Fast Video Cataloger 10.4 Lets Editors Find Scenes by Describing Them

New visual search finds shots in a local video library from a plain-language description, with nothing tagged in advance and no footage uploaded.

STOCKHOLM, SWEDEN, October 2, 2026 /EINPresswire.com/ — VideoStorm has released version 10.4 of Fast Video Cataloger, its Windows video cataloging software. The main new feature is visual search. An editor types what a shot looks like, for example “drone shot over a city at night”, and the program lists the scenes in the library that match best. It searches the images themselves, not keywords, so it also finds footage that nobody has tagged.

Finding a shot usually means scrubbing through hours of footage that was never logged, or was logged with other words than the ones the editor is searching for today. With visual search the editor describes the shot the same way they would describe it to a colleague. The description works together with the other filters in the Search window, so a search can be limited to one folder or one project keyword.

MORE SHOTS LIKE THIS ONE

When an editor has found one good shot, Find similar scenes sorts the whole catalog by how much each scene looks like it. That brings up the rest of a sequence, or every take from the same location. The best matches come first. By default each video gets a limited number of results, so one long shot does not fill the list.

The search index is built on the user’s own computer while videos are indexed. Older catalogs can be updated in a separate pass that shows how much of the library is covered and can be paused and resumed. The footage and the index never leave the user’s machines. With a shared catalog the search runs on the Fast Video Cataloger server. The server also makes it available to scripts through the REST API and to AI assistants through MCP.

“You remember what a shot looks like long before you remember what you called it,” said Fredrik Lönn, co-founder of VideoStorm. “Visual search lets you look for it the way you remember it.”

TRAINED TAGS THAT LEARN FROM CORRECTIONS

Version 10.4 also improves Trained Tags, which were introduced in version 10.3. A trained tag is a custom AI keyword that the user teaches the software from a handful of example thumbnails. A wrong match can now be right-clicked and marked Not this. The keyword is removed, and the tag stops matching similar frames. The Trained Tags window also points out examples that make a tag worse, such as near-duplicates.

AVAILABILITY AND PRICING

Fast Video Cataloger works alongside editing programs such as Premiere Pro and DaVinci Resolve. It does not replace them. It is where editors search and organize their footage, with visual browsing, structured metadata and everything stored locally. Version 10.4 also has many smaller fixes and improvements. The full list is in the version history in the user guide.

Fast Video Cataloger 10.4 is available now for Windows 10 and 11. Pricing starts at $9.90 per month, $97 per year, or a one-time perpetual license of $197. A free 30-day trial is available at videocataloger.com.

Fredrik Lönn
VideoStorm Sweden AB
email us here
Visit us on social media:
LinkedIn
Instagram
Facebook
YouTube
TikTok

Find any shot by describing it – visual search in Fast Video Cataloger

Media gallery