DOCKER PIPELINE

Ingest to export, step by step

This page walks you through the Shotloom pipeline — not infrastructure deep dives. Drop in multicam clips, Docker services sync and montage in background, download platform-ready exports. The question isn't "how is it built inside" but "can I bootstrap tonight and ship a rough cut tomorrow?"

Ingest

2–4 camera clips

Sync

Audio alignment locally

Analyze

VAD + face scoring

Montage

AI rough cut draft

Export

YT · TT · IG · FB

Docker API

Dashboard at 127.0.0.1:8765

Worker services

Sync, analyze, render

Your disk

Projects dir on your host

Fig 1. Shotloom pipeline — ingest → sync → analyze → montage → export

Docker services that run the pipeline

Bootstrap once

Run ./scripts/bootstrap.sh to pull images and start compose services. API binds to localhost by default. Open the dashboard, drop in your first multicam session — no cloud account required.

Background workers

Sync, VAD, face scoring, montage, and export run as Docker services while you keep working. Long sessions don't glue you to the progress bar. Check status in the dashboard Media tab.

Refine in dashboard

After the rough cut lands, tune montage rhythm, subtitle style, and platform durations in Settings and Captions tabs. Re-export without re-ingesting — same aligned streams, new profiles.

SESSION FLOW

What happens to your multicam session

You drop 2–4 camera files into the dashboard. Ingest queues the session. Sync aligns audio across streams locally. Analyze runs VAD and face scoring for 9:16 crop decisions. Montage builds an AI rough cut you can tune. Captions get styled SRT. Export queues YouTube, TikTok, Instagram, and Facebook profiles — all on hardware you control.

Elena, Marco, Luca, and Sara decide from one real session: sync accuracy, montage usability, caption quality, export coverage. No lock-in, no forced cloud timeline. If the pipeline helps, bootstrap stays; if not, you learned fast without a subscription.

"Can I bootstrap tonight?" → Yes: ./scripts/bootstrap.sh, open 127.0.0.1:8765.
"Will long sessions block my day?" → No, workers run in Docker background.
"Does footage stay local?" → Yes, projects dir on your host by default.

Shotloom — Dashboard · session_ep03

Ingest
podcast_ep03_cam{A,B,C}.mp4 → queued

Sync
aligning 3 camera streams locally... done

Analyze
VAD + face scoring for 9:16 crop... done

Montage
rough cut draft — 68% → tweak in Settings

Captions
styled SRT ready

Export
YouTube 16:9 + TikTok 9:16 queued
stored on your disk — no cloud upload

Your first session is waiting

If you've read this far, bootstrap with Docker, drop 2–4 multicam clips, and see if ingest → sync → montage → export fits your publish workflow. That single test is what matters.

Get started with Docker First-run guide See the features