Turn Original Storyboard Panels Into a Timed Commercial Animatic
Turn Original Storyboard Panels Into a Timed Commercial Animatic
A storyboard says what happens. It doesn't say how long any of it takes.
An animatic is where you make that second claim. You give each original panel a provisional duration, set a little temporary sound against it, and watch whether the sequence actually says what you thought it said. Ten seconds or thirty, the number matters less than this: the number came from you, and you can change it.
Two things the animatic is not. It is not a folder of images played at an assumed rate — that's a different import behavior with a very different result, and it's the easiest way to lose an afternoon. And it is not proof. A timed study can show that a proposed order of events is legible in ten seconds. It cannot show that an actor, a camera, a rolling jar, and a sound recordist can produce that ten seconds on a real day.
Everything below about application behavior comes from published documentation rather than a session I ran. Where the exact wording of a setting decides the outcome, confirm it in your version; menu labels move.
Prepare panels that can survive an edit
Start with the scene, not the software. You want a setup, a change, and a response: something is normal, something disturbs it, somebody reacts. If your board has a mood but no change, the animatic will faithfully reproduce a mood you could have described in a sentence.
Then give every panel an identity that outlives its position. Not panel_003.png — that name means "third thing in the folder," and the moment you reorder or swap, it lies. Use something like SH03_jar-roll_revB.png. The identity travels with the drawing; the position lives in the timeline, where you can move it without breaking the paper trail.
Keep the storyboard note attached to the panel. A board panel usually carries a caption — a line of action, a fragment of dialogue, a camera indication. That note is the reason you chose a hold length, and six weeks from now it will be the only way to reconstruct why. A plain text or spreadsheet hold sheet, one row per panel with its identifier, note, and current duration, does more for a revisable study than any amount of timeline tidiness.
Before you add anything musical or decorative, write down the timing question in one sentence. Not "does this feel right" — something a colleague could disagree with. Then mark what the drawings genuinely cannot establish: how a performer's weight shifts on a turn, how fast a hand actually moves, whether the camera drifts in the middle of a hold. Those marks aren't disclaimers. They're the list of things the study is going to leave open on purpose.
Import held stills rather than accidentally creating a frame sequence
The trap is in the import dialog.
If you select a numbered folder of panels and let the application treat it as an image sequence, each image becomes one frame. Adobe's page on importing numbered images as a sequence — the version of record updated 7 January 2026 — describes exactly that behavior: the images occupy consecutive frames of a movie. At 24 frames per second, eight panels become a third of a second. Your commercial is a blink.
The alternative is to import the panels as individual still images and set each one's duration yourself. Adobe's still-image import documentation, updated 1 April 2026, covers setting a still's duration and changing it for use on the timeline. That's the route you want: one panel, one deliberate hold, adjustable after the fact.
Whether you use a numbered sequence or separately held stills is the whole decision. A numbered sequence can be useful if you've genuinely rendered every frame of an animation — that's a different job with a different purpose. For storyboard panels, it throws away the thing you came for.
Premiere is one route, and it's the one I'll describe, because mixing steps from several applications produces instructions that work in none of them. If you work in a storyboard tool instead, Krita's Storyboard Docker documentation records a different model: scenes carry durations in seconds and frames, and reordering scenes affects their associated keyframes, while deleting a scene can preserve keyframes and redistribute the duration. It's an honest alternative — the drawing tool and the timing tool are the same tool — but pick one and finish the job in it.
Whichever you choose, set the sequence frame rate before you import anything, and write it down. The holds in the next section assume 24 frames per second, and that assumption is doing arithmetic work.
Give each beat time for the information it carries
Holds aren't emotional units. They're units of information and attention.
A panel that establishes where two people stand relative to each other needs enough time for a viewer to build that map — roughly, long enough that you stop reading the drawing and start waiting. A panel that shows a hand already in motion needs much less. A reaction needs enough time for the recognition to land, and no more, or it starts to look like the character is deciding something.
Here's the example. It's invented for this article: no panels were drawn, no timeline was built, no audio exists. It's specified in enough detail that you can construct it and check the arithmetic yourself.
The scene: a small kitchen. A jar on a shelf above the counter gets knocked loose by a cabinet door, lands on the counter, and rolls. Nia, working at the counter, tries to catch it. Dee is at the pass at the far end of the counter — the end the jar is rolling toward — with her back turned. The board, in eight panels:
- SH-01 — Wide: counter at left, pass at right. Nia at the counter; Dee at the pass, back to the room.
- SH-02 — The shelf; the cabinet door touching the jar; the jar tipping at the shelf edge.
- SH-03 — The jar on the counter, in profile, rolling toward the right of frame.
- SH-04 — Nia's head turning toward frame right.
- SH-05 — Nia's right hand reaching into frame right, fingers open.
- SH-06 — Dee's head beginning to turn over her shoulder.
- SH-07 — Nia's hand closing on the jar at the counter's edge; the jar a few degrees off vertical.
- SH-08 — Nia upright, jar in hand; Dee looking at her; Nia's expression composed.
The timing question, written before any sound: in the first four and a half seconds, does a viewer register both that the jar is rolling toward Dee and that Dee has not turned around yet?
Version A holds the setup generously — 2.00 seconds on SH-01, 1.50 on SH-02, 1.00 on SH-03 — then SH-04 at 0.75, SH-05 at 0.75, SH-06 at 1.00, SH-07 at 0.50, and a long 2.50 on SH-08. At 24 fps that's 240 frames, 10.00 seconds exactly.
