TIMEPRINT DB Public beta Download
Windows x64
Local-first
No account

The database your media folders never had.

Timeprint DB catalogs large private media collections on Windows. Tag a person once and see everything you've tagged with them, across every collection and every drive. Your tags stay attached when files are renamed, moved or re-copied, and a changed drive letter costs one relocate step.

It never moves, renames or copies your files on its own. It reads; your folders stay exactly as you arranged them.

Download the free beta
v0.2.0 beta · ~203 MB · Windows 10/11 x64
No account. No cloud. No subscription.
A person's page in Timeprint DB: a portrait and bio, then videos and photo sets from two drives; the items on the unplugged drive are shown greyed out as offline but still listed

Drive 2 unplugged. Its four items are still listed, greyed out, with their tags and people.

Demo library. The person is AI-generated.

1
person record, linked across every collection
0
re-tagging after a rename or move
0
bytes uploaded by the app (AI note)
Chapter 01
Your situation

You've had this happen. Probably more than once.

A machine rebuild renumbers your drive letters. Or you finally consolidate onto a bigger disk. Then the catalog starts reporting duplicates that aren't duplicates, and missing files that are sitting right there.

Years of hand-made tags come loose from the files they described. You re-tag four thousand items by hand over three weekends. After that you stop reorganising your storage. The layout isn't right; touching it just costs too much.

The one part of a collection you can't re-download is the part you made yourself. It shouldn't depend on where the bytes happen to sit.

Sounds familiar? I'd like to hear how it happened to you →
Chapter 02
What it's for

Built for one job: keeping a collection re-enterable.

  1. See everything you've tagged with one person, in one view, across drives.
  2. Land on the twenty seconds you kept a two-hour file for.
  3. Keep the catalog's grip on every file when your storage changes shape.

One person, one record

The same person across separately stored collections and separate drives.

Rename, move, re-copy: same item

Tags, people and timepoints stay with the file. No re-tagging, no phantom duplicates.

Unplugged isn't lost

Disconnecting an external drive doesn't wipe or corrupt the catalog.

Nothing moved or extracted

Your folders stay as they are. ZIP photo sets stay zipped, where they are.

A catalog you can carry

A collection is a .vdb file plus its cache folder. Copy both to another drive or machine.

Work and private, kept apart

Environments: home libraries in one, work in another, or one per client. Each has its own collections, recent files, people and settings, so presenting one shows none of the others' collections, files or people. Out of sight, not locked.

The limit, stated plainly: identity comes from file content, so a rename or a move is invisible to it — but re-encoding a file makes it a new file.

Chapter 03
How it works

Every promise above, and the feature behind it.

Content fingerprinting

→ A file is recognised by what it is, not where it sits, so identity survives storage changes.

Portable paths + offline-drive detection

→ Each item stores an absolute and a root-relative path. A new drive letter costs one relocate step.

Global cast

→ One person record linked into every collection: the person axis across drives.

.vdb file + cache folder

→ The catalog moves machines. The cache holds thumbnails, cast photos and scripts, so copy both.

Watched roots + ZIP photosets in place

→ The catalog keeps up with the disk, and indexing costs no extra disk space.

Query, scripting and MCP

→ Find the gaps in your own curation, automate the rest, and let an AI assistant help (below).

D:\Archive\
├── main.vdb the catalog
├── main.cache\ thumbs · cast photos · scripts · backups
└── media\… left exactly as it is

People are records, not filter chips: each one has a page with everything you've tagged with them. Cast portraits come from any frame, through a built-in editor with crop, rotate, exposure, shadows, highlights and clarity.

person pages with full media lists portrait editor for cast photos bulk cast matcher tag merge & rename retag 500 items at once

Timepoints → re-enter the twenty seconds

Mark an instant or span a whole scene. Each timepoint has a title, a note, tags and its own thumbnail, drawn as markers under the scrubber, so a three-hour file reads at a glance. They're searchable and taggable like any other item.

Player with timepoint markers on the timeline and the timepoints sidebar

Scriptable — and open to the AI you choose

A JavaScript API over videos, photosets, cast, global cast, tags and raw SQL, with an editor and live log.

The built-in MCP server lets an AI assistant work on the whole catalog through that API: query it, tag it, link people, tidy it up. Use the assistant you already have, or an open-source model running on your own machine. The server is off by default and listens only on this PC; each assistant gets its own key and limits, and its writes wait for your OK unless you mark that key trusted. With a hosted model, the catalog data it reads goes to that provider; with a local model, nothing leaves the PC.

retag-shorts.js
const hits = db.videos.search({
  untagged: true, maxDuration: 60
});

for (const v of hits)
  db.tags.assign(v.id, 'shorts');
Chapter 04
Why it keeps happening

Every existing way makes you give something up.

Tools that take ownership of your files

They import, rename to a hash, restructure your folders. Your tags survive. The library you spent years arranging does not.

Tools that key everything to the file path

They respect your folders — until a path changes. Then the catalog and the files quietly drift apart.

A plain folder tree

It respects everything and outlives every tool, but it can't give you a person axis: the filename is the metadata.

Timeprint DB uses identity derived from file content, kept up to date through every way a file's path can change, while the files stay where they are.

Chapter 05
Straight answers

The fair questions, answered first.

"Windows warns me it's unsigned."
It will. The installer isn't code-signed yet; this is a solo developer's beta. Here's exactly what you'll see. If that's a dealbreaker today, fair — come back when it's signed.
"What if you stop working on this?"
A .vdb is a plain SQLite database; any SQLite tool opens it. Exports are keyed by content fingerprint, so your tags can be re-attached to the same files. You're not holding a file only my software can read.
"Will it touch my files?"
Scanning only reads them. It never moves, renames, copies or extracts on its own; file operations happen only when you choose them in the app.
"Do I have to import 45,000 items to find out?"
No. Point it at one folder you know well. The first ten minutes are below.
What changes

You reorganise your drives on a Tuesday afternoon, just because you felt like it. Then you open the catalog and everything is where you left it: still tagged, still linked to the right people.

The archive stops being something you have to be careful around.

Next
Not claimed yet

Everything above is in the build you can download today. These are next.

01Auto-update and a code-signed installer
02A backup screen — backups already run automatically
03macOS and Linux builds
04Ratings on videos and cast, with rating sort
05Smart / dynamic playlists and M3U export
06Series, seasons and parts modelling
07Co-cast browsing — who else appears here
08RAR and 7z archives — ZIP only today
09Windows shortcuts to videos and exact timepoints
10Markdown notes on videos and timepoints
Beta
Help shape it

It's a wild beta. Tell me where it breaks.

If chapter 01 sounded familiar, I want to hear your version. Timeprint DB is built by one person, and so far it's been shaped by one person's collection. Now I need yours: how it's organised, what you can't find, and where the app gets in your way.

  • Early accessGet new builds first, and a note when something you asked for ships.
  • 30-min callShow me how you manage your collection today. No pitch, just questions — and your answers decide the roadmap.
  • Free licenceDo a 30-minute call or send a written "where it broke" report, and keep full access after the beta.

Point it at one folder. See what it finds.

Timeprint DB · Windows x64 · one download, no account, and the app sends no telemetry.

Download for Windows — free

v0.2.0 beta · Windows 10/11 x64 · ~203 MB

screenshot
Windows SmartScreen prompt
Windows will show this because the beta isn't code-signed yet. Click More info → Run anyway. If Smart App Control is on, Windows may block the installer outright — tell me and I'll help.
The first 10 minutes
  1. screenshot
    add one folder

    Point it at one folder

    Choose one you know well — a few hundred items is plenty.

  2. screenshot
    tag a person

    Tag one person

    Add a person under Cast and tag them on a few items.

  3. screenshot
    the person's page

    Open that person

    Everything you tagged, from every folder and drive, in one view. Got a second drive? Add a folder from it and watch the page fill in from both.

That's the whole idea. Then tell me where it broke →

Free while it's in development. A paid tier later would be a one-off licence, never a subscription — and your .vdb always opens in any SQLite tool.