OTOY Forums

Is this worth to get now or should I just wait for Octane integration?

Depends on what you need, but I can tell you my renders look a heck of a lot cleaner than they ever have before. I'm going to test the animation soon.

it works really well already, and even more so once we add a final AOV shortly. I would recommend getting it if you need denoising right now. Even with the module SDK in 3.1, there is still the issue that the Altus process requires a lot of memory and basically it's own GPUs (including two frames rendered separately from Octane), so work would have to done to make it meaningful in the live viewer workflow. As it is now, the simplest integration in 3.0x would be to use the new render job node to have a script to launch Altus as a separate process that is run as the raw AOVs are completed by the job node and saved into the right folder. The job node could then run Altus and generate denoised frames on its own GPU, export the back to the job node folder, while Octane is freed up or moves to the next frame/job segment. The RJ node can group multiple frames/rendertargets etc, using a normal base RT as input, so this could work pretty well if done right. Once we have the last AOV done, we'll figure out the best way to look at doing an RJ node to run Altus this way.

I am thinking about getting this a try, is there anywhere step by step some kind of tutorial how to use it - i mean it regard to Octane?

Next steps as:

1) Add the final AOV in Octane core for full Altus support (working on this now)

2) Release a sample ORBX scene with the needed render layers already set up, once (1) above is working.

3) Shortly after (2) is out, release (maybe built in to Octane) a render job script node that can create the two frames+needed AOVs from an existing render target (connected via an input pin), which can also launch Altus after all output is completed by the job node. This can further be used for animations, and easily hooked up in plug-ins that support script nodes (even invisibly). Some work needs to be done around memory usage and GPU sharing, but this could be exposed in the node settings as well.

4) Once (3) above is working for users locally, and is confirmed to be stable, we will then work an the ORC version of this node to automate denoising for cloud rendered images/animations in a way that matches the local workflow.

5) Work with Innobright on license tailored for Octane users that works across local/network/ORC jobs. TBD - input is very welcome on pricing and usage expectations different than what is offered right now.

It is awesome that this is finally getting worked on. Thanks!

Goldorak wrote:Next steps as:

1) Add the final AOV in Octane core for full Altus support (working on this now)

2) Release a sample ORBX scene with the needed render layers already set up, once (1) above is working.

3) Shortly after (2) is out, release (maybe built in to Octane) a render job script node that can create the two frames+needed AOVs from an existing render target (connected via an input pin), which can also launch Altus after all output is completed by the job node. This can further be used for animations, and easily hooked up in plug-ins that support script nodes (even invisibly). Some work needs to be done around memory usage and GPU sharing, but this could be exposed in the node settings as well.

4) Once (3) above is working for users locally, and is confirmed to be stable, we will then work an the ORC version of this node to automate denoising for cloud rendered images/animations in a way that matches the local workflow.

5) Work with Innobright on license tailored for Octane users that works across local/network/ORC jobs. TBD - input is very welcome on pricing and usage expectations different than what is offered right now.

But instead of going through all these steps for an external denoiser what is the status of your AS feature? It's Feb 2017 and in 2 months time will be 1 Year from the announcment of it..won't go through the pain of listing features that has been implemented elsewhere in less than 2 months..or less... Any news? 3.1?

And meanwhile over at the Corona ranch they have an Altus equivalent built into the frame buffer, no need for likely hack and buggy 3rd party implementation, with additional cost to users.

Just saying.

I also think Otoy adding it's own denoising to Octane would be the best, this in combination with adaptive sampling would be huuuuge, I can not think how much the render times are going to decrease with this two combine.

any news on this?? more than 1 year after some plans about denoise features, and 80% of my renders still unviable comercial noise .... 3.1 when?