AWS Deadline Cloud for VRED

AWS Deadline Cloud for VRED is a Python-based package that supports creating and running Autodesk VRED render jobs within AWS Deadline Cloud. It provides a user-friendly VRED submitter plug-in for your Windows-based workstation. You can choose from a set of common render options and offload the computation of your rendering workloads to AWS Deadline Cloud to reduce the load on local compute resources to pursue other tasks.

Requirements

The AWS Deadline Cloud for VRED package requires:

1. VRED Pro or VRED Core 2025/2026 and its requirements
   • Note: at the time of writing, it is strongly recommended to use NVIDIA GPU driver 553.xx (until future driver levels pass hardware qualifications for VRED.)
2. Python 3.11 or higher;
3. Operating System support:
   • Windows 10+ (for the in-app VRED Submitter plugin, Worker, Standalone AWS Deadline Cloud client submitter)
   • Linux (for the Worker, Standalone AWS Deadline Cloud client Submitter)
   • macOS (for the Standalone AWS Deadline Cloud client Submitter)
4. Optionally: ImageMagick static binary (to support tile assembly when region rendering with raytracing is applied).

Important: This integration of AWS Deadline Cloud into VRED requires bring your own licensing (BYOL) for VRED. You must have valid VRED licenses available for your render farm fleet.

Submitter

This VRED integration for AWS Deadline Cloud provides an in-app Submitter plug-in. It must be installed on the Windows workstation that you will use to submit render jobs from within VRED Pro.

Before submitting any large, complex, or otherwise compute-heavy VRED render jobs to your farm, we strongly recommend that you construct a simple test scene file that can be rendered quickly. You can then submit renders of that scene to your render farm fleet to ensure that its setup is correctly functioning.

Getting Started

If you have installed the submitter using the Deadline Cloud submitter installer you can follow the guide to Setup Deadline Cloud submitters for the manual steps needed after installation.

If you are setting up the submitter for a developer workflow or manual installation you can follow the instructions in the DEVELOPMENT file.

VRED Submitter Plug-in

The VRED Submitter plug-in creates a menu (Deadline Cloud) and menu item (Submit to Deadline Cloud) in VRED's menu bar, which can be used to submit render jobs to AWS Deadline Cloud. This menu item launches a Submitter UI to create a job submission for AWS Deadline Cloud using the AWS Deadline Cloud client library. It automatically determines the files required for submission based on the loaded scene file (includes Source/Smart references as dependencies.) Additionally, the Submitter provides basic render options (in the Job-specific settings tab), and builds an Open Job Description template that defines the render pipeline workflow. From there, the Submitter submits the render job to the render farm queue and fleet of your choice.

Launching the Submitter

1. Open VRED.

2. Open a VRED scene file.

3. Open the Deadline Cloud menu and click the Submit to Deadline Cloud menu item to launch the Submitter.

   Note: If you have not already authenticated with Deadline Cloud, then the Authentication Status section at the bottom of the Submitter will show NEEDS_LOGIN.

   a) Click the Login button. Then, in the web browser window that appears, authenticate using your IAM user credentials.

   b) Click the Allow button (your login will then be authenticated and the Authentication Status section will show AUTHENTICATED).

4. In the Submit to AWS Deadline Cloud dialog, configure appropriate settings (including the render settings that are listed in the Job-specific settings tab).

5. Click the Submit button to submit your render job to AWS Deadline Cloud.

VRED Software Availability in AWS Deadline Cloud Service Managed Fleets

Please ensure that the version of VRED that you want to run is also available on the fleet (worker nodes) when you are using AWS Deadline Cloud's Service Managed Fleets to run render jobs; these fleet hosts do not have any rendering applications pre-installed. The standard way of installing applications is described in the service documentation.

Licensing Requirements: VRED requires valid BYOL licenses on all worker nodes. You must configure your license server to be accessible from your render farm, ensuring that licenses are available for concurrent rendering tasks.

Submitter Interfaces

The VRED Submitter provides multiple interfaces for configuring render jobs. These interfaces will vary slightly since the values for certain settings (including animation clips and viewpoints/cameras) are only known during runtime (within VRED).

Submitter GUI (within VRED)

Settings applied in this interface will persist for the entire duration that a given scene file is open (this includes closing and reopening the Submitter window). Initial settings in the Submitter UI are populated from VRED's general render options (which are saved in the scene file). Saving the scene file won't persist the Submitter UI options, but certain options (including frame range to render) will be repopulated if saved in VRED's general render settings.

Once a job has been submitted, a local Deadline settings file will be saved next to the submitted scene file with the file name <SceneName>.deadline_render_settings.json. This file defines some submission options that will be reloaded for this scene the next time you open the submitter window.

