C4D's Net Render?

[havensole]

Radiance has said often in the past that the distributed rendering (on a frame by frame basis) should be able to be done through a program like Dr.Que. I've been trying to write a distributed render que engine for Blender, but have been too side tracked lately. I would imagine that as long as c4d can be called via a command line this should be possible. Basically you would need a script to open the c4d file, launch the export script, send the files to the next node on the list (or tell that node to access a shared folder), go to the next frame, repeat. This is of course a very over simplified explanation.

[tangent]

C4D's native render just can't make things look real.

Is this with GI -- AR before R12, or anything but R12 Prime now -- or the base rendering engine? It's not fair to compare Octane to C4D on realism otherwise.

R12 constitutes a big leap forward in "realism by default" with the new linear light model, too.

The only problem is that if I can't work it into a studio pipeline, that's a deal-breaker.

The only external renderer that NET Render knows how to talk to are RenderMan compatible ones, if you have a version of C4D with the CineMan feature. (That's only in R11.x with AR, R12 Visualize, or R12 Studio.) Renderman compatibility is on the Octane wish list, but so are a lot of other things. The current plugin based solution for exporting scenes to Octane isn't NET Render compatible.

The new Python support in R12 might be enough to help you roll your own NET Render alternative. The feature really isn't all that great. Most of the value of this feature is just the render node licenses, not the code that manages the distributed render. I think a lot of users end up replacing it, either with hand-rolled code or a third-party render farm management package.

I know there are multi-GPU processing options out there like the NVidia Tesla

This has approximately nothing to do with your stated problem.

You can only get about 6 Tesla cards into a single PC, and to achieve even that, you'll probably have to upgrade the power supply. A 6-card Tesla box with Octane isn't going to replace a 100-core render farm. Unless you can get equivalent power to your current farm in a single box, you still have to solve the farm management problem first.

[havensole]

As far as the Tesla bit of your answer, He could add a few Cubix xpanders along the farm. even the new rack mount ones, which would solve the power issue.

Best thing I think would be to hire a programmer for a few months (check universities and get some student work. works out as they're cheap and would love the experience) if you don't have one on staff and give them the problem. I don't imagine it would be a terrible problem to handle, but having someone paid to do it usually makes it happen a lot quicker. Once Octane gets RIB support this should be a lot easier to work out, versus the current exporting of every frame. Even the upcoming Collada support will probably make this easier to do. At that point you are just tell each node what frame to render as you only, should have to, export once.