Geneva 1.1 was released as a command‑line video editor where each edit is a JSON document and titles or graphics are plain HTML and CSS rendered without a browser. The tool ships a single binary for Linux (x64, arm64), macOS (Apple silicon) and Windows (x64), and pulls codecs from FFmpeg libraries while the planner, compositor and layout engine are written in Rust. Install is a one‑liner on Linux/macOS via curl -fsSL https://raw.githubusercontent.com/geneva-render/geneva/main/scripts/install.sh | sh or PowerShell on Windows【https://raw.githubusercontent.com/geneva-render/geneva/main/scripts/install.sh】.
Editing model and command set
Every operation—trim, concat, convert, subtitle burn, probe—turns into a JSON document that the engine renders. Example commands include:
geneva trim clip.mp4 -o short.mp4 --from 00:01:00 --to 00:01:30
geneva concat part1.mp4 part2.mp4 -o merged.mp4 --crossfade 1s
The JSON can reference HTML assets for overlays; a lower‑third card can be written in HTML/CSS and positioned with flexbox or absolute coordinates. The same card scales automatically because percentages are relative to the frame size.
Performance compared to script‑based pipelines
Benchmarks on a 4‑vCPU Intel Xeon (no GPU) show Geneva rendering a 13‑second opening in about 10 seconds, whereas Playwright + ffmpeg took 64 seconds and Remotion 19 seconds. A 3‑minute broadcast package rendered in 52 seconds with Geneva, compared to 244 seconds for Playwright and 324 seconds for Remotion. The tool copies unchanged H.264 frames directly when the optional x264 library is installed, reducing encode work.
Limitations to be aware of
Geneva is purely a CLI—there is no graphical editor or hosted service. CSS support is limited to flexbox, absolute positioning, gradients, filters, clip‑path, blend modes, background‑clip, and keyframe animations; CSS grid, floats, static transforms and JavaScript are not supported. The README notes that the codebase was largely generated by AI models, which may affect trust in edge‑case behavior. Pricing is not applicable; the project is MIT‑licensed and free to use.
When to give it a try
If you already automate video pipelines and want deterministic renders without a UI, try Geneva on a small clip (e.g., a 10‑second social teaser) to compare render time and output size against your existing ffmpeg scripts. Watch for future updates around parallel chunk rendering, which could further speed up long‑form renders.
