A frame you grabbed six months ago is only useful if you can say why you kept it

Film stills rot fast in a library that only sorts by date added. You save a frame because of what a director did with light, or blocking, or a colour a production designer chose on purpose — and none of that survives a folder named Screenshots next to four hundred others from the same week. The fix is not a better folder structure. It is tagging what made the frame worth keeping, once, at the point you save it, so a query line can find it again by the reason rather than by the date.

This is a different habit from collecting stills off the web. A frame you grab yourself from a film you're watching carries information a downloaded still never does — you know the scene, the minute mark, what happens before and after, whether the framing was a deliberate choice or an accident of where the camera happened to be. Losing that context the moment the frame lands in a folder is the actual problem this workflow solves, and it solves it by capturing a little of that context at the moment you already have it, rather than trying to reconstruct it later from the image alone.

0. Grab frames at a resolution worth keeping

Whatever you're using to pull a still off a film — a screen recording tool, a frame-accurate player with an export function, a photo of a projected frame if that's genuinely what you're working from — grab at the highest resolution the source allows. A still you'll want to study for lighting or composition needs the detail; a still you're keeping only as a memory jog for "there's a shot like this in this film" doesn't, and it's fine to grab that one smaller. Decide which kind you're making before you grab it, not after, because re-grabbing the same frame from the same film a second time because the first export was too small is the single most avoidable waste of time in this whole workflow.

1. Keep the source film in the filename or the folder, not just your memory

Rename the frame, or the folder it lives in, with the film and the scene before you forget which is which — there-will-be-blood/derrick-fire-01.png rather than IMG_4821.png. Coii Ref indexes files where they already sit, so this step happens in Finder, once, and the workspace picks it up on the next scan. A folder per film also gives you a natural unit to point a canvas at later, when you are pulling a board together for a specific look rather than browsing the whole collection.

2. Tag by what the shot is doing, not by the film a second time

The folder already says which film a frame is from, so the tags should carry information the folder cannot: tag:handheld, tag:wide, tag:silhouette, tag:practical-light. Tag the technique and the mood, and a search across your whole library — every film you have collected from — starts answering questions like "what did handheld coverage of a fight scene actually look like" instead of only "what is in this one folder."

tag:any:handheld,steadicam rating:>=4      camera movement, the good ones
tag:silhouette -is:favourite                good but not yet starred
type:image tag:practical-light              lit by something in frame
has:notes no:tags                           written up, not yet filed

3. Rate on a second pass, not the first

Do not rate while you are grabbing frames — you are pattern-matching on "interesting," which over-rates half the folder. Come back a day later and rate against a narrower question: would I actually put this on a board for a project. A four-star pass done cold catches what enthusiasm missed the first time, and it is the rating that makes rating:>=4 mean something later. If rating consistently is the part that usually falls apart after the first enthusiastic week, rating and filtering your best references goes into the habit in more depth than fits here.

4. Write down what you'd forget by the time you search for it again

A note is the only place to put the thing a tag cannot express: the scene this is from, what happens two shots before and after it, or why the lighting is doing what it's doing. has:notes becomes a genuinely useful filter once a fraction of your stills carry one — it separates the frames you understood at the time from the ones you only liked. Writing notes on reference images covers what makes a note worth writing versus one you'll skim past in a year.

4.5 Keep a short list of technique tags rather than inventing one per shot

Decide on maybe fifteen or twenty technique and mood tags before you start grabbing in earnest — handheld, static, wide, close, silhouette, practical-light, backlit, symmetrical, and whatever else genuinely recurs in what you find yourself drawn to. Resist adding a new tag for every frame that doesn't cleanly fit an existing one; instead, ask whether it's close enough to an existing tag to use that one, or whether a note is the better home for what makes it different. A short, consistently applied vocabulary is worth far more at search time than a long one nobody remembers to reuse.

5. Put a working set on a canvas, and let the rest stay in folders

Not every frame belongs on a board — most of your library is reference material you might never look at again, and that's fine, it's still findable by query. When you're actually building a look for a project, pull the handful that matter onto a .coii-canvas next to the folder for that project. It's a plain JSON file, so it stays with the images it references and opens the same way on another Mac.

