Folders, collections and quick filters: a library that keeps up with your footage
StoryFolder —
Part of: StoryFolder 0.4: AI Autofill, local transcription, and an assistant that can see your library
Last updated: 2 September 2026.
Folders file videos, collections hold hand-picked shots from across different videos, and a quick filter saves a search as a destination you can jump straight back to: what you typed, the filter chips you built and the card layout, together. Collections, quick filters and video-level metadata fields are new in StoryFolder 0.4; folders are not, and are here as context rather than as news. None of it manages where your media lives: StoryFolder indexes the videos you gave it, and which drive a file is on belongs to a different tool.
What is the difference between a folder, a collection and a quick filter?
What they hold, and whether their membership changes on its own.
| Holds | Membership | Changes when metadata changes | |
|---|---|---|---|
| Folder | Videos | You put them there; folders nest | No |
| Collection | Shots, from any number of videos | You add them by hand, with a count on the rail | No |
| Quick filter | Nothing; it is a saved search | Whatever matches right now | Yes |
| All items | The whole library | Everything | Not applicable |
A folder is the filing cabinet: a video lives in one, folders live inside folders, and opening one shows and searches that folder plus everything underneath it, not just its direct children. A collection is the pile on the desk: you select shots in the editor or the library, choose Add to collection, and they stay in that collection whether or not anything about them changes later. An open collection has its own search box and filter bar, so a big selects pile is still something you can type into. That is the point of a collection, and the reason it is not a quick filter under another name.
A quick filter is the saved question, and the whole question gets saved. Save as quick filter appears as soon as there is anything to save, a typed query on its own included; name it and it becomes a row in the rail with its own address. Arriving there puts the query back in the box and applies the chips and the card layout; leaving puts your own view back, so a quick filter is a place you visit rather than a state that follows you around. Change any of the three and the row offers to save the change. Results are recomputed on arrival, so a shot that has since earned the rating you filtered on is simply there.
How do you organise a real client archive?
Put the stable things in folders, the facts about the whole job in video metadata, the picks in collections, and the questions you keep re-asking in quick filters.
- In the library, create a folder and drag a project into it. Nest by account and year if that is how your work divides; folders do not follow disk paths and moving a project between them does not move a file.
- Open Settings → Video Metadata and add the fields that describe a whole job rather than a shot. Custom fields are on StoryFolder Pro.
- Fill those in once per project instead of once per shot. That is the point of them, and it is the fix for a schema where the client name was being typed on every row.
- Select shots across two or three videos and add them to a collection: the delivery set, the ones the client already saw, the ones you keep coming back to.
- Build the filter you rerun most, save it as a quick filter, and give it a name that says what it answers rather than what it contains.
Everything in the rail renames in place on a double-click and carries a right-click menu, and every folder, collection and quick filter has its own address you can keep as a link. Deleting a collection or a quick filter asks first; neither ever deletes a video.
What does one search actually cover?
The whole library, from All items, across four things at once: the titles and descriptions of your videos, the fields you type into on a shot, the fields you type into on the video itself, and transcript text. Every result names which of those it matched on, so a video that came back because somebody said a word on camera is visibly different from one that came back because you tagged it. / puts the cursor in the box from anywhere in the library, and ⌘F, Ctrl+F on Windows, does it even from inside another field.
Searching or filtering from All items covers the whole library, filed videos and all, while opening a folder narrows the same search to that folder and its subfolders. Finding it anywhere and finding it in this account differ by where you are standing and nothing else.
Matching is literal. Typing wide finds the shots with wide in a value, and leaves alone the shots that are wide but were never described. Type two words and both have to match, in any order and in the same value, so dock warehouse finds a Location reading Warehouse dock and dock forklift finds nothing. Nothing here inspects the picture. The index is only as good as what you or AI Autofill recorded, and the transcript half only exists for videos where somebody was talking, which is why searching by what is in a shot and what is said in it is one problem with two layers.
Text, dropdown, tag and number values are the ones the box reads, on a shot or on a video alike, so typing a client name finds the project it is filed against and typing 24 finds the shots whose Focal length says so. Rating and checkbox fields have no text to match and are reached with a filter chip instead. Number is new: it filters with is more than, is less than and is between, and it exports to a spreadsheet as a real number you can total. What each field type is for covers the rest of them. The chips and the box work together: filter to a client, then type the phrase. Results are ordered by why they matched, with the conditions you built ranked above a field hit and a field hit above a transcript hit.
Why keep video metadata as well as shot metadata?
Because some facts are true of the whole job and typing them on every shot is how a schema rots. Client, Job# and Status describe a project; Shot Size and Usable describe a moment.
Video fields are a flat list, with no categories to arrange, because their values sit on the video itself. You add them in Settings → Video Metadata, fill them in on the project, and then filter whole projects by them or find them by typing their value into the same box everything else goes into.
The boundary worth knowing is that AI Autofill does not touch video fields. There is no Autofill switch in that pane at all, because a client name is not in the frame. Those facts are typed once by the person who knows them.
How do quick filters cut a 461-shot review down?
By turning the criteria into the destination, so the queue is already small when you arrive rather than something you assemble again each time.
Filter to the client, then to the shot type you are auditing, then save it. The next time that question comes up it is one click on the rail, and if the underlying values have moved the answer moves with them. Two things make that trustworthy in practice: the result count, which tells you immediately whether the filter is doing what you think, and the reason line on each result, which tells you why a particular shot is in front of you.
The queue carries through to what you hand over. An export or a publish takes the shots your current query and filters leave standing, so "filter to four stars, then publish the selects" gives you the selects.
Where a quick filter is the wrong tool is when the set has to stop moving. A delivery list that keeps changing because somebody edited a field is not a delivery list. That is a collection, chosen by hand and staying chosen.
When do you need something bigger than this?
When the problem is the files rather than the footage. Shared permissions, ingest from a watch folder, proxy storage other people can pull from, a catalogue of which drive holds what, or a round trip back into an NLE are all outside what a local visual index does.
StoryFolder indexes the videos you import, not your storage, and it does not know or track where the originals live. What a three-person team actually needs before reaching for an asset manager is a longer answer, and it is often "less than you think".
FAQ
Are folders new in StoryFolder 0.4? No. Folders predate 0.4. Collections, quick filters and video-level metadata fields are the new library pieces.
Can a collection hold shots from more than one video? Yes, that is what it is for. Select shots anywhere, choose Add to collection, and the rail shows the collection with its shot count. An open collection has its own search box and filter bar.
Does searching from All items cover every project? Yes. A search or a filter run from All items covers the whole library including videos filed in folders; opening a folder narrows it to that folder and its subfolders.
What does a quick filter actually save? The whole search: the query, the filter chips and the card layout, and a query with no chips can be saved on its own. Its results re-answer on arrival, where a collection stores the shots themselves and does not.
Is library search fuzzy? No, it is substring matching against your text, with every word of a multi-word query having to match in any order. The transcript editor's own search box is the fuzzy one.
Can I search by a Client or Job number kept on the video? Yes. Text, dropdown, tag and number fields are searchable by typing whether they sit on a shot or on the video; rating and checkbox are filter-chip only.
Do I need Pro for any of this? Collections and quick filters are free. Custom fields, shot or video, are on Pro.
Does StoryFolder track which drive my footage is on? No. It indexes the videos you import, not your storage.