ClipCatalog logo ClipCatalog
EN

Find B-roll by what's on screen.

Windows 100% local video processing Free trial · No time limit

You know the shot is somewhere in your archive — but pulling it out means scrubbing through hours of footage on the off-chance you'll recognize it. ClipCatalog auto-tags every clip on your PC the first time you index a folder, then lets you stack a couple of tags — sunset + ocean, or crowd + concert — to pull every match from every shoot you've ever filmed. Seconds instead of hours.

No per-clip pricing. No cloud uploads. One $99 license to make every clip you own browseable by what's actually visible in it.

Filter local video clips by automatically detected scenes, objects, and locations using the Detected Content filter — no manual labeling required.

Library-wide, not per-video

Tools like OpusClip cut clips out of one video at a time. ClipCatalog goes the other direction — it indexes your whole back-catalog so you can pull shots that match a tag combination from across every shoot you've ever filmed.

Tag vocabulary, not free-text guessing

Tag search uses a discrete, browseable vocabulary you can actually drill into — so you know what's in your library before you query it. For the rare shots the vocabulary doesn't name, switch to natural-language search with Relaxed/Balanced/Strict strictness. Two retrieval modes, one local index.

Pay once, no per-clip fees, no cloud

Cloud DAMs and AI clip services bill per asset, per seat, per minute, or per month. ClipCatalog is a one-time $99 license — re-index a folder any time at no extra cost, and never ship your footage to a third party. Your tag index is yours, on disk, encrypted.

The 'I know I shot something like this' problem

Tag search flips B-roll work on its head: instead of starting with a 12-drive haystack and hoping to recognize the right shot mid-scrub, you start from a shortlist of clips the AI has already flagged as matching your filter. The same four-second coffee-cup cutaway that used to take an hour now lives one Match-All click away — because every clip is pre-tagged the moment it's indexed.

Without tag search

  • Twelve drives of source footage, no consistent naming, no way to search visually
  • Manual logging would take weeks and goes stale the moment the next shoot lands
  • Cloud DAMs charge per asset, per seat, per month — and want your footage on their servers
  • The shot you need is on the SSD you mailed back to the client

With ClipCatalog

  • Open the Detected Content filter, see the tag vocabulary the AI built from your own footage, and start from a shortlist
  • Stack the right tags, hit Match-All, and a result strip of candidate shots appears in seconds — no scrubbing, no guessing
  • Drag the clip you need straight into Premiere, DaVinci Resolve, or Vegas Pro from the result strip — files never leave your machine
  • Future shoots auto-tag in the background, drives stay searchable even when unplugged, and no monthly bill ever shows up

How searching B-roll by visual content works

Three things have to be true for visual B-roll search to actually replace scrubbing: tags worth searching, library-wide coverage, and a query path that doesn't require remembering filenames. ClipCatalog handles all three locally.

Detected content →
1

Point at your footage folders

Add one folder or several — internal drives, external SSDs, archive NAS. ClipCatalog scans for video files and queues each one for local AI tagging. Your folder structure stays untouched.

2

Local AI tags every clip

A vision model runs on your hardware — GPU when available, CPU fallback otherwise — and produces a structured tag set per clip: scenes, objects, actions. Nothing is uploaded. See the AI visual tagging feature page for how the tag vocabulary is generated and which model size to pick.

3

Pull B-roll by tag

Open the Detected Content filter, pick a tag like beach, or combine beach + sunset with Match-All to narrow further. For shots the tag vocabulary doesn't name, switch to natural-language search and type a free-text description — set strictness to Relaxed, Balanced, or Strict.

Example searches that become easy

Tags are discrete and enumerable, so you can browse what's actually in your library — not just guess at it. The Detected Content filter handles single-tag and multi-tag queries; combine tags with Match-All (AND) or Match-Any (OR). For shots that don't fit any tag in the vocabulary, the natural-language filter takes a free-text description.

sunset — every dusk shot across three years of travel footage (single-tag search)
crowd + concert — wide audience reactions for the live-music recap (Match-All)
kitchen + cooking — usable cutaways for the food channel's next episode (Match-All)
cityscape OR skyline — every establisher across your urban work (Match-Any)
dog + beach — pet shots near water for the client reel (Match-All)
Natural-language: a person walking alone on a foggy road — a shape the tag vocabulary doesn't name, retrieved by free-text description with Balanced strictness

Who searches B-roll by visual content?