6. Let video stills and frame grabs from clips live in the same search

If you're keeping actual clips alongside your frame grabs — a scene you might re-grab from later, or a reference you want to scrub through rather than look at as a still — index the clip in the same folder. It's read on the same pass, gets the same tags and rating, and a proxy is written at canvas resolution so a board mixing stills and footage scrolls evenly. type:video ext:mov tag:practical-light finds the clip version of the same lighting setup you tagged as a still.

7. Cross-reference across films when you're actually building a look

The real payoff of tagging by technique rather than by film shows up when you're pulling reference for a project and the answer isn't "show me everything from this one film" but "show me every practical-light shot I've ever saved, from any film, so I can see the range." A query across the whole library does that in one line, something a folder-per-film structure alone can't answer no matter how well it's organized:

tag:any:practical-light,silhouette rating:>=4    the whole library, one look
in:there-will-be-blood tag:wide                  narrowed to one film

That's the actual argument for tagging technique instead of just filing by film: the folder answers "what's from this film," and the tag answers "what looks like this," and a real reference pull for a project almost always needs the second question answered across everything you've ever saved, not just the one film you happened to grab it from.

8. Keep a running note on what you were actually watching for

If you're working through a director's filmography or a specific period of film history with intent, a short note on the folder itself — not on any one image, just a line in a readme beside it — about what you were looking for that pass saves you from re-explaining it to yourself the next time you open the folder. This is a small habit and it's the one people skip most, because it doesn't feel like it's part of the actual collecting. It is.

What to do when the folder is thousands of frames from dozens of films

A film-stills habit compounds faster than most reference libraries, because one rewatch with intent can produce two hundred frames in an evening, and a determined weekend can produce two thousand. At that scale, the folder-per-film structure stops being optional — it is what keeps tag:handheld from returning six hundred results with no way to narrow by film — and tagging on the way in, rather than batching it later, is the difference between a library you trust and a backlog you avoid opening. Six months of batched, unlabelled grabs is not a weekend project to fix retroactively; nobody remembers which film three hundred unlabelled frames came from, which is exactly why the folder-and-tag habit above needs to start on day one rather than be adopted once the pile is already unmanageable.

The watcher keeps the database honest as you keep renaming and reorganizing folders per film, so tidying up in Finder six months from now does not cost you the ratings and notes you already wrote — a folder you split into two, or a film you decide deserves its own subfolder for two different cuts, is just a rename and a move as far as the watcher is concerned, not an event that requires re-tagging anything.

At real scale, the tag vocabulary itself becomes something worth being deliberate about. A handful of technique tags used consistently across hundreds of films — handheld, wide, silhouette, practical-light, and whatever else actually recurs in what you collect — beats fifty tags used once each, because the whole value of tagging is being able to ask "show me everything like this" across a library too large to browse by eye. If a tag only ever applies to one still, it isn't really a tag; it belongs in that still's note instead. Deciding which of the two a given piece of information is — reusable category, or one-off context — is most of what makes a tagging system still work at a thousand films instead of ten.

If you're starting from an existing pile of ungrabbed frames sitting loose in one folder, building a library from folders you already have covers the first index specifically, including what to do about the material you've already collected without any of this structure in place.

One more thing worth deciding early rather than late: whether you're keeping stills only, or stills alongside the clips they came from. Both are valid collections and they don't need to be the same one — a lot of people keep a lean folder of curated stills for boards and moodboards, and a separate, much larger folder of raw clips they might re-grab from later. If you go that route, tag both folders with the same technique vocabulary so a search across the whole library still returns both kinds of result together, rather than requiring you to remember which folder to search first.

For collections that lean more toward footage than stills, organizing video references goes further into proxies and clip-level tagging. If your current system is Eagle folders with keywords, the comparison with Eagle covers what carries across and what a switch would cost. And if tagging consistently across hundreds of frames from dozens of directors is the part you're not sure about yet, tagging reference images consistently is worth reading before you're three hundred frames in and inventing a new tag for every one.

None of this requires paying for anything to try it — the trial runs for thirty days with every feature on, which is enough time to grab frames from an actual project and see whether the tags you chose on day one still make sense on day twenty.

Thirty days is also enough time to notice whether the folder-per-film habit actually survives contact with a real project deadline, which is the more honest test than trying it out on a slow week with nothing due. Grab frames from whatever you're actually watching for work this month, tag them as you go, and by the end of the trial you'll know whether the system above is one you'd keep using or one that only worked because you were being careful on purpose.