Depth-Anything TensorRT in TouchDesigner

TouchDesigner implementation for Depth Anything and Depth Anything v2 with TensorRT monocular depth estimation.

Features

• One click install and run script
• In-TouchDesigner inference

Usage

Tested with TouchDesigner 2023.11340 & Python 3.11

Install:

1. Install Python 3.11
2. Install CUDA Toolkit 11.8
3. Download TDDepthAnything repository
4. Open accelerate.bat with text editor and set path to Python 3.11 in set PYTHON_PATH=. (ex. set PYTHON_PATH="C:\Program Files\Python311\python.exe")
5. Run accelerate.bat

Tip

You can use same .venv for TDDepthAnything & TouchDiffusion. Copy all files from TDDepthAnything folder to TouchDiffusion and run accelerate.bat. In TouchDesigner extension, on settings tab select TouchDiffusion folder also.

Accelerate models:

1. Download Depth-Anything model or Depth-Anything v2 model
2. Copy model to checkpoints folder
3. Run accelerate.bat
4. Select model version (1 - Depth-Anything, 2 - Depth-Anything v2)
5. Select model size (s - small, b - base, l - large, g - giant)
6. Select width & height (default is 518x518)
7. Wait for acceleration to finish

TouchDesigner inference:

1. Add TDDepthAnything.tox to project
2. On Settings page change path to TDDepthAnything folder and click Re-init
3. On Depth Anything page select path to engine file (for ex. engines/depth_anything_vits14.engine) and click Load Engine

Acknowledgement

Based on the following projects:

• Depth-Anything - Unleashing the Power of Large-Scale Unlabeled Data.
• Depth-Anything TensorRT C++ - Leveraging the TensorRT API for efficient real-time inference.
• TopArray - Interaction between Python/PyTorch tensor operations and TouchDesigner TOPs.