Standalone Submitter (external to VRED)

This interface can be accessed via command line and offers most of the options presented in the Submitter GUI.

Description of Interface Options

Below are descriptions of the options available in both of the above interfaces.

Render Options

• Output Directory (Render Output): Directory where rendered images will be saved
• Output Filename (Render Output): Base name for output files (without filename extension)
• Output File Format (Render Output): Image format (PNG, EXR, JPEG, TIFF, etc.)
• Render Viewpoint/Camera: Name of the viewpoint or camera from which to render
• Image Size Presets: The available presets for image size and resolution (HD 1080, 4K, etc.)
• Image Size: Width and height in pixels; can be set via Image Size Presets setting (HD 1080, 4K, etc.)
• Printing Size: The printing size in centimeters (width and height), influences image size.
• Resolution (DPI): Dots-per-inch scaling factor, influences image size.
• Render Quality: Quality level (Analytic Low/High, Realistic Low/High, Raytracing, Non-Photorealistic (NPR))
• DLSS Quality: Deep Learning Super Sampling quality (Off, Performance, Balanced, Quality, Ultra Performance)
• SS Quality: Super Sampling quality (Off, Low, Medium, High, Ultra High); overridden by DLSS quality setting
• Use GPU Ray Tracing: Enable GPU-accelerated raytracing (if hardware supports it)

Animation Options

• Render Animation: If checked, allows specifying an animation type (which can also include a specific animation clip), corresponding frame range and frames per task.
• Animation Type: The type of animation (Clip or Timeline).
• Animation Clip: The name of the animation clip to render
• Frame Range (Start Frame, End Frame, Frame Step): Start and end frames with optional step size. (format: 'a', 'a-b', or 'a-bxn', where 'a' is the start frame, 'b' is the end frame, and 'n' is the frame step). Negative frames are supported.
• Frames Per Task: The number of frames that will be rendered at a time for each task within a render job.

Tiling Settings

• Enable Region Rendering: When enabled, the rendered output image will be divided into multiple tiles (subregions) that are first rendered as separate tasks for a given frame. These tiles will then be assembled (combined) into one output image for a given frame (in a separate step).
• Tiles in X/Y: Number of horizontal/vertical tiles to divide the specified rendered output image for a given frame

Note: Region rendering (tiling) is designed for scene files that have been configured for Raytracing. When you enable "Enable Region Rendering", the "Use GPU Ray Tracing" option will be automatically enabled to ensure proper tile rendering. Applying region rendering to scene files that aren't configured for Raytracing may result in solid black-rendered output.

How to use Tile Rendering with Service Managed Fleet (SMF): To use tile rendering with SMF, add the imagemagick conda package from the conda-forge channel to your Queue Environment. This provides the ImageMagick binary needed for tile assembly. For detailed instructions on configuring conda packages, see Configure jobs using queue environments. Once configured, simply enable "Enable Region Rendering" in the submitter ("Use GPU Ray Tracing" will be automatically enabled) and submit your job.

Invoking the Standalone Submitter

The standalone submitter is invoked via command line and relies on the following environment variables, which you can substitute with appropriate values:

# Windows
set FARM_ID=farm-<farm-id>
set QUEUE_ID=queue-<queue-id>
set FLEET_ID=fleet-<fleet-id>
deadline bundle gui-submit .

# Linux
export FARM_ID=farm-<farm-id>
export QUEUE_ID=queue-<queue-id>
export FLEET_ID=fleet-<fleet-id>
deadline bundle gui-submit .

Viewing/Submitting a Job Bundle

Before submitting a render job, the Submitter first generates a Job Bundle, and then relies on the AWS Deadline Cloud Client package to submit that Job Bundle to a specified render farm. If you would like to examine that job bundle, then you can use the Export Bundle button in the Submitter to export the Job Bundle to a location of your choice. If you want to submit the exported Job Bundle manually outside VRED, then you can use the Standalone AWS Deadline Cloud Client to submit that same Job Bundle to your specified render farm in a platform-agnostic manner.

Standalone AWS Deadline Cloud Client render jobs should use the appropriate Job Bundle Template obtained through this link. There, you will find one Job Bundle Template for supporting tile-based rendering (template.yaml) and another Job Bundle Template for non-tile rendering (template_tiling.yaml). When submitting a render job that doesn't rely on tiling, you can use the standard job.

Please also ensure that your Job Bundle directory has a scripts subdirectory containing VRED_RenderScript_DeadlineCloud.py, which is a required pipeline rendering component for the job bundle. You should also include parameter_values.yaml (values for the fields defined in the Job Bundle Template) and asset_references.yaml (for defining file dependencies).

