Important: Ensure all computers involved in rendering are set not to hibernate automatically.

Prepare in advance

1. Choose the packaged resource in File > External Data, configure the rendering settings (engine, sample rate, etc.), and save.
2. Store the blend file in a shared directory accessible by all workers.
3. Make sure you download the latest manager and worker files
   

Ready to use

1. Install the renderworkshop_manager.zip plugin in blender on the computer you are working on, and enable renderworkshop.
   
2. Unzip renderworkshop_worker_xxx.zip on the computer where the rendering task is to be performed and run the executable, please note that your operating system needs to be compatible with the system of the worker file.
   
3. Open the blend file which need to render.
4. Open the N panel and select renderworkshop.
5. Click start server to enable the manager service.
   
6. On worker computers, configure the settings in the config file (config.json) and launch the worker. For GPU rendering, adjust "Cycles rendering devices" in Blender's Edit > Preferences > System.
   
   
   • server_ip: The manager's IP address.
   • server_port: The port used by both manager and workers.
   • blender_path: The absolute path to the Blender executable, preferably matching the manager's version.
   • retry_second: Setting the timeout for a worker to find a manager
   • cycles_both_CPU: Cycles renderer can be set up to use both CPU and GPU for rendering
     
     
7. The manager's parameter list will display connected workers.
8. In the worker list, fill the content. After setting the worker content correctly, click the hollow circle button in front of the list. The manager will automatically check the correctness of the content with the worker. If there is an error, please check the prompt message. If it passes the test, the worker will not be editable, the button in front will become a solid circle, and this worker will participate in the rendering task. If you need to modify the content, please click the button in front of the list again.
   
   
   1. specify the network path to access the blend file.The location of the blend file on the target computer. For example, the file that needs to be rendered on 192.168.20.112 is at X:\Render\test.blend
   2. If you are using Cycles rendering, set the device to the Cycles rendering devices set by your worker.（The default is Auto, which will automatically detect the GPU and system of the target computer for judgment. If the judgment is wrong, please manually set it to the correct option.）
      
   3. If the worker is Mac OS, you can set the GPU backend to metal. GPU backend support for Vulkan will be added when official Vulkan support is available.（The default is Auto, which will automatically detect the GPU and system of the target computer for judgment. If the judgment is wrong, please manually set it to the correct option.）
   4. The number of threads is set to 0 to automatically adjust the rendering threads according to the performance, the maximum is 1024.it is recommended that you do not change this setting.
      
      
9. Refresh the scene list, select the scene to render, and configure the parameters:
   • For single images, set the frame and the number of tiles (recommended: 2-20)(Larger image sizes and a larger number of workers can be set to a larger number of slices).
   • For animations, set the start and end frames and Split (the number of frames per task.It is not recommended that the split frame rate be set too small)
10. Click render. Images will be saved in the blend file directory under the scene name. Animations will be saved in a folder named after the scene within the blend file directory.

Update Log

V0.02    2024.10.24

1. Support merge image with AgX , Filmic, Filmic log and Khronos PBR Neutral look
2. Support check workerlist
3. Disable render setting when rendering
   
4. Fix Cycles engine render bug

1. Support automatic identification of Cycles Device and GPUBackend
   
2. The worker is no longer allowed to accept additional data while it is running.
3. Add md5 check to blend file
4. improve the server message queue

1. Info adds a new progress bar
2. Rewrite the workerlist inspection process
3. merge_image uses the copied new temporary scene and no longer takes up the original scene synthesis
4. recv_data adds timeout feedback
5. Optimize the timeliness of info feedback

1. Supports heartbeat detection to ensure that tasks can still be assigned to other online workers even after the worker goes offline.
2. Support Cycles render both CPU+GPU
   
3. Adjust worker inspection method
4. Refresh the list to support saving original data
5. Fixed a large number of worker problems
6. Fix bugs in scene batch rendering
7. Fix tiff format problem
8. Fix progress bar for checking re-rendering loss
9. Some detailed adjustments and improvements

1. Now when rendering animation, the tasks of all scenes will be integrated into one task. There is no need to wait for the task of one scene to be completed before starting the next rendering, which is more efficient.
2. Remove the restrictions on compositing, and now rendering can enable compositing.
3. The worker can now display the local IP
4. Optimize some code

V0.08 2024.11.21

1. Supports the cancel rendering function
2. Support automatic detection of blend file path
3. Support automatic checking when connecting to workers
4. Support displaying host name
5. Support displaying worker rendering status
6. Support customization of heartbeat detection time and maximum number of retries in preferences
7. Numbers will be automatically added to the rendered image and the original file will no longer be directly overwritten.
8. It also supports displaying the progress bar of each project during rendering.
9. Use a separate folder for temporary files of rendered images
10. Fixed the problem of incorrectly ending the rendering state in rendered images
11. Fixed the problem of premature ending when merging pictures
12. Fixed the problem of skipping a certain scene when rendering pictures in multiple scenes
13. Fixed the problem that caused parsing errors when the worker received multiple messages at the same time.