Network Rendering

The options in the Network Rendering rollout allow you to distribute a texture baking job over a render farm. Due to the way 3ds Max handles network texture baking and because Flatiron needs to synchronize data between the involved render stations a few special considerations are required for network rendering:

To use network rendering, the checkbox in the Network Rendering rollout must be checked and the hostname (or IP address) and port of a running Backburner manager must be provided. A network job is initiated just like local texture baking by pressing the Bake button in the Baking rollout. Flatiron will put a placeholder image to your selected render output directory and already assign it to the scene depending on your slot and shell settings. When the network rendering job is finished it will replace this temporary image file with the final baked image.

The Network output folder field must contain a network directory that is accessible for all computers that participate in the render job. If this directory is not accessible and mapped to the same network path for all computers they will store the required temporary output locally or not at all and the final render result will be corrupted. When the render job is initiated Flatiron creates a temporary subdirectory under the selected network directory which is used to collect and process the outputs from the separate render computers. This directory is automatically removed again after the render job has been finished and was processed completely by Flatiron.

Flatiron will monitor the render job and process the temporary data to generate the final baked map in the normal output directory (selected in the Baking rollout). That directory can be local and does not need to be accessible by other computers.

Additional details about Multi Pass baking

Flatiron has an internal render job tracking which is used to generate the correct render results in Multi Pass mode. When you are using Single Pass baking everything is handled through backburner by default and Flatiron’s job tracking and merge options that are described in the following paragraphs are not relevant.

Flatiron’s internal job tracking requires that the 3ds Max instance that initiated the network rendering is kept open while the job renders. If the Max instance is closed while a job is still running, the monitoring will be interrupted and Flatiron can no longer automatically process the temporary data to generate the final map. In this case the rendering job will continue even though Flatiron no longer monitors it and the processing of the data can be started at a later time manually: Make sure in the Backburner monitor that Flatiron’s render job is finished and then use the button Manual Merge which is provided in the Network Rendering rollout. Select the job.desc file for this render job (it is stored in the job’s temporary directory under the selected network output directory) and Flatiron will finish the map generation process and put the final result into your output directory. If Flatiron’s monitoring is not interrupted, the whole process will be done automatically and you don’t need to use the Manual Merge button to complete the job.

Warning: using Manual Merge with a netrender job that is not yet finished will result in a corrupted result. If you need to use it, always make sure that the render job is completed.

You can initiate multiple render jobs without having to wait for previous jobs to finish. The status information in the rollout will display only the most recently initiated job but Flatiron internally tracks and processes multiple parallel netrender jobs.