get more control like at the dirt AO - strength and detail could be implemented.

Here a test with an adjusted HDRI. I think for a quick biased rendering it's a good result. The HDRI can be adjusted for faking GI effects on the walls.

General I would like to see more biased options at Octane. Unbiased rendering is nice, but often I havn't enough time during my projects. For example rendering a 10.000x5.000 interior pano needs approx. 1 hour per full biased Vray calculation. This or better is my goal I try to archive per Octane and a few graphic cards. I hope biased rendering will be more in focus of the development. Videos like the one from youtube shows, that people need speed, speed and speed.

Micha3D wrote:Here a test with an adjusted HDRI. I think for a quick biased rendering it's a good result. The HDRI can be adjusted for faking GI effects on the walls.

General I would like to see more biased options at Octane. Unbiased rendering is nice, but often I havn't enough time during my projects. For example rendering a 10.000x5.000 interior pano needs approx. 1 hour per full biased Vray calculation. This or better is my goal I try to archive per Octane and a few graphic cards. I hope biased rendering will be more in focus of the development. Videos like the one from youtube shows, that people need speed, speed and speed.

Orc.otoy.com

We have a monster 300 gpu render farm in the LA office and stopped using it for production once Orc launched. You want your animations 1000x faster, and are willing to pay for it, (~1$/GPU hr) the option is there. We are now rendering 18k 60 fps stereo cube map studio content for VR because of the cloud service. It will be built into octane and all plugins before 3.0.

Also, rendering to light field is a game changer too.

There are also ways to get quicker noise free renders through coherent mode settings.

The idea with your render farm sounds good, I will keep it my mind.

Interesting could be the provocative question, if a company sells render software and render power, will be there more motivation to enhance the speed or more too add complex calculated special effects.

Micha3D wrote:The idea with your render farm sounds good, I will keep it my mind.

Interesting could be the provocative question, if a company sells render software and render power, will be there more motivation to enhance the speed or more too add complex calculated special effects.

Octane on local GPUs will get faster with v3 - we are due for a solid speed boost with next gen NVIDIA/AMD GPUs in 2016.


The cloud service is intended to give you a force multiplier effect for harder jobs beyond static 1-2 MP renders. Think of render jobs that aren't likely to be possible in 24 hours without millions of $$ of GPUs - such as assigning 1 Titan X speed GPU for every frame in a 100-1000 of frames of animation, generating light fields, giant tiled renders etc. all within a few hours.

Specifically for animations, the time it takes Octane to render 1 frame on a Titan X GPU may potentially be same the time it takes to render all your frames on orc.otoy.com for a typical shot if capacity is there.

I'm curious for v3.

Since the users hardware isn't use 24/7 wouldn't be useful to create an user cloud computing with credits for power and power could bought back? So, my graphic cards could spend power 24/7 and if needed, I could get back the power within a short time.

That's actually an awesome idea micha3D!!!

Micha3D wrote:I'm curious for v3.

Since the users hardware isn't use 24/7 wouldn't be useful to create an user cloud computing with credits for power and power could bought back? So, my graphic cards could spend power 24/7 and if needed, I could get back the power within a short time.

We thought about this. Asset privacy and the ingestion of GBs of data (scenes are big) to users is an issue that pushed this out.

On the other hand, we now have thousands of GPUs on the cloud and a solid backend. We are about to launch a transparent way for Octane and host apps to use ORC, even while you edit your scene, as if it were a high latency network render node. I think all that will help