Fourteen Frames That Cost Two Days
A director reaches picture lock on a 78-minute feature. The composer has been working from a rough cut for six weeks, building themes and writing cues. The director sends the locked picture - and then remembers that the editor trimmed 14 frames from a scene transition on reel three to fix a continuity problem after the "locked" file was exported. The composer's cue for that scene now lands 14 frames late. The climactic music hit timed to coincide with a door slamming shut now plays while the door is still open.
Fourteen frames is less than half a second. It takes two days to fix because the composer has to re-export six cues from a session built against the wrong cut.
The music-to-picture workflow has specific failure points that are entirely avoidable with the right process. This post covers that process from the spotting session through stem delivery - including how timecode, tempo mapping, and the Tempo to Frame Sync Calculator prevent the frame-offset problems that consume time on nearly every first-time composer collaboration.
Workflow standards in this post draw from the Berklee Online Film Scoring curriculum, Professional Post Sound workshop documentation, and published composer workflows documented in Film Score Monthly and similar industry publications.
How Music and Picture Stay in Sync
The synchronization between music and picture rests on a shared temporal reference: timecode. The film's edit has a timecode start point - typically 01:00:00:00 for the program start. Every frame of the locked picture has a unique timecode address. When the composer builds a DAW session, they import the locked picture and align the video to the same timecode start. Every cue they write anchors to a specific timecode position. As long as the picture doesn't change, the music hits its marks on every playback.
Why picture lock is non-negotiable: Once a composer begins working against a cut, any change - even a single frame trim - breaks the timecode alignment of every cue that follows that change. A one-frame trim in reel two means every cue from reel two onward has shifted and must be re-checked. The composer doesn't need to re-write anything, but they need to re-export everything downstream of the change, after verifying the new hit points still land correctly.
Tempo mapping is how composers sync musical rhythm to specific picture events. A cue at 120 BPM in strict 4/4 may not naturally land a beat on the exact frame where a character fires a gun or a cut to close-up occurs. Tempo maps let the composer vary BPM or insert meter changes so musical accents align with picture events without sounding rushed or dragged. The Tempo to Frame Sync Calculator converts between BPM and frames-per-beat at any frame rate, letting you verify whether a given tempo places accents on clean frame boundaries before a note is written.
The formula for frames per beat:
Frames Per Beat = (Frame Rate x 60) / BPMAt 24fps and 120 BPM: (24 x 60) / 120 = 12 frames per beat. If the picture event falls on frame 14 of a bar, 120 BPM won't place a beat there. At 102.9 BPM: (24 x 60) / 102.9 = 14 frames per beat exactly. The 17.1 BPM adjustment is audible if made abruptly, but achievable through a gradual tempo ramp across 8 bars.
Typical Film Scoring Timeline
| Phase | Timing | What Happens |
|---|---|---|
| Temp mix (rough cut) | During editing | Editor places temp music as reference for director and producer feedback |
| Spotting session | Within 1 week of picture lock | Director, composer, editor walk the film identifying cue positions and intent |
| Score writing | 4 to 12 weeks post-lock | Composer writes and records cues; delivers stems for approval |
| Stem approval | Rolling through score period | Director approves or requests revisions per cue |
| Score recording | 1 to 4 weeks before mix | Live players or virtual instruments finalized and recorded |
| Music deliverables | 1 week before final mix | Composer delivers stems aligned to locked picture timecode |
| Final mix | 1 to 2 weeks before delivery | Re-recording mixer combines music stems with dialogue and effects |
This assumes a low-to-mid-budget production with a dedicated composer. Micro-budget projects using library music compress this timeline significantly but lose the custom tempo mapping that makes a score feel built for the film.
Three Real-World Scoring Collaborations
Example 1: Short Drama, Composer Plus Library Supplement
A 22-minute drama. The director hired an emerging composer for $1,500 to write original themes for four key scenes, supplementing with Musicbed library tracks for transitional cues. The spotting document listed 11 cue positions with timecode in/out, emotional intent, and a temp reference track for each.
Problem: Two library cues were edited to picture at the wrong frame rate. The library tracks were 25fps; the film was 24fps. In the Resolve timeline at 24fps, the library cues drifted by 4% over their duration - barely noticeable at the start, but a full half-second out by the end of a 2-minute cue.
Fix: The editor re-imported the tracks as 25fps audio, letting the NLE handle the sample rate conversion. Four minutes to fix - but only after correctly diagnosing the frame rate mismatch as the cause.
Example 2: Feature Documentary, Remote Composer
A 74-minute documentary, LA-based director and Berlin-based composer. Score budget: $18,000. Eight weeks from picture lock to mix. The director delivered an H.264 QuickTime with embedded SMPTE timecode at 24fps. Spotting notes were a shared Google Doc: timecode in, timecode out, cue name, duration in seconds, director notes per cue.