Optional: Worker Setup - Customer Managed Fleet (CMF) (for Windows and Linux only)

While using AWS' Service Managed Fleet (SMF) is highly recommended, you can also set up a customer managed fleet (CMF) worker for your own farm using the steps below:

1. Create your own farm in the AWS Console (under the AWS Deadline Cloud service). Note the Farm ID, Fleet ID, and AWS Region (substituting them in step 3).

2. Install the AWS Deadline Cloud worker agent:

   pip install deadline-cloud-worker-agent
   

3. Configure the worker agent in C:\ProgramData\Amazon\Deadline\Config\worker.toml or similar (and substitute the field values below):

   farm_id = "farm-<farm-id>"
   fleet_id = "fleet-<fleet-id>"
   profile = "<my_aws_profile_name>"
   worker_logs_dir = "c:/users/username/desktop"
   
   [capabilities.amounts]
   "amount.attr.worker.gpu" = 1

4. Switch to the AWS profile (substitute below) that was used to create the farm.

   # on Windows:
   set AWS_PROFILE=my_deadline_profile
   # on Linux:
   export AWS_PROFILE=my_deadline_profile

5. Create a batch file to continually re-run worker agent:

   (Windows: create run-worker.bat)

   :do
   python -m deadline_worker_agent --run-jobs-as-agent-user
   goto do

   (Linux: create run-worker.sh)

   #!/bin/sh
   while true; do
       python -m deadline_worker_agent --run-jobs-as-agent-user
   done

   chmod +x run-worker.sh

6. Optional: Follow the instructions in the ImageMagick Installation section (if you would like to support region rendering/tiling)

7. Run the worker: run-worker.bat (Windows), ./run-worker.sh (Linux)

ImageMagick Installation

For render jobs using region rendering (tiling), ImageMagick must be available on the worker nodes. Install ImageMagick so the magick command is available in the system PATH. Or, set the MAGICK environment variable to point to a specific ImageMagick executable.

Windows

1. Visit the ImageMagick downloads page https://imagemagick.org/script/download.php
2. Download and install the 64-bit static binary release
3. Choose one of the following options:
   • Option A: Add ImageMagick to your system PATH so magick command is available globally
   • Option B: Set the MAGICK environment variable to point to the specific executable:

     setx MAGICK "C:\Program Files\ImageMagick-7.1.1-Q16\magick.exe"

4. Logout, login.

Linux

Important: use a static ImageMagick binary to prevent missing file format decoder dependency issues:

1. Apply the commands below, adjusting the MAGICK environment variable as appropriate (depending on ImageMagick binary path/version):

wget https://imagemagick.org/archive/binaries/magick
chmod 755 magick
yum install fontconfig libX11 fribidi  # or dnf install fontconfig libX11 fribidi
export MAGICK=$PWD/magick

You can also add these steps (above) to an existing Conda package (or make another one) and reference that package's name in the render job submission (under the Conda Packages setting). At the time of writing, Customer Managed Fleet (CMF) instances only support tiling since there isn't yet an official Conda package that provides a static ImageMagick binary (including dependencies) on Linux. However, it is possible to create that Conda package.

IMPORTANT: Without ImageMagick, tile assembly will fail and region rendering jobs will not complete successfully.

Environment Variables

The following environment variables can be used to configure the Submitter amd:

Environment Variables for VRED Submitter

Optional

• CONDA_CHANNELS: Override default conda channels for job environments (example: s3://conda-bucket/Conda/linux-64)
• CONDA_PACKAGES: Override default conda packages (example: vredcore=2026*)

Environment Variables for Fleet/Worker Nodes

One of these environment variables must be set:

• VREDCORE: Path (including filename) to VRED Core executable (takes precedence if VREDPRO environment variable is also set)
• VREDPRO: Path (including filename) to VRED Pro executable

Optional

• VRED_DISABLE_WEBINTERFACE: Disable VRED web interface (enabled by default)
• VRED_IDLE_LICENSE_TIME: License release timeout in seconds (set to "60" by default)
• FLEXLM_DIAGNOSTICS: FlexLM diagnostic level for licensing (set to "3" by default)

Environment Variables for Tile Assembly

• MAGICK: (Optional) Path to ImageMagick binary executable (required for region rendering jobs). If not set, the default magick command from PATH will be used.

Troubleshooting

In general, the AWS Deadline Cloud Monitor provides valuable insights into the activities that occurred while a worker node was processing a VRED scene file. Logs from these activities can be generated and shared with support teams. Depending on the scope/nature of an issue, there may be additional avenues worth investigating. Below are suggestions.

