Storyboarding is iteration with structure. You start with the script and what the director is chasing: emotional beats, camera language, how characters move through space. Your reference work is shaped by questions that change every time you present: "Does this angle feel rushed?" "Can we see her face here?" "Is this wide shot holding too long?" A reference system that supports that means tracking decisions at two levels at once — what the shot is, and whether you kept it.
Coii Ref handles storyboarding because shots live as files in folders the whole time, tags layer on top rather than requiring you to rebuild organization when direction shifts, and canvases are JSON files you can move into a Slack thread, drop into revision history, or lock into an approved set without remaking them.
This is what a week of refinement looks like when the framework stays stable and everything else moves.
Monday morning — Scene 12 first pass: blocking the diner confrontation
The script reads: INT. ROADSIDE DINER - MORNING. A character confrontation. Twenty seconds. Two actors, two sides of the counter. You have a folder of reference—diners you've photographed, film stills, location scout images from the production's shoot.
Open the workspace. The project folder is /Volumes/ProjectDrive/Editorial/ and
it holds Reference-Shots/, Storyboards/, and Previs-Footage/. The app
writes its database and proxies beside them.
Start fresh with a search for this location:
tag:location:diner type:image in:/Volumes/Reference-ShotsForty-seven images. Diners in daylight, night, morning light. Wide shots of the space, close-ups of details. Some from reference libraries, some from production scouts, some from films.
Pull up a canvas called "Scene 12 Blocking" and start laying out the sequence. You need: wide shot of the space, medium on the actor at the counter, their angle on the other actor, the other actor's angle back. Four beats minimum.
Drag the wide shot onto plate one. It shows you the diner depth and counter position. Plate two: medium on actor A at the counter, looking toward actor B. Plate three: reverse angle from actor B's side. Plate four: close-up on the counter where something gets set down—the prop beat that punctuates the argument.
You have the frame sequence. Rate the wide shot 5 (keeper). Rate the mediums 3 (works, refine). The close-up is 4 (maybe better framing exists).
Monday, 10 am — Search for tighter actor angles
The director wants to see the reaction more clearly on actor A. You need a shot that is medium-to-close, A's face taking up more frame, but still enough counter space to see what they're reaching for.
Search:
tag:angle:medium-close tag:actor-face rating:>=3 type:image in:/Volumes/Reference-ShotsYou have eight images. Three are diners, three are restaurant interiors. Two look promising. Drag the one with the tighter framing onto a second canvas called "12 Revision A" — you're not replacing the blocking canvas, you're making a branch.
Monday, 2 pm — Cross-cut testing: does the rhythm work if we cut faster?
The editor on the previous episode did something interesting with this kind of confrontation scene: tighter cuts, shorter holds. Search for that show's color notes and shot language:
tag:show:previous-episode type:image tag:angle:tight rating:>=4Fourteen images. These are not diners — they're different locations — but they have the shot language you're tracking: fast cutting, close faces, reaction time built small. Pull four of them onto a canvas called "Rhythm Reference" and compare their timing against your diner sequence.
Rate each one: how long does this close-up hold before the cut lands? How is the reverse angle composed? Make a note: "Fast cuts 0.5 sec hold / Slow cuts 2.5 sec hold." You are not copying the rhythm; you are measuring it.
Tuesday — Scene 14: an outdoor approach across open ground
New scene. The character walks across a field toward a house. Forty seconds, establishing shot into closer approach, land at the front door. The production has not location-scouted this yet, so your reference is research.
Search your archive for approaches:
type:image tag:mood:approach tag:distance:wide-to-close added:>2026-08-01 rating:>=3These are images you have been collecting: paths through fields, driveway approaches, the specific light of a figure crossing open ground toward a structure. Fifteen images.
Canvas: "Scene 14 Approach Sequence." Plate one: the field wide shot, showing depth. Plate two: 80 percent of approach, closer but still in landscape. Plate three: driveway entry, the house filling more of frame. Plate four: front door arrival.
This is not a storyboard yet. This is the shot progression that the storyboard will follow. Show the director and ask: "Does this pacing feel right? Do we hold the wide shot longer?"
Rate as you edit: wide shot 5 (perfect), the mid-distance angles 3 (find better ones), arrival 4 (works but tighten it).
Note in the app: "Need better mid-distance angle. Actor moving left-to-right."
Tuesday, 4 pm — Find the actor approach angle
Search for that specific piece:
tag:subject:figure type:image tag:motion:walking tag:composition:centered in:/Volumes/Reference-Shots rating:>=4Thirty-two images. Centered figures walking toward camera in field settings. You need the one where light and field depth are closest to what the diner location scout shot. Compare them on a canvas called "14 Figure Approach Options" and rate the closest match 5.
Wednesday — Lighting pass: matching continuity across locations
You have been collecting light references all week: diners at different times of day, fields in morning and late afternoon, interior transitions from bright to dark. Search everything tagged with lighting notes:
tag:lighting:golden-hour type:image rating:>=3 added:>2026-09-01Organize by scene:
tag:scene:12 tag:lighting in:/Volumes/Reference-ShotsCreate a canvas called "Lighting Continuity Check." Plate one: scene 12 diner light. Plate two: scene 14 field light. Are they plausible in the same day? Does the transition work? Is one too blue, one too warm, or does the script jump enough in time that the shift feels intentional?
Rate together as a continuity check. Five stars if the lighting logic works. Four if you need to adjust.
Thursday — Director presents feedback
You show four canvases:
- Scene 12 Blocking (initial)
- Scene 12 Revision A (tighter actor angle)
- Scene 14 Approach Sequence (pacing test)
- Lighting Continuity Check (day logic)
Director feedback: "12 blocking is good, but let's go with Revision A. Scene 14 is too slow—cut the mid-distance angle, go wide straight to close. And the lighting — can we make the approach happen in magic hour instead of morning? It feels more urgent."