The Tempo to Frame Sync Calculator confirmed the main theme at 84 BPM would place a beat every 17.1 frames at 24fps - not a clean frame boundary. The composer adjusted to 84.7 BPM, which places a beat every 17.0 frames, allowing the theme to sync cleanly to whole-frame hit points throughout the film. All 23 cues were delivered as 48kHz / 24-bit WAV stems with timecode start points in the file names. The re-recording mixer aligned every stem without a single manual offset adjustment.
Example 3: Music Video, Director Self-Scoring
A 4-minute music video. The director wanted ambient underscore under two spoken-word sections not covered by the original track. No external composer - they used Logic Pro X to write the layers themselves, without knowledge of Logic's tempo map tools.
Using the Tempo to Frame Sync Calculator, the director found that their ambient loops at 72 BPM would place a phrase boundary every 2 seconds at 24fps (72 BPM divides into 60 seconds as 0.833 beats per second; at 24fps, one beat = 20 frames = 0.833 seconds). Phrase boundaries landed on the visual moments the director wanted without requiring tempo map automation.
Step-by-Step: Spotting Session Through Delivery
Step 1: Schedule the spotting session within three days of picture lock. Attend with director, composer, and ideally the editor and supervising sound editor. Walk the film in real time. The composer notes: where music starts and ends (timecodes in and out), the emotional intent of each cue, any instruments the director has in mind, and specific picture events that need a musical hit.
Step 2: Deliver the locked picture with timecode reference in writing. Export ProRes 422 or high-bitrate H.264 with SMPTE timecode embedded (not burned in as a visible overlay). Include a spotting document with timecode in/out, duration, and intent per cue. Confirm the frame rate in writing: 23.976fps, 24fps, 25fps, and 29.97fps are all called "24fps" colloquially but produce different tempo calculations and must be specified exactly.
Step 3: Establish a revision protocol before the first cue is delivered. Agree on a maximum number of revisions per cue, what constitutes a revision versus a new cue, and the turnaround time for feedback. Write it down, even between friends. It protects the relationship when revision requests escalate.
Step 4: Review each delivered cue against the locked picture. Import the audio file into your NLE at the timecode start point the composer specified. Play it against picture in real time. If a hit point is more than 2 frames off its target, flag it with timecode-referenced notes before accepting the cue.
Step 5: Verify stem delivery format before the mix. Standard: one WAV file per cue (or per stem group), 48kHz / 24-bit, starting at the cue's timecode in point, no fades applied. File names should include the cue number and timecode start: "M32_Strings_01-24-38-14.wav" tells the mixer everything they need without consulting the cue sheet.
Step 6: Deliver the cue sheet to the mix stage. The cue sheet lists every music cue: cue number, cue name, timecode in/out, duration, composer, publisher, and original score vs. licensed music. It is required for ASCAP/BMI/PRS performing rights registration, festival music clearance documentation, and streaming platform delivery.
Pro Tips and Common Mistakes
Pro Tip: Send the composer a "work picture" with burned-in timecode before picture lock so they can sketch themes and confirm session setup. Note clearly that this is not the locked version and timecode positions may change. A composer who enters the spotting session with themes already drafted can focus the conversation on placement and intent rather than starting from scratch.
Pro Tip: When evaluating temp music, note the BPM of any tracks the composer needs to match in feel. Use the Tempo to Frame Sync Calculator to check whether that BPM syncs cleanly to your key picture events. If a temp track happens to sync well to a specific visual moment, it may be because the editor unconsciously trimmed the cut to align with the music. If so, the original score needs to match that BPM to preserve the same feel.
Common Mistake: Sending a compressed low-bitrate H.264 as the picture reference. If the composer's decode stutters or drops frames, they cannot accurately identify hit points. Send ProRes 422 or an H.264 at a minimum of 20 Mbps for 1080p. The extra upload size is worth the confidence that the composer's picture reference matches yours frame-for-frame.
Common Mistake: Sending a new cut to request cue revisions. If the composer receives a new video with editorial changes, they have to re-evaluate every hit point from scratch. Instead, give timecode-referenced feedback: "01:24:38:14 - chord hit needs to move 6 frames earlier to catch the cut to close-up." That is actionable in minutes. A new cut requires a full re-sync session.
Frequently Asked Questions
What is the difference between a stem and a cue?
A cue is the complete music track for a specific section of the film, starting at a defined timecode and ending at a defined timecode. A stem is one isolated instrument group within that cue. A single cue might have four stems: strings, brass, rhythm, and solo instrument. Stems give the re-recording mixer the ability to balance the score against dialogue and effects without going back to the composer for a remix. Not every budget allows for stems - a single stereo cue file is acceptable for lower-budget projects.
How far before the final mix should picture lock happen?
For a feature with an original score, picