Stoatworks Labs

Services

The bits you can't download

Some jobs need more than an app. They need somebody to build the pipeline, run it through the event and hand it over at the end, and these are the two I do that way. Both are built on tools I wrote and give away, so there's no black box in here you can't open up and look inside.

01

Managed event content

Nextcloud and my own apps, run for you

Every conference has the same problem: two hundred presenters, four hundred versions of the same deck and a runner with a memory stick. I run the content pipeline for you instead, on a private Nextcloud with my own apps on top, set up for your rooms and your schedule.

Built on

  1. 01

    Collect

    A folder and an upload-only drop link for every session, made in bulk from your running order, whether that's a CSV, typed into the browser or live-synced from a Google Sheet. Each presenter gets emailed their own link, and one expiry date covers the whole batch.

  2. 02

    Normalise

    Keynote, PowerPoint, Google Slides and ODP all get converted to PDF with the presenter notes kept in a sidecar, so the notes survive the conversion that usually eats them.

  3. 03

    Distribute

    Each theatre gets an account that sees its own folder and the shared ones (holding slides, fonts, schedules) and nothing else, so a room team only ever sees its own content.

  4. 04

    Go on site

    Once the drop links expire, the whole tree mirrors to a second Nextcloud you take to the venue, so the show runs off a local server that couldn't care less about the site WiFi.

  5. 05

    Hand over

    Sessions stay in a tracked list you can edit, close early or remove, and the same session data sets up a DaVinci Resolve project for post.

More to comeI'm still writing up the rest (how engagements work, hosting options, what I run and what you run, and pricing), and it'll go here when it's done.

02

DIT & ISO offload

Recordings off the switcher and into the edit, while the session runs

A multi-room event makes recordings faster than anyone can carry them about. I build the ingest so nobody has to: ISO streams pulled off each switcher continuously, landing in the edit suite and the archive at the same time, ready to cut before the session's even finished.

Built on

  1. 01

    Pull, don't copy

    The ATEM shares its drive over FTP and will happily serve a file it's still writing. I pull only the bytes past the last known offset, using FTP's own resume (the same thing curl -C does). Re-copying the whole file every ten minutes can't keep up, and falls behind for good within twenty minutes.

  2. 02

    Nothing on the show laptops

    The ingest is one container on the central server, reaching each room over the event's private mesh. There's nothing to install on a machine that's running a session, and no software on the critical path.

  3. 03

    Two destinations at once

    Each pull writes to the edit suite's NAS and to Nextcloud at once, so editors work off local storage while the archive copy builds itself.

  4. 04

    The other recorders too

    vMix machines are pulled the same way over SMB, so program records and ISOs arrive on the same clock and in the same tree.

  5. 05

    Sized before it's built

    Four to six active ISOs plus program comes to 50–80 GB for a three-hour session. I do that sum against your actual uplink first and design to it, rather than finding out on day one.

More to comeI'm still writing up the full workflow (chain of custody, proxy generation, naming conventions and what you get handed at wrap), and it'll go here when it's done.

Talk to me

Got an event with this shape?

Tell me the room count, the recording setup and the uplink you're stuck with, and I'll tell you whether this would help, or whether you just need one of the free tools and an afternoon.

Drop me a line