Repository navigation
Cast library
The cast library (#41) keeps a few head-and-shoulders photos of each regular for covers and the weekly issue: from their streams' named faces, Wikimedia Commons, their own channels and image search, each checked against their known face and labelled (neutral, laughing, angry...). The war room's Cast window keeps or drops them.
It needs the faces extra (uv sync --extra faces) and [thumbnail] photos set
(Covers, Photos). clippa serve runs the daily find and the Cast window, so it
needs the faces extra too (clippa setup adds it for the serve role; clippa doctor --role serve checks).
Image search is Brave's Search API (about $5 per 1,000 searches): make a key at
api-dashboard.search.brave.com and put it in secrets/brave_secret.txt ([paths] brave_key)
or the Windows user environment variable BRAVE_API_KEY; without it the other sources still
work. The same key serves the web and Reddit searches of the search swarm.
The library is photos/<Person>/ folders of several photos each, kept by clippa cast
beside the single files (a person's best kept photo wins over their <Person>.jpg).
-
clippa cast grab <stream>: each named person's sharpest faces in a stream clippa cast set Jer --alias "The Quartering" --channel https://www.youtube.com/@TheQuartering-
clippa cast find Jer: Wikimedia Commons, the channels' avatars and video thumbnails, image search clippa cast add Jer <file|url>-
clippa cast keep|drop|label ...andclippa cast list
Every photo is checked for the person's face first (only theirs if they're enrolled in
faces/known.json), cropped 4:5 and labelled by a Haiku call ([cast] label_model, about
$0.0002 a photo; --no-label or "" skips it). It never takes photo agencies' pictures
([cast] blocked_domains), but image search photos have no known licence, so prefer stream and
Commons ones for covers (covers do, given the choice). clippa doctor flags anyone with fewer
than [cast] min_photos (3) kept.
The worker's cast stage (after faces in [worker] stages) grabs each new stream's named
faces for anyone with fewer than [cast] target (6) photos, and serve runs clippa cast auto
every [cast] auto_hours (24), a find from every source for the same people (someone it found
nothing for waits [cast] retry_days). A photo is kept without asking when it comes from a
stream, Commons or a channel ([cast] auto_keep_sources; never image search, whose licence is
unknown), the person is enrolled, their face matches at [cast] auto_keep (0.65) or better and
the label has no doubt; the rest wait as New in the Cast window. Costs: a Haiku label per photo
and a Brave search per person tried, cents a day.
Covers ([thumbnail] enquirer_prompt_version v6) and the weekly ([weekly] prompt_version v3)
are told each person's labels and pick the photo that fits the story; the war room's cover
editor offers them too ("Jer · angry"). Everyone with a kept photo is added to [scoring] known_figures (with their aliases) unless [cast] known_figures = false.
Home · Getting started · clippa doctor · These pages are built from docs/wiki/ in the repo, so edits made here are overwritten: how to edit them
Run it
- Overview
- Getting started
- Roles and services
- Configuration
- Pipeline
- Review
- Reviewing
- War room
- Clippa messenger
- Scheduling and uploads
- Costs
- Checkup
- Audit
- Backups
Install
- Install
- Install checklist
- Kitsu
- Discord app
- YouTube sign-in
- Windows
- macOS
- Transcribe on Modal
- NAS recorder
- NAS hub
- Claude on the NAS
- Phone access
- Model providers
- Adding a show
- YouTube channel
Features
- Pre-shows
- Fetching VODs
- Speakers
- Faces
- Covers
- Cast library
- B-roll
- Clip art
- Weekly issue
- X posts
- TikTok
- News watcher
- Character pieces
- Documentaries
Research
Working on Clippa