# Build an Infinigen conda package for Deadline Cloud
<a name="examples-conda-infinigen"></a>

The [infinigen-1.19.0](https://github.com/aws-deadline/deadline-cloud-samples/tree/mainline/conda_recipes/infinigen-1.19.0) rattler-build recipe packages [Infinigen](https://infinigen.org/), a procedural 3D scene generator from Princeton Vision & Learning Lab. The package includes the following:
+ Infinigen v1.19.0 procedural scene generation engine.
+ `bpy 4.2.0` (Blender as a Python module) for headless rendering through Cycles. No standalone Blender installation is needed at runtime.
+ Terrain C\+\+ shared libraries (CPU and optional CUDA) compiled from Infinigen source for landscape generation.

To build this recipe, deploy the [CUDA farm CloudFormation template](https://github.com/aws-deadline/deadline-cloud-samples/tree/mainline/cloudformation/farm_templates/cuda_farm) to create a Deadline Cloud farm with a CUDA GPU fleet for the build. The build requires a GPU worker (for CUDA terrain compilation) and takes approximately 6 minutes.

Submit the build from the `conda_recipes` directory:

```
./submit-package-job infinigen-1.19.0
```

To reuse conda environments between jobs, attach the [conda\_queue\_env\_improved\_caching.yaml](https://github.com/aws-deadline/deadline-cloud-samples/blob/mainline/queue_environments/conda_queue_env_improved_caching.yaml) queue environment to your queue. The dependency closure of Infinigen contains many gigabytes of packages, so caching saves significant time and bandwidth.

For a job bundle that uses this package, see [Generate procedural 3D scenes with Infinigen on Deadline Cloud](examples-jb-infinigen.md).