> ## Documentation Index
> Fetch the complete documentation index at: https://docs.comfy.org/llms.txt
> Use this file to discover all available pages before exploring further.

# VAEDecodeHunyuan3D - ComfyUI Built-in Node Documentation

> Complete documentation for the VAEDecodeHunyuan3D node in ComfyUI. Learn its inputs, outputs, parameters and usage.

> This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! [Edit on GitHub](https://github.com/Comfy-Org/embedded-docs/blob/main/comfyui_embedded_docs/docs/VAEDecodeHunyuan3D/en.md)

The VAEDecodeHunyuan3D node converts latent representations into 3D voxel data using a VAE decoder. It processes the latent samples through the VAE model with configurable chunking and resolution settings to generate volumetric data suitable for 3D applications.

## Inputs

| Parameter           | Data Type | Required | Range       | Description                                                                             |
| ------------------- | --------- | -------- | ----------- | --------------------------------------------------------------------------------------- |
| `samples`           | LATENT    | Yes      | -           | The latent representation to be decoded into 3D voxel data                              |
| `vae`               | VAE       | Yes      | -           | The VAE model used for decoding the latent samples                                      |
| `num_chunks`        | INT       | Yes      | 1000-500000 | The number of chunks to split the processing into for memory management (default: 8000) |
| `octree_resolution` | INT       | Yes      | 16-512      | The resolution of the octree structure used for 3D voxel generation (default: 256)      |

## Outputs

| Output Name | Data Type | Description                                                        |
| ----------- | --------- | ------------------------------------------------------------------ |
| `voxels`    | VOXEL     | The generated 3D voxel data from the decoded latent representation |