Temporary sound: two elements and nothing else. A small clack on SH-02 when the door touches the jar, and a low scrape under SH-03 for the roll. No music. That's deliberate — an energetic bed would make both versions feel equally urgent and cover the exact thing you're trying to hear, which is whether the cause is present in the first place.
Compare timings, replace one panel, and inspect the export
Build version A. Watch it once, then watch the first four and a half seconds three more times and ask only the question you wrote down.
If you can't tell which way the jar is rolling, you've found a panel problem, not a timing problem. That's worth saying plainly, because the instinct is to lengthen the hold. More time on a panel that doesn't state its direction produces a longer ambiguous moment. The fix is the drawing: rebuild SH-03 so the jar reads in profile with the counter's end visible at frame right, and keep it filed as the same panel's next revision — SH03_jar-roll_revB against SH03_jar-roll_revA.
Replace the media on the existing timeline clip rather than deleting the clip and dragging in the new file. Then check the hold. A newly imported still carries the application's default still duration, and a fresh clip brings that default with it — so a 1.00-second hold can quietly become five seconds, which is the kind of change that ruins a comparison without looking like anything happened. Confirm SH-03 still reads 1.00 second in version A, and listen to the roll sound against the new panel, because a redrawn rolling jar may want its scrape a touch earlier or later than the old one did.
Notice what just happened: you made two changes, but only one of them belongs to the timing question. Comparing version A against a version B built on the ambiguous panel would have told you nothing about setup length, because the direction problem would have contaminated both. Fix the panel first. Compare second.
Version B then changes exactly one thing. SH-01, SH-02, and SH-03 drop to 0.50 seconds each — setup falls from 4.50 seconds to 1.50 — and SH-04 through SH-08 keep their holds unchanged. Total: 168 frames, 7.00 seconds. Same panels, same order, same sound — and here that last part needs a condition attached. Both cues stay attached to their panels rather than to absolute times, so the clack rides with SH-02 and the scrape travels with SH-03 into their new, shorter holds. That is what makes this one changed interval. After the setup is re-timed, check each cue's placement in both versions: a clack left sitting at 2.00 seconds would land under SH-04 or SH-05 in version B, and the two versions would then differ in more than the holds.
Now you have something to look at. If the compressed setup loses the rolling direction, or loses the fact that Dee's back is turned, then those three seconds were carrying real information and you should spend them or find a panel that carries it faster. If version B reads the same, the setup was longer than the scene needed, and you've found three seconds to give back — perhaps to SH-08, where the composed expression may be doing more work than 2.50 seconds allows.
Export both versions. Name the files so the revision is recoverable: something like animatic_jar_vA_SH03revB_PROVISIONAL and its B counterpart. Then watch the exported files, end to end, on something other than the editing timeline. The timeline shows you what you assembled; the file shows you what you made. Check that the export used the sequence's frame rate — a 0.75-second hold is exactly 18 frames at 24 fps and is not a whole number of frames at 25 — that the first frame isn't black, and that the roll sound sits where you put it.
What the comparison settles, and what it doesn't
You finish with a short file that says what it is: a provisional timed study, two versions of one interval, one replaced panel logged by identity, and a hold sheet that lets someone else reconstruct every duration in it. It's revisable in a way a conversation about pacing isn't. When the editor says the reveal comes too late, you can point at the interval and move it.
What it doesn't settle is everything that involves a body. Whether Nia's reach reads as a near-miss at that speed, whether a jar can be made to roll that line on a real counter without wandering, whether the catch looks easy or lucky, whether the scrape you're imagining is recordable in that kitchen — none of that is in the file. All of it is now written down as a question, which is the useful outcome. A ten-second study against a thirty-second brief isn't a failure either; the animatic is answering a question about a scene, not about the length of the buy.
The distinction to hold onto is between a movement that has been made imaginable and a movement that has been demonstrated. An animatic does the first one well, on purpose, with drawings. The second one needs a performer and a camera, and no amount of careful timing pretends otherwise.
Frequently asked questions
What should be written down before any temporary sound is added?
A one-sentence timing question a colleague could disagree with, plus marks for what the drawings cannot establish—for example how a performer's weight shifts, how fast a hand moves, or whether the camera drifts during a hold. Those marks are the study's deliberate open items, not disclaimers.
Why is importing a numbered folder as an image sequence a trap for storyboard panels?
The application makes each image one frame. At 24 fps, eight panels become about a third of a second. The useful route is importing the panels as individual still images and setting each one's duration yourself.
What changes from Version A to Version B in the jar example, and what must stay attached?
SH-01, SH-02, and SH-03 drop to 0.50 seconds each, so setup falls from 4.50 to 1.50 seconds; SH-04 through SH-08 keep their holds. Total is 168 frames, or 7.00 seconds at 24 fps. The clack and scrape must stay attached to their panels, not absolute times, or the two versions differ in more than the holds.
If the jar's rolling direction is unclear, should the hold be lengthened?
No. The article calls that a panel problem, not a timing problem; more time on an ambiguous panel produces a longer ambiguous moment. Rebuild SH-03 so the jar reads in profile with the counter's end visible at frame right, and file it as the panel's next revision.
What should be checked after exporting both versions?
Watch the exported files end to end on something other than the editing timeline. Check that the export used the sequence frame rate—a 0.75-second hold is exactly 18 frames at 24 fps and not a whole number at 25—that the first frame is not black, and that the roll sound sits where it was placed. Name files so the revision is recoverable.