Feature request: Network Render

[KyungSoo]

I have a plan to build a small render farm (8 nodes) with Octane Renders.
For that, network rendering is an essential feature.

If you have any plan for it, please let me know.

[Xenomorph]

- Integration in render-queue systems for animation/batch rendering**

** = Octane Render 1.0 feature

please refer to the feature page

http://www.refractivesoftware.com/features.html

[radiance]

Hi,

As Xenomorph said, you can use a free network queue system like DR queue to batch schedule animation renderjobs,
with the upcoming command-line versions included with our commercial releases.

Distributed / single frame rendering is not on our list for v1.0 currently, but we might add it in the near future.
It's not that easy to implement as octane is interactive, eg imagine what needs to happen when you change something,
or move the camera, you'll have to contact all the slaves, and update the data etc... so it could get really slow.

Also, the raw speed of octane makes a single frame render already as fast as a say 10 pc distributed single frame render with current CPU engines,
so it does'nt have that much value anymore for octane...

Radiance

[KyungSoo]

Thanks.

[havensole]

I am also working on a free network rendering app that "should" be easier to configure then Dr. Que, or some of the other options out there. Most of this depends on the options available to the CLI, so I "should" have something available shortly after the v1 release.

[cecofuli]

Hi radiance,

this is my first post here. I'm interesting about your software. Just to know. VRay-RT work very well in DR for a single image. Why Octane can't do the same?
In VRay-Rt the interactivity is very fast with 10+ rendernote.
Obviously, at the begin, if the scene is very heavy and we don't have a goog LAN, you must waiting some minutes, but after, you change shader, view etc... in a second very fluid.
So, IMO, a Single IMAGE Network Render is a PRIORITY. Speed is the future and with this feature you can obtain a big success

[Sam]

So, IMO, a Single IMAGE Network Render is a PRIORITY. Speed is the future and with this feature you can obtain a big success

It should be more easy to do than biased renderers
Because you just add all the samples of all the nodes into 1 render
There's some work to do on the network side of things
Making a production proof network renderer is not easy

Im sure Octane will support it, its just a matter of time
There's plenty other features that are missing, more important than that

[parva]

VRay-RT work very well in DR for a single image. Why Octane can't do the same?
In VRay-Rt the interactivity is very fast with 10+ rendernote.

Maybe I'm wrong but there VrayRT uses (currently) only CPU there is no real comparison with a GPU Renderer like Octane.

[cecofuli]

Yes, right now VRay-RT commercial release is only for CPU. But radiance write:

what needs to happen when you change something,
or move the camera, you'll have to contact all the slaves, and update the data etc... so it could get really slow.

But in VRay-RT distribute rendering (still frame) CPU based is very fast. Why not on Octane?

[radiance]

Yes, right now VRay-RT commercial release is only for CPU. But radiance write:

what needs to happen when you change something,
or move the camera, you'll have to contact all the slaves, and update the data etc... so it could get really slow.

But in VRay-RT distribute rendering (still frame) CPU based is very fast. Why not on Octane?

i don't understand why it's such a priority.
you can put the equivalent of 100 networked PCs in one machine with 2-4 GTX400's.

that beats having to write a real-time control network protocol that needs to coordinate 100 PC, sends commands, received images over the network and shows them.

it might be possible, but if we do that now, it will take us many months to code. I'd like to focus on more often asken features first.

Radiance

Radiance