Anyone who maintains a growing back-catalog of footage they keep dipping back into. A few real shapes:

Editors with multi-project archives

Three years of corporate, documentary, and brand work on rotating drives. Pull every interior office shot, every product close-up, every street establisher — without remembering which client paid for which shoot.

YouTubers needing reusable B-roll

Every cooking video, gym session, or city walk you've ever published is also raw cutaway material for the next one. Tag it once, reuse it forever — no per-clip subscription.

Filmmakers cataloging raw footage

A documentary feature with four hundred hours of source. Find every shot of the subject's hometown, every workshop interior, every transitional landscape — without per-tape manual logging.

Course creators pulling visual examples

Hundreds of recorded lessons plus screen-share and field footage. Find every clip that visually demonstrates a concept — a whiteboard, a city street, a piece of equipment — and drop it into the next module.

Brand-content teams reusing shoots

One product shoot becomes a year of social cutdowns. Find every clip of the new bottle, the lifestyle b-roll, the hero close-up — without re-watching the whole take.

Stock-shooters maintaining a personal archive

A backlog of raw self-shot footage you slowly mine for portfolio pieces. Pull every silhouette, every empty street, every aerial — and decide what's worth finishing.

What to expect from B-roll visual search

ClipCatalog's tagging pipeline is designed to be practical and honest. Here's what's true before you start.

Speed depends on your hardware

A capable GPU makes initial indexing fast; CPU-only is slower but works. Either way, it's a one-time cost per clip: once the library is indexed, tag searches return their first results in seconds.

Windows only for now

ClipCatalog is currently available for Windows 10 and 11. A GPU helps but isn't required — the app picks the faster option for your hardware.

Why local-first matters for B-roll archives

Unreleased footage is some of the most sensitive material a creator handles. Client deliverables under embargo. Unreleased product shots. Footage of people who haven't signed releases. A cloud DAM that uploads every clip for tagging is asking you to trust their access controls — and their billing department's continued goodwill — forever.

ClipCatalog tags clips on your hardware. The video stays on the drive. Tags live in a local SQLite database on your machine. Nothing leaves until you choose to share it.

If you're comparing local-first video tools side by side, see the privacy-first video management roundup for how ClipCatalog stacks up on AI tagging, semantic search, and library-wide retrieval.

Searching B-roll by visual content — FAQ

Does this upload my footage anywhere?

No. Tagging runs entirely on your machine using a bundled local vision model. Once the model is on disk, no network is needed for indexing or search.

How is this different from OpusClip or similar clip-extraction tools?

OpusClip and similar tools turn one long video into shorter clips. ClipCatalog goes the other direction — it indexes your whole library so you can pull matching B-roll from across every shoot you've ever filmed.

What's the difference between tag search and natural-language search?

Tag search uses a discrete, enumerable vocabulary — you pick from the tags the AI detected, combine them with Match-All / Match-Any, and get exact matches. Natural-language search takes a free-text phrase and ranks clips by semantic similarity with three strictness levels: Relaxed, Balanced, or Strict. Both run locally; use whichever fits the shape you're after.

Can I narrow results further once I have a tag-search shortlist?

Yes. Layer tags with face filters, transcript filters, footage-type filters (dialogue / voiceover / scenic), date range, camera, resolution, and so on. Each filter you add cuts the result list down.

Does the free trial include AI tagging?

Yes — up to 500 videos and 10 hours of total duration, with full access to all features including tag search, face detection and grouping, and natural-language search. No account or credit card required.

Does it work on Mac or Linux?

ClipCatalog is currently available for Windows only (Windows 10 and 11). Mac and Linux support is not on the near-term roadmap.

Relevant comparisons

If you are evaluating this workflow against other tools, start with these side-by-side pages.

Try ClipCatalog free — up to 500 videos

No account required. Your footage stays on your computer.

500 videos free No credit card · no account 100% local — footage never leaves your PC