Distributed Rendering and Network Rendering are different methods of sending render tasks to multiple render server machines. They both have their advantages and can be more efficient with certain tasks and in certain situations. This article aims to provide information on some of those strengths.
The answer to whether to use Distributed Rendering or Network Rendering really depends on the projects, expected results, and personal preferences. You are welcome to find more information on DR in the table below.
| Network Rendering |
Distributed / Swarm rendering |
|
| Rendering Stage |
|
|
| Efficient for |
|
|
| Control |
Full control:
|
Partial control:
|
| Job Resubmission |
|
|
| Automatic transfer of missing assets |
|
|
| Software Configuration |
|
|
| Operating System Support (Linux, macOS, Windows) |
|
|
| Supported V-Ray versions |
|
|
| Inaccuracy in individual machines |
|
|
* Distributed / Swarm (Clarify)
Rendering Stage
Network Rendering is most efficient for the final rendering stage where no further project changes are required.
Distributed Rendering is suitable for both final rendering and design stage rendering. During the project development stage, the artist can speed up the interactive rendering process by harnessing the resources of other machines in addition to the power of the artist's workstation (render-host).
Efficient for
Network Rendering: jobs can be rendered sequentially or parallelly, which makes it more efficient when rendering images and animations.
Distributed Rendering is very efficient for rendering large-resolution still images with more than one machine which significantly decreases the rendering time. It should be noted that the project information is sent separately to each node, and the loading time varies according to project complexity. In some cases with larger projects, the main machine may finish the render, before the worker machines have started. This can also be observed when the render time of the project is very short, for example, less than 1 min.
Control / Job Resubmission
Network Rendering offers control over rendering multiple jobs sequentially or parallelly and also setting a job prioritization. In case of an accidental error, the job is automatically resubmitted.
Distributed Rendering on the other hand allows selecting machines manually or automatically.
It offers to set CPU utilization and add or remove render servers during the rendering process. In case of accidental error, the job needs to be manually resubmitted.
Missing assets / Inaccuracy in individual machines
Network Rendering: does not support transferring missing assets. Inaccuracy in individual machines will not affect the rendering of other machines.
Distributed Rendering: on the other hand can transfer missing assets from the render host to render servers automatically. Inaccuracy in individual machines will affect the final rendering. Crash in the render-host machine leads to losing the information from all machines.
Software Configuration
Network Rendering: there are many render management software on the market, and each of them requires a different software configuration.
Distributed Rendering offers native support and does not require additional software configuration. Host Application (excl. 3ds Max) is not required for executing Distributed Rendering jobs.
Operating System Support
Network Rendering: the Operating System depends on the Host Application support.
Distributed Rendering supports any Operating System; Linux, macOS, and Windows, via V-Ray Standalone. Jobs can also be submitted to machines with different Operating Systems.
Supported V-Ray versions
Network Rendering is supported by third-party software, and Chaos cannot guarantee it supports the latest V-Ray integrations.
Distributed Rendering is developed by Chaos and it’s always up to date with any V-Ray integrations.
Related topics:
Further information for Distributed / Swarm rendering can be found on docs.chaos.com page for each host application. For example: 3ds Max | SketchUp
See also: Distributed Rendering Troubleshooting Guide