RenderDeck is a self-contained render manager for Unreal Engine. Submit a sequence, a level and a pipeline preset; RenderDeck queues the job, sends it to the next free machine and renders it, with live progress, failure detection and a report on every finished shot.
Free, open and self-hosted. One command to run.
Download link contains a minimal example project in Unreal. Please follow the included guide.pdf on how to run it.
Nothing sits in the middle but a FastAPI server and a SQLite database. Workers connect over a websocket, claim what they are allowed to claim, and report back as they render.
Batch a shot list directly from the web panel, or POST jobs straight to the API.
Priorities, tags and path maps resolve every job to a sequence, a level and a config.
Idle workers pull the next job for their project, or take only the work assigned to them.
The worker renders the shot with Movie Render Queue on the chosen preset, streaming progress back.
Every finished job files a report: durations, frame counts, failures, full history.
Not a generic farm manager with a plugin bolted on. Designed for Unreal Engine: a level sequence, a map and a Movie Render Queue preset, resolved to real content paths and rendered in Unreal on the machine that picks it up.
Nudge priorities, retry, stop, reassign, bulk-clear. Filter by status, tag, worker or search, then act on the selection.
Machines announce themselves with their specs, take distributed work or a fixed assignment, and can be drained or shut down from the browser.
Maps, scenes and shots keep the real content paths. Rename a naming convention once and every job resolves against the new one.
Save MRQ configurations as named presets (proxy, HD, 4K, ACES) and pick one per job instead of per artist.
Server-sent events push job, worker and progress changes to every open tab, with toasts for the things you need to notice.
Success, error and cancelled all file a report. Average render times, status breakdowns and a searchable history of finished renders.
The server bootstraps its own virtual environment on first launch, installs its requirements into it, and re-launches itself inside it. Nothing touches your system Python, and there is no bundler in sight.
# start the server (creates ./venv on first run) $ python3 server.py [bootstrap] Re-launching inside venv -> ./venv/bin/python Starting server... ✅ Server running. Press Ctrl+C to stop. # open the UI $ open http://localhost:6789/ # or kick the tyres with a simulated farm $ python3 server.py --demo --demo-seed
Try the live demo, a seeded project with a thousand shots and a simulated render farm, or download it and point it at your own machines.
Download link contains a minimal example project in Unreal. Please follow the included guide.pdf on how to run it.
