I developed a tool to render in batch mode single or a sequence of images from one or more scenes, using one or many computers (using CPU or GPU).
It works only on Linux and with cycles.
You have to install the Postgres database (very simple in Linux)
You can ‘submit’ a task through an ‘add-on’.
The ‘slaves’ are shell clients that you launch and loop to procedd sub-tasks.
See the .ods file, Installation sheet for details.
choose temp and dest directory different from your current directory:
++ the submit part copy <blende file>_EXR.blend in the destination directory
++ a lot of files are generated in the temp directory.
It is the third beta revision, still a WIP.
The <blende file>_EXR.blend generated may be used for quick compositing, the render layers are replaced by EXR mutli-layers files.
To do list:
- deal with blender version parameter (so that the clients will have a warning if their blender version is older),
- deal with CPU/GPU limitations. You can launch ‘client’ with -d GPU option, but still lacks how to do when the submiter don’t want GPU renders and how to choose the graphic card
- deal when one client take too long the render a sub task,
- add token for clients (see if it’s necessary),
Updated 2014-09-10: Windows and OSX adaptations
> the bldclient.bash has been updated to bldclient.py, to be easier to adapt for windows and OSX. See WARNING in the .py files
If someone can test it in a single PC or network configuration, I will be glad to know if it is working (and how to parameter it for the network part).
CollaborativeRenderV1R4.zip (149 KB)