Skip to content

Installation

Quickstart

No git clone, no configuration file.

1 — Download
curl -LO https://raw.githubusercontent.com/depictio/depictio/stable/docker-compose.yaml
2 — Start
docker compose up -d

Open Depictio

Service URL Notes
Depictio localhost:5080 Single-user mode — no login required
API docs localhost:8058/docs Interactive OpenAPI interface

  • Customise credentials

    Copy .env.example to .env to change the S3 password or switch to multi-user mode.

    Advanced configuration

  • Multi-user or public mode?

    Set DEPICTIO_AUTH_SINGLE_USER_MODE=false in .env to enable accounts and login.

    Authentication modes

Server Deployment

Three ways to run the same Depictio server, which differ in how its services run.

  • Kubernetes · Helm chart

    Deploy Depictio on a Kubernetes cluster with the official Helm chart.

    Ideal for production environments and scalable deployments.

    Installation guide

  • Docker Compose · containers

    The recommended way to run Depictio. One compose file starts every service in containers, object storage included. Upgrading from before v1.12.0? Copy the former MinIO store's data first: Upgrading to v1.12.0.

    Ideal for development, testing, and small-scale deployments.

    Installation guide

  • Python package · pip / uv, no containers

    Install depictio[local] with uv or pip: depictio local up then runs every service as a local process on your machine.

    Ideal for reviewing a template on your own results, on a laptop.

    Installation guide

CLI & Configuration

  • Depictio CLI

    Command-line tool for data ingestion, project management, and interacting with the Depictio API.

    CLI guide

  • Configuration

    Configure authentication, S3 storage, backups, and advanced features via environment variables.

    Configuration guide

  • Try in GitHub Codespaces

    A cloud workspace with the Python package already installed, which runs depictio local up in a terminal of the editor when it opens, so the server's start-up shows there: nothing to set up on your machine, a GitHub account is enough. The viewer opens on its own once the server is ready, in single-user mode, with the Iris and Penguins examples.

    Open in Codespaces