vred_render

VRED Renderer Job Bundle

Overview

This job bundle is for rendering VRED scenes using either VRED Core or VRED Pro in headless mode. It uses VRED's Python API through the VRED_RenderScript_DeadlineCloud.py script, which:

• Controls and executes the actual rendering process
• Converts job template parameters into VRED API calls
• Manages all rendering settings (quality, resolution, tiling, etc.)
• Handles file references and error management

Available Templates

This job bundle provides two template options:

• Basic Template (template.yaml): Standard rendering functionality
• Tiling Template: Additional support for region/tile-based rendering
  • Includes all features of basic template plus tile-based rendering capability
    • Adds a "Tile Assembly" step that combines rendered tiles into final images
  • To use this template, rename template_tiling.yaml to template.yaml

Requirements

For Service-Managed Fleets (SMF)

• Access to 'deadline-cloud' conda channel
• GPU-enabled worker
• (Optional) For tile assembly when using the template_tiling.yaml template, ImageMagick must be available in the rendering environment. The easiest way is to add the imagemagick conda package from the conda-forge channel to your Queue Environment. For detailed instructions on configuring conda packages, see the Configure jobs using queue environments.

For Custom-Managed Fleets (CMF)

• VRED Core or VRED Pro (2025.X/17.X or 2026.X/18.X)
• Windows or Linux operating system
• GPU with appropriate drivers
  • Please review the system requirements and software dependencies.
• Environment variable setup:
  • VREDCORE pointing to VRED Core executable, or
  • VREDPRO pointing to VRED Pro executable
• (Optional) If using the template_tiling.yaml template, ImageMagick must be available in the rendering environment for tile assembly.

Required Files

• A job template (yaml) file named template.yaml
• scripts/VRED_RenderScript_DeadlineCloud.py: a controller script that must be present in job bundle with the provided job template
• VRED scene file (.vpb) and/or its referenced files

Job Submission Examples

CLI Submission

Submit with default parameters:

deadline bundle submit .

Submit with custom parameters:

deadline bundle submit . \
  -p SceneFile="path/to/scene.vpb" \
  -p OutputDir="path/to/output" \
  -p ImageWidth=1920 \
  -p ImageHeight=1080

GUI Submission

Launch the GUI submitter to interactively configure parameters:

deadline bundle gui-submit .

Job Parameters

Input/Output

• SceneFile: VRED scene file (.vpb) to render (must be relative path)
• OutputDir: Directory for downloading rendered outputs (must be relative path)
• OutputFileNamePrefix: Prefix for output filenames (default: "output")
• OutputFormat: Image format (default: PNG, supports multiple formats including EXR, JPG, etc.)

Note: All PATH type parameters must use relative paths from the current working directory. Absolute paths are not supported.

Render Settings

• ImageWidth: Output width in pixels (default: 800)
• ImageHeight: Output height in pixels (default: 600)
• DPI: Resolution in dots per inch (default: 72)
• RenderQuality: Quality preset (Analytic Low/High, Realistic Low/High, Raytracing, NPR)
• View: Specific viewpoint/camera to render from
• GPURaytracing: Enable GPU-accelerated ray tracing

NVIDIA DLSS Options

• DLSSQuality: Deep Learning Super Sampling quality setting (Off, Performance, Balanced, Quality, Ultra Performance)
• SSQuality: Supersampling quality setting (Off, Low, Medium, High, Ultra High)

Frame Control Settings

• StartFrame: First frame to render (default: 0)
• EndFrame: Last frame to render (default: 20)
• FrameStep: Frame increment - e.g., 2 for rendering every second frame (default: 1)
• FramesPerTask: Number of consecutive frames to render in a single Task (default: 1) Batching frames can improve rendering efficiency by reducing overhead from task initialization. Example with FramesPerTask=5:
  • Task 1 renders frames 1-5
  • Task 2 renders frames 6-10
  • And so on...

Animation Settings

• RenderAnimation: Enable animation rendering (true/false)
• AnimationType: Animation type (Clip/Timeline)
• AnimationClip: Name of animation clip to render

(Optional) Region/Tile-based Rendering

This feature is available when using the tiling template (template_tiling.yaml). Tile-based rendering divides each frame into smaller tiles that are rendered independently and later assembled into the final image.

When using tile-based rendering, GPURaytracing must also be enabled to prevent solid black tile outputs.

• RegionRendering: Enable tile-based rendering (true/false)
• NumXTiles: Number of tiles to divide the image horizontally (default: 1)
• NumYTiles: Number of tiles to divide the image vertically (default: 1)

Process

1. Image is divided into tiles based on NumXTiles and NumYTiles
2. Each tile is rendered as a separate task
3. After all tiles are rendered, a final "Tile Assembly" step combines them into the final image

Expected Output

The renderer produces image files in the specified output directory following this naming pattern:

• Single images: <OutputFileNamePrefix>.<OutputFormat>
• Animation frames: <OutputFileNamePrefix>-#####.<OutputFormat>
• Region rendering temporary files (Individual tiles): <OutputFileNamePrefix>_YxX_AxB.<OutputFormat> Where:
  • Y: Vertical tile number
  • X: Horizontal tile number
  • A: Total number of horizontal tiles
  • B: Total number of vertical tiles

Notes

• When using conda environments, ensure your queue has access to the required conda channels (e.g., deadline-cloud, conda-forge)
• For optimal performance with large images, consider enabling region rendering
• GPU ray tracing requires compatible hardware
• DLSS features require NVIDIA RTX GPUs
• Tile rendering: Requires GPURaytracing to be enabled. Scene files must be properly configured (such as adding sufficient lighting) to support ray tracing, otherwise rendered images may appear black.