OpenVFX

README

BYOM Video

Local-first video workflow CLI. Transcribe, cut, caption, and plan exports from your own machine — no cloud required.

Alpha. Schemas and commands are still evolving.

Requirements

• Go 1.22+
• ffmpeg and ffprobe on PATH
• Python 3.10+ with faster-whisper for real transcription (optional for metadata-only runs)
• Ollama for local model expansion (optional)

Install

curl -fsSL https://raw.githubusercontent.com/mirelahmd/OpenVFX/main/install.sh | sh

This installs the byom-video binary and sets up a Python environment with faster-whisper for transcription. Requires Go 1.22+ and Python 3.10+.

Or build from source:

git clone https://github.com/mirelahmd/OpenVFX.git
cd OpenVFX
go build -o byom-video ./cmd/byom-video
./byom-video version

Quickstart

# Check dependencies
./byom-video doctor

# Initialize workspace
./byom-video init

# Metadata-only run (no Python needed)
./byom-video pipeline media/clip.mov --preset metadata

# Full shorts pipeline
BYOM_VIDEO_PYTHON=.venv/bin/python ./byom-video pipeline media/clip.mov --preset shorts

# Inspect and export
./byom-video runs
./byom-video inspect <run_id>
./byom-video export <run_id>

What It Does

Area	Commands
Pipeline	pipeline, run, batch, watch
Inspection	inspect, artifacts, validate, open-report
Export	export, ffmpeg-script, export-manifest, concat-plan
Clips	clip-cards, review-clips, selected-clips, enhance-roughcut
Agent planning	plan, review-plan, approve-plan, execute-plan, revise-plan
Inference mask	mask-plan, review-mask, revise-mask, mask-decisions, mask-decision
Expansion	expansion-plan, expand-local-stub, expand, verify-expansions
Maintenance	cleanup, retry-batch, rerun, doctor

Local Model Setup (Optional)

# Install faster-whisper for transcription
python3 -m venv .venv && .venv/bin/pip install faster-whisper

# Install Ollama for local model expansion
ollama pull qwen2.5:7b
./byom-video models doctor
./byom-video expand <run_id> --dry-run

What's Not Here Yet

• No web UI
• No cloud provider execution (OpenAI, Anthropic, etc.)
• No DaVinci / Premiere integration
• No Docker workflow

License

MIT