Griptape Nodes: Depth Anything 3 Library

A Griptape Nodes library for monocular depth estimation using Depth Anything 3 models.

Overview

This library provides nodes for estimating depth from images and videos using state-of-the-art Depth Anything 3 models. The depth maps can be output as grayscale images (closer = brighter) or colorized using the INFERNO colormap.

Requirements

• GPU: CUDA (NVIDIA) or MPS (Apple Silicon) is required for inference
• Griptape Nodes Engine: Version 0.66.2 or later

Nodes

Depth Anything 3 Image

Estimates depth from a single image.

Parameters:

Parameter	Type	Description
model	Dropdown	Select the Depth Anything 3 model to use (see Available Models)
input_image	ImageArtifact	Input image for depth estimation
colorize	Boolean	When enabled, outputs a colorized depth map using the INFERNO colormap. When disabled, outputs a grayscale depth map where closer objects are brighter. Default: False
output	ImageArtifact	The resulting depth map

Depth Anything 3 Video

Estimates depth from a video, processing each frame independently.

Parameters:

Parameter	Type	Description
model	Dropdown	Select the Depth Anything 3 model to use (see Available Models)
input_video	VideoArtifact	Input video for depth estimation
colorize	Boolean	When enabled, outputs a colorized depth video using the INFERNO colormap. When disabled, outputs a grayscale depth video where closer objects are brighter. Default: False
output	VideoUrlArtifact	The resulting depth video

Available Models

The following Depth Anything 3 models are available from HuggingFace:

Model	Description
depth-anything/DA3-SMALL	Smallest and fastest model
depth-anything/DA3-BASE	Base model, good balance of speed and quality
depth-anything/DA3-LARGE	Large model with improved accuracy
depth-anything/DA3-LARGE-1.1	Improved large model (v1.1)
depth-anything/DA3-GIANT	Highest quality, slowest inference
depth-anything/DA3-GIANT-1.1	Improved giant model (v1.1)

Models are cached after first use, so subsequent runs with the same model will be faster.

Installation

Prerequisites

• Griptape Nodes installed and running
• A CUDA-capable NVIDIA GPU or Apple Silicon Mac

Install the Library

1. Clone the repository to your Griptape Nodes workspace directory:

   # Navigate to your Griptape Nodes workspace directory
   cd `gtn config show workspace_directory`
   
   # Clone the repository with submodules
   git clone --recurse-submodules https://github.com/griptape-ai/griptape-nodes-depth-anything-3-library.git

2. Add the library in the Griptape Nodes Editor:

   • Open the Settings menu and navigate to the Libraries settings
   • Click on + Add Library at the bottom of the settings panel
   • Enter the path to the library JSON file:

     <workspace_directory>/griptape-nodes-depth-anything-3-library/griptape_nodes_depth_anything_3_library/griptape-nodes-library.json
     

   • You can check your workspace directory with gtn config show workspace_directory
   • Close the Settings Panel
   • Click on Refresh Libraries

3. Verify installation by checking that the "Depth Anything 3 Image" and "Depth Anything 3 Video" nodes appear in the node palette under the "Depth/Estimation" category.

Usage

Image Depth Estimation

1. Add a Depth Anything 3 Image node to your workflow
2. Connect an image source to the input_image parameter
3. Select a model from the model dropdown
4. Optionally enable colorize for a color depth map
5. Connect the output to your next node or a display

Video Depth Estimation

1. Add a Depth Anything 3 Video node to your workflow
2. Connect a video source to the input_video parameter
3. Select a model from the model dropdown
4. Optionally enable colorize for a color depth video
5. Connect the output to your next node or a display

Note: Video processing can take significant time depending on the video length and model size.

Output Format

• Grayscale mode (colorize: False): Outputs a single-channel image where brightness indicates proximity. Closer objects appear brighter (white), distant objects appear darker (black).

• Colorized mode (colorize: True): Outputs an RGB image using the INFERNO colormap, where colors range from dark purple/black (far) through red/orange (mid) to bright yellow (near).

Troubleshooting

Library Not Loading

• Ensure the git submodule is initialized. If you cloned without --recurse-submodules, run:

  git submodule update --init --recursive

CUDA/MPS Not Available

• Verify your GPU drivers are up to date
• For NVIDIA GPUs, ensure CUDA is properly installed
• For Apple Silicon, ensure you're running on macOS 12.3 or later

Out of Memory Errors

• Try using a smaller model (e.g., DA3-SMALL or DA3-BASE)
• For videos, consider processing shorter clips
• Close other GPU-intensive applications

Additional Resources

• Depth Anything 3 Paper
• Depth Anything GitHub
• Griptape Nodes Documentation
• Griptape Discord

License

This library is provided under the Apache License 2.0.