Does somebody here have experience with cycles distributed network rendering?

At Wayne state we try to figure out how and if it is even possible to render a single image over multiple cpus of different PCs. The science department has an interest for this.

in case somebody here knows this subject and would open to answer some question could you get in touch with me via [email protected]?

class

sheepit-renderfarm supports this, and it is open source. Maybe there’s something you can use from them? https://github.com/laurent-clouet/sheepit-client
It is Java though so hopefully that isn’t a problem.

You could also check this out: https://www.blendernation.com/2018/02/22/crowdrender-v0-1-3-adds-easier-faster-network-rendering/

This is how it can work in principle, there’s probably a more convenient solution out there.

You can split the image up using render border and crop, then stitch the result back together.

Render from the commandline and run a script (–python) that sets the relevant variables, for example, to render the left half of an image:

bpy.context.scene.render.use_border = True
bpy.context.scene.render.use_crop_to_border = True

bpy.context.scene.render.border_min_x = 0
bpy.context.scene.render.border_max_x = 0.5

bpy.context.scene.render.border_min_y = 0
bpy.context.scene.render.border_max_y = 1

For stitching, you can use imagemagick.

Note that this obviously won’t work with any compositing operation that requires access to neighboring pixels.

Unless you render a tile width/height (depending on where you split) extra in each, in which case it could.

You’re talking about a guard band, it needs to be as wide as the largest radius of all the neighboring-pixel operations, which may be less or more than one tile.

That’s of course possible, but then you need to also crop the resulting image differently from the setting in Blender.

Thread: Cycles Distributed Rendering
&
Cycles Network Rendering - Pigeon Build

The Grid currently has the combined processing power of 8,212 cores runs Rocks version 6.1 and CentOS 6. These resources are managed by Altair’s PBS Professional 12 job scheduler.

I log in via ssh and run a job script to submit work to the queue.

I have to ask the guys maintain this but to my understanding PBS launches Blender on each selected device and then quites it.
so one cannot Lauchen x amount of Blender and keep them in slave mode.

I have the feeling that just because of this blender cannot be used the same way as on renderframs that have blender as an engine installed. Or am I wrong?

Mods because there is already a distributed rendering here - maybe we can join the two. Threads ? My initial search didn’t don’t discover it.