OCAP Installation

OCAP (Open-World Agent Capture) is the desktop interaction capture system we'll use to record screen, audio, and input data. This section covers the installation and configuration process.

Directory Structure

First, set up the required directory structure on your EC2 at the C:\ root level:

C:\scripts\
├── pid\
│   └── (PID files will be written here - see fork documentation)
├── temp_recordings\
│   └── (OCAP will write recordings here)
├── meeting\
│   └── (meeting metadata will go here)
└── (scripts go here)

C:\projects\
└── (OCAP repository will go here)

Downloading the OCAP Fork

Download the custom OCAP fork as a ZIP file:

Repository: https://github.com/shaundano/elephant-ocap

Extract the ZIP file into the C:\projects\ folder.

Note - Why Use This Fork?

This fork includes several customizations that allow it to run on non-GPU EC2 instances and includes additional features like microphone capture. See the Fork Discussion for technical details.

Installing Miniconda3

OCAP requires Conda as a package manager (used by Open World Agents). Install Miniconda3 via PowerShell:

Download Miniconda3

Invoke-WebRequest -Uri "https://repo.anaconda.com/miniconda/Miniconda3-latest-Windows-x86_64.exe" -OutFile ".\Downloads\Miniconda3-latest-Windows-x86_64.exe"

Run the installer. If given the option to install for all users, accept it.

Installation Complete

Once installed, you should see (base) in your PowerShell prompt, indicating that Conda is active.

Creating the Conda Environment

Create a new Conda environment for OCAP:

Create Conda Environment

conda create -n ocap-env python=3.14

Activate the environment:

Activate Conda Environment

conda activate ocap-env

You should now see (ocap-env) in your terminal prompt.

Installing OCAP

Navigate to the OCAP project folder in PowerShell, then install it in editable mode:

Install OCAP

pip install -e .

Note - Editable Mode

The -e flag installs in editable mode, which is like debug mode or hot refresh. If dependencies change in the Python project, it should automatically react. The dependencies are defined in pyproject.toml.

Installing GStreamer Dependencies

Follow the official OCAP documentation and install GStreamer dependencies:

Install GStreamer Dependencies

# Install GStreamer dependencies first (for video recording)
conda install open-world-agents::gstreamer-bundle

The regular documentation says to also run pip install ocap, but since we downloaded it locally, we don't need to do that.

Running OCAP

You should now be ready to run OCAP. Execute:

Run OCAP

ocap my-recording

This will write files to the default path, which is C:\.

Success Indicators

If everything is working correctly, you should see output similar to:

Troubleshooting

Common Issues

Warning - Don't Give Up!

Getting OCAP to run was one of the hardest parts of this project to debug. Once it worked, I didn't even really understand why it started working.

Environment Activation

Make sure your Conda environment is activated. You should see (ocap-env) on your terminal line.

Missing GStreamer Components

Error: no module 'gi' found

OCAP uses many different GStreamer components, and sometimes they'll be missing. Try:

1. Re-installing the GStreamer plugins from open-world-agents
2. Adding the conda-forge channel (should be enabled by default, but sometimes it isn't):

Add Conda Forge Channel

conda config --add channels conda-forge

Why Use the Fork Instead of pip install?

By downloading the project directly onto your EC2, you're able to rely on your own local repository rather than pip install ocap, which uses the most recent version. This gives you visibility into the dependencies you need in pyproject.toml and allows you to customize the code if needed.

---

Next: Fork Discussion →