Octane Standalone vs Blender Plug-in Render Speeds Tweet

[jerzy]

I’ve noticed that when doing final animated renders using the official blender plug-in render times are significantly slower than when rendering as an exported orbx file within Octane Standalone. (6-8sec vs 30-40sec)

Running OSX 10.11.6, Blender 3.08 - 15.1, 1x GTX 980 ti, 2x 980s, 4x 580s.

Is this typical for most users?

[DrawFun]

Hi jerzy,

Thanks for your post.

It takes extra time for blender to resolve scene and transfer data to octane server, which is done in exporting if you render in StandAlone with an exported orbx.
We will keep optimizing the transferring process in coming releases(e.g. get rid of the invisible objects in 15.1 and 15.2).
It will be very helpful if you can upload(or PM me) a demo scene. Thanks.

Cheers,
DrawFun

[J.C]

I’ve noticed that when doing final animated renders using the official blender plug-in render times are significantly slower than when rendering as an exported orbx file within Octane Standalone. (6-8sec vs 30-40sec)

Running OSX 10.11.6, Blender 3.08 - 15.1, 1x GTX 980 ti, 2x 980s, 4x 580s.

Is this typical for most users?

Jerzy, what's the compilation time in standalone and in Blender? What's the render time after compilation?

[jerzy]

Thanks for the replies,

Drawfun,
Unfortunately I can't upload the original file because of nondisclosure aggrements.

J.C.,
After some tests it does seem that Blender is compiling the scene for every single frame sent to the octane server.

In standalone the first frame of the file takes takes a total of 25 seconds. Around 12s to compile then 13s to render. Each following frame is between 7-8s with no compiling.

In Blender the first frame takes approximately 22 seconds to compile and 18s to render (40sec Total). Each following frame takes 17-18 seconds to render with around 10-12 seconds to compile (28-30sec Total)

The 980ti is the only internal card. All others are connected via an Amfeltec GPU cluster. I decided to test just the internal card vs all connected cards and came up with the following averages for a 10 frame render.

Only 980 ti, Standalone 30s, Blender 42s
All GPUs, Standalone 8s, Blender 30s.

So when using all the cards in blender your only seeing around a 29% increase in speed. In standalone its closer to 4x the speed.

[J.C]

To reduce compilation time you need to set the proper type of objects and animation mode. Movable proxy is the best for animated objects while scattered and global for non moving ones. Bear in mind that if you have any objects with modifiers Octane will need to Apply them every frame that can take lots of time if there are CPU intensive ones.

More detailed information is here:
https://docs.otoy.com/BlenderH/BlenderP ... yTypes.htm