Distaph
The Distaph icon: a single violet thread glowing against black.

Spin a lifetime of footage into stories.

Distaph catalogs every clip you've ever recorded, keeps your editorial judgments in files you own, and spins real moments from across the years into rough cuts that open in DaVinci Resolve.

Why the name

A distaff is the staff that holds raw wool while a spinner draws it out and twists it into one continuous thread. Your footage is the wool: years of it, tangled across drives and projects. Distaph holds all of it and helps you draw out the strands that belong together.

How it works

You record. Distaph does the logging. You decide what matters.

  1. Drop footage in your intake folder

    Old archives and yesterday's clips alike. That's the whole import step.

  2. Sync

    Each file is stored by its content hash, dated from its own metadata, transcribed with Whisper, captioned frame by frame, and made searchable. All of it runs on your Mac.

  3. Review in Obsidian

    Proposed project tags, undated clips and likely duplicates arrive as a checklist. Tick what you agree with and apply. Nothing is filed without you.

  4. Ask for a video

    Describe it, or start from a narrative. Claude chooses and orders real moments; Distaph checks every one and builds the timeline.

  5. Finish in Resolve

    Import the FCPXML and do the part that's yours: pacing, color, sound.

What it does differently

Plenty of tools turn one shoot into one rough cut. Distaph is built for the long run: everything you've filmed, organized the way you think about it, still useful in ten years.

Narratives outlive videos
A narrative is a named, ordered thread of real moments, separate from any project or finished video. It keeps growing as you keep filming. One narrative can become many videos, and one video can draw on many narratives.
Your judgments live in your files
Transcripts and timecodes are facts the machine can always regenerate. Which project a clip belongs to, what order a story goes in, what counts as a good take: those are yours, kept as plain markdown in an Obsidian vault. Either side can be rebuilt without touching the other.
Your library is a git repository
Every sync and every review is a commit, so you can see how your archive and your thinking about it changed over time. Transcripts are versioned as text. Distaph never pushes anywhere.
Local and open
Transcription, captioning and search run on Apple Silicon. Footage never leaves your drive. The code is MIT-licensed.

It can't put words in your mouth

When AI edits from a transcript, it's tempted to tidy what you said. Distaph doesn't give it the chance. A cut is a list of pointers to segments that were actually recorded, never text:

{"name": "projector-build", "segments": [
  {"id": "seg_fe2298880c97_r1_0000", "pip": true},
  {"id": "seg_18cc16b61141_r1_0019"},
  {"id": "seg_18cc16b61141_r1_0042"}
]}

Before anything is built, every pointer is checked against the database. One that doesn't exist stops the cut:

$ distaph validate --cut cuts/projector-build.json
{
  "checked": 3,
  "missing": ["seg_18cc16b61141_r1_0042"],
  "bounds": [],
  "drift": [],
  "ok": false
}

The timeline is assembled by code, from the database, using only pointers that passed this check. There's no path for invented text to reach it.

Nothing to memorize

Open Distaph and you get a menu of everything it does, in the order you'd do it. Pick a narrative or project from a list instead of remembering its name. Cataloging runs right there; making things hands off to Claude Code with the context already written, and brings you back when you're done.

Every action is also a command: distaph sync, distaph search "…", distaph build.

Your library is yours

The app and your data never mix. Distaph lives in Applications; everything it knows about your footage lives in one folder you own.

~/Distaph/
├── distaph.toml   where your footage lives
├── vault/         projects, narratives, reviews — open in Obsidian
├── facts/         transcripts and metadata as text, versioned
├── cuts/          cut lists, versioned; timelines rebuilt on demand
└── db/            working database, rebuilt from facts/

If you add a remote to back it up, make it private. It's a record of your life.

Install

You'll need an Apple Silicon Mac, uv, ffmpeg, Claude Code for the making features, and DaVinci Resolve to finish in.

git clone https://github.com/JulianKingman/distaph \
  /Applications/Distaph
/Applications/Distaph/macos/build-app
distaph init

Then point ~/Distaph/distaph.toml at your footage drive and open Distaph from Spotlight.

Where it stands

Early and opinionated. Working now:

  • Content-addressed ingest and metadata dating
  • Local transcription and keyframe captions
  • Semantic search over lines and shots
  • The review checklist in Obsidian
  • Validated cuts with picture-in-picture
  • FCPXML export to Resolve
  • Versioned libraries and the launcher

Next: summaries at every level from clip to archive, richer narrative tooling, and suggestions for what to make from what you already have. The design is written down in the specs; issues and pull requests are welcome.