Skip to content

Indexing

Indexing (a scan) is how ArchiveMind learns what is in a repository. This page explains the stages in plain terms and what to expect on large collections.

The stages of a scan

ArchiveMind processes files through a pipeline of stages:

  1. Discovery — walking the folder tree and listing every file.
  2. Classification — deciding what kind of file each one is (photo, document, audio, video, archive, or generic), based on its file extension.
  3. Hashing — computing a fingerprint of each file's contents. Hashes are what make duplicate detection possible.
  4. Metadata and text extraction — reading useful information from inside files: photo dates and camera details, document text, audio tags, and so on. Extracted text is what makes full-text search work.
  5. Thumbnails and previews — generating small preview images for photos and other supported formats.
  6. Optional visual and media enrichment — where enabled, building the image-understanding index used by visual search, and media previews such as audio waveforms (some of these need optional tools — see Audio and video).

You can search as soon as the scan starts — results appear progressively as files are discovered and processed.

How long will it take?

It depends, honestly, on two things:

  • How many files there are. A few thousand photos take minutes; a multi-terabyte archive of hundreds of thousands of files can take hours.
  • How fast the drive is. An internal SSD is much faster to scan than a USB hard drive or a network share. The drive's speed is usually the bottleneck, not your computer's processor.

Large collections may take considerable time — this is normal. The interface shows progress while enrichment work continues in the background.

0% CPU does not mean stuck

If ArchiveMind is running but using little or no CPU, that usually means it is idle (a scan finished, or it is waiting on a slow drive), not stuck. See Running but 0% CPU.

While a scan is running

  • You can keep using search — results improve as the scan progresses.
  • You can leave the window open or closed; the scan runs in the app's background process either way (remember: closing the window does not quit the app).

Do not delete the database while troubleshooting

If a scan seems slow or stuck, do not delete the catalogue database as a first step — you would lose everything already indexed, and the cause is usually something simpler. Work through Indexing appears stuck first.

Stopping and rescanning

A running scan can be stopped with the Stop Scan button — the stop is cooperative, so the scan winds down within a few seconds rather than being cut off mid-file. Stopping (or quitting the app) during a scan is safe: your original files are only ever read, and everything already written to the catalogue stays searchable.

There is no pause button, and a stopped scan does not resume from an exact checkpoint. What makes stopping cheap anyway is that rescans are incremental by default: when you scan the repository again, files that are unchanged since the last scan (same size and modification time) are recognised and skipped, so the new scan spends its time only on what is new or changed.