Multiple Octane Instances render the same frame Tweet

[zukan]

I'm trying to render with multiple instances of Blender simultaneously on the same computer to speedup rendering of an animation. Just as described here: viewtopic.php?f=9&t=49745

This works with Octane Standalone and other Octane plugins but not in Blender. When I start rendering using Blender, each one of the instances render the same frame instead of the different timeframes I've manually set.

[J.C]

Have you set the proper frame range on each instance? If it does not work maybe it's worth trying running blender in terminal?

[grimm]

You might try to run each instance on separate GPUs.

[zukan]

Yes I've tried all those things and it does not work, same frame rendered for each separate frame range. Does it work for you?

On a 40 frame animation I've split into 2 instances:
frame 1 is the same as frame 21
frame 2 is the same as frame 22
and so on...

[grimm]

The thread you liked to is talking about 3dsmax and c4d, it might be an issue with how the Blender plugin works. I'm not sure how you would run two Octane servers and point them to different cards at the same time. This would have to be something the server would understand and handle.

[zukan]

Could a dev please weigh in with some insight as to whether this will be a possible functionality in Blender or not? It is very vital for planning future workflows with this plugin.

Geometry loading times for animation is a huge bottleneck in the current version of Octane, whether something will render quickly or not can be taken care of with more GPU power but speeding up compilation times is a completely different beast to take on, as revealed through many experiments done by users here in the community.

I've done tests limiting my GPU's that normally run in PCIe 3.0 16x, all the way down to 4x speed and the upload process itself is negligible compared to the time it takes for the CPU to compile the scenes. Yet even people running top of the line i7's and Dual Xeon workstations have disappointing compilation times as explained here: viewtopic.php?f=9&t=52567

So the one quick and dirty method to get around this problem is to run multiple instances of Octane. This is a very vital workaround for animators who need multiple frames delivered quickly. Geometry optimizations can only take you so far, there are many situations where we still need to render big scenes that take very long to calculate. It really pains me to see that Blender of all plugins just cannot play along nicely with this important time-saving feature.