OTOY Forums

Is it possible to let the mesh import work on all available CPU cores, so that the import time decreases?

As far as I can see there are 3 parts to Octane:

1) The Importer part using import scripts - these are done by the people making the scripts.Not sure about that.Jimstar has got the Max exporter version so that you can run it in Asynchronous mode, so that as soon as the first frame of an animated sequence is exported.It starts rendering that frame and then the others, whilst in the background the exporter continues to export the remaining frames.So this is a sort of Multi threading.Using it , it certainly removed a 20 minutes wait before the first frame of a 250 frame video started rendering for me.Even so this is still only using just over 2 cores.(1 for Octane to Render in the foreground and 1 for the exporter to keep processing .obj files in the background.)

2) Voxalization using the CPU - This just uses one of the CPU cores at present and could be multi threaded by the Team to cut down the time preparing the scene.

3) Rendering using the GPU - This uses all of the GPU cores so no where to go on that side of things.

It would be very interesting to extend the asynchronous mode (1), so that 2 instances of Octane will be open. One is importing the mesh while the other is already rendering. Of course that would mean that you will consume twice Ram, but for smaller projects it could speed up everything.

For my part, I think multi threading the voxalization process is the best option.

Taking 10 minutes to voxalize a large scene in Octane before rendering commences using just over 1 core , down to about 1.5 to 2 mins before it starts to render using 6 cores.

Of course if you do that then there will be no capacity to do something else in the foreground whilst Octane Renders up the scene in the background.(a sort of multi threading if you have something else you can do whilst waiting for the scene to process)

So what does octane power tools do?

I think it gets rid of parts 1 & 2 on my list and links the mesh, camera and materials directly from the modelling program to Octane.Like for example if Octane was plugged straight into the modelling program itself. A bit like VRAY or Mental Ray is within Max.

Obviously this would be the ultimate way to go for speed , if it could be achieved within future versions of Octane.

I think for that to happen the team would have to expose octane's program in some way , so that plug ins could be written directly within each modelling App to call Octane.

However there is of course other talk here about animation facilities directly from within Octane. We''ll just have to see how this all pan's out.

Time will tell !

There are 4 parts to Octane, export, import, voxelization and render...
Export is done for around 1.4mio tris at 23sec.
For the import/voxelizing, you must wait 67sec.

I think, there is some capability to speedup the import, without the use of multithreaded tasks...

face