Three notes. No rebuild. You have the canvases and you can pivot.
Thursday, 2 pm — Apply feedback: new versions
Scene 12: The revision canvas was already your second take. Rate all the old blocking shots 2 (archive, not active). The Revision A images stay 5. Save a new version: "Scene 12 Approved" with only the final four frames.
Scene 14: The director wants wider approach, then close-up at door, skipping the mid-distance. Search again:
tag:scene:14 type:image tag:distance:wideFind the perfect field approach. Rate it 5 on the canvas called "14 Director Cut." No wide-to-close progression—just wide, then land close at the door.
Lighting: Search for magic hour references:
tag:lighting:golden-hour type:image tag:mood:approach rating:>=3Compare them to your diner lighting. Are they warm enough? Do they read as the same day? Save a new continuity canvas with magic hour field shot plus the original diner angles.
Friday — Building the scene archive
You have been through one full revision cycle. Four scenes are locked (or locked enough for previs). Create a view that pulls everything approved:
tag:scene:12 OR tag:scene:14 OR tag:scene:13 OR tag:scene:15 rating:>=4 -is:favouriteThirty-two images. These are your frame-locked sequences. Export them. This is what goes to the editor so they know your shot intent.
Separately, save another view for your own reference:
tag:location:diner OR tag:location:field type:image rating:3These are the "maybe still useful" takes—not approved, not archived, but the ones you might return to if the director asks for an alternative.
A two-week revision: when one scene becomes two
Three days after lockdown, the director asks: "Can we split scene 12? Right now it's 20 seconds and it feels long. What if the confrontation starts in the diner but finishes outside?"
This is a rebuild, not a tweak. But you have everything tagged. Search:
tag:scene:12 type:image in:/Volumes/Reference-ShotsSplit the old sequence: keep the diner opening, archive the climax, search for outdoor confrontation framing:
tag:mood:confrontation type:image tag:distance:medium in:/Volumes/Reference-ShotsNew canvas: "Scene 12b Exterior Continuation." You already have a confrontation library because you tagged it. The split scene is faster to build than the first take because the reference system is organized for exactly this kind of pivot.
Managing versions: keeping scripts and shots in sync
The script is on version 12 draft. Your boards are tagged by scene number, which does not change, and by version date.
tag:version:draft-12When the script jumps to draft 13 and scenes get renumbered, you do not reorganize. You tag the new version:
tag:version:draft-13Search for scenes that existed in draft 12 but have a different number now:
tag:scene:12 tag:version:draft-12A scene that was 12 is now 11. Add the tag tag:scene:11 without removing
tag:scene:12. Queries using the old scene number still work for historical
reference, and the new number labels the current cut.
Canvases live in the project folder as JSON. Name them by version:
12-diner-blocking-draft-11.coii-canvas12-diner-blocking-draft-12.coii-canvas(revised)12b-exterior-continuation-draft-13.coii-canvas(new scene from split)
Three files, three decisions, all recoverable.
Where this scales to feature-length
By the end of a six-week storyboard pass, you have boards for fifty to eighty scenes, each with three or four versions, tagged by scene number, version date, location, lighting mood, shot language, and approval status.
The reference folder holds two thousand images: production scouts, film stills, archive research, color studies. A search pulls the right subset for any question: "Show me everything approved for exteriors that is tagged magic hour" or "Show me wide shots from draft 10 that the director rated as keepers."
Saved views become reference templates. When you start a similar scene—an outdoor approach in a different context—your previous approach sequence is one query away. The images are tagged; you are not starting from scratch. What is a reference manager answers whether storyboarding really requires one, or whether simpler tools suffice.
Boards move to a shared drive or locked project folder. Each one is a plain JSON file with a date in its name, so the full decision history is in git or your file system without needing a separate version control tool.
Comparing approved directions: why saved views matter
By week 2, you have explored multiple directions. The question directors ask most is: can I see all three side by side? Building a moodboard from existing folders covers this exact comparison workflow. A saved view becomes your most important tool. You save:
tag:direction-a rating:>=3
tag:direction-b rating:>=3
tag:direction-c rating:>=3Each pulls up every image rated high enough on that direction. Open them on
adjacent monitors and you have the comparison without reopening files. When the
director says "hybrid — direction B's silhouette but direction A's fabric," you
tag the winning images tag:direction-final and rate them 5. Everything else
stays archived and findable.
Rating by version prevents folder chaos. An image can be both version:draft-10
and archived:direction-a. Moving between states is one tag change, not one
folder move. When you want every approved frame for scene 12 across all versions:
tag:scene:12 rating:>=4 -tag:archivedOne query finds it all.
Rate as you edit, not later
Rating is not cleanup. It is the only thing that separates a searchable library from a folder of images you scroll through. Rate as you drag an image onto a canvas: keeper-quality, revision-quality, archive-quality, or still-exploring. By the end of week one, your best frames are marked, rejected ones are marked differently, and unseen images are at default. A search for any version brings up the keepers first. That is the difference between a professional workflow and hope.
At the end, the editor receives boards organized by scene, with multiple takes where decision was hard, and every image rated so they know which one the director held for longest.
How to organize references by project covers the folder structure that makes this work at feature scale. How to rate and filter covers tagging strategies for film and animation reference. How to build a moodboard is the technical step if you have not used canvases before. What is a reference manager answers whether you need one for previs at all. And the reference manager for storyboard artists is the feature list if you want to know what tools live inside this workflow.
For a broader comparison of how DAMs and reference managers differ in scope and speed, see DAM vs reference manager — it explains why storyboard reference lives in a reference tool instead of a larger asset management system.