Common Issues

1. Worker Reports that VRED Executable is not Found

   • Verify the setting of the VREDCORE or VREDPRO environment variables
   • Check VRED installation path and permissions

2. VRED Hangs at Startup or lists "builtins.builtins.exec blocked by python sandbox"

   • Load VRED using the -insecure_python program argument, disable Python Sandbox in VRED's preferences.
   • Alternatively, if you must use VRED's Python Sandbox (against our recommendations), then verify that the Python module allowlist (python-sandbox-module-allowlist.txt) had its contents entered into VRED's preferences.
   • Verify that your Job Bundle Template is applying the -insecure_python program argument
   • On Windows, check VRED logs in %TEMP%\VREDPro\log

3. VRED Submitter Menu is Missing

   • Ensure that DeadlineCloudForVRED.py is in the correct directory for the version of VRED being used
   • Ensure that DeadlineCloudForVRED.py is listed in VRED's Preferences screen:
     • Edit menu → Preferences → General Settings → Script
   • Ensure that AWS Deadline Cloud for VRED was installed correctly
   • Check the VRED Console for any unusual errors

4. Worker is Producing Unusual Crashes

   • Ensure that the worker has ample resources (CPU and GPU memory) for processing the scene file in question
     • For SMF workers, define appropriate render farm resource requirements and update your host requirements configuration:
     • Target ample resources:
       • GPU: A10G, L4, L40S. 24-48GB of GPU memory
       • CPU RAM: 32GB-64GB
       • Try to leverage G5, G6, and G6e machine instance families.
     • For CMF workers, define similar render farm resource requirements (to those above) and update your host requirements configuration.

5. Tile Assembly Failed

   • Verify that ImageMagick is installed and accessible either via:
     • The magick command in system PATH, or
     • The MAGICK environment variable pointing to the executable
   • Check ImageMagick executable permissions
   • Ensure sufficient disk space for tile processing

6. Rendering Process is Slow

   • Verify that OpenGL hardware acceleration is enabled and functional
   • Verify that ample hardware resources are allocated
   • Consider tile rendering your scene file
   • Allocate ample worker nodes
   • Disable GPU Raytracing (if not required)
   • Scale down quality level and image resolution

7. VRED Licensing Issues

   License checkout failed
   

   • Verify VRED license server accessibility
   • Check available floating licenses
   • Ensure license server can handle concurrent requests from render farm
   • Check diagnostic worker output feedback in Deadline Cloud Monitor

VRED Rendering Issues (Detailed)

1. Raytraced Renders Appear Solid Black

   • Verify that the scene file is configured for Raytracing
   • Region rendering requires Raytracing for proper tile generation
   • Check render quality settings in the scene file
   • Verify that X Server is correctly configured and operational when rendering on Linux
   • Lighting: Raytracing requires proper light sources
     • consider increasing light intensity 5-10x or add area lights/HDR environment
   • Materials: Verify materials have proper diffuse colors and raytracing-compatible properties
   • Camera: Check exposure settings and camera position
   • Quick fix: Add area light and verify that materials aren't pure black

2. Region-based Rendering/Tiling Appears Solid Black

   • Render regions only work with raytracing enabled in VRED and will be enabled by default (see the above scenario)

3. Rendered Output Colors Appear Inconsistent on Windows v.s. Linux

   • Check the tone mapper being applied to the rendered Camera/view.
   • Try changing the tone mapper and all Color Space/Range-related settings to the sRGB/Reinhard/Linear color space.
   • Use a compatible/hardware qualified NVIDIA driver level (553.xx)
   • Consider distributing ICC monitor profile (if necessary) per Autodesk recommendations

Versioning

This package's version follows Semantic Versioning 2.0, but it is still considered to be in its initial development, thus backwards incompatible versions are denoted by minor version bumps. To help illustrate how versions will increment during this initial development stage, they are described below:

1. The MAJOR version is currently 0, indicating initial development.
2. The MINOR version is currently incremented when backwards incompatible changes are introduced to the public API.
3. The PATCH version is currently incremented when bug fixes or backwards compatible changes are introduced to the public API.

Security

We take all security reports seriously. When we receive such reports, we will investigate and subsequently address any potential vulnerabilities as quickly as possible. If you discover a potential security issue in this project, please notify AWS/Amazon Security via our vulnerability reporting page or directly via email to AWS Security. Please do not create a public GitHub issue in this project.

Telemetry

See telemetry for more information.

License

This project is licensed under the Apache-2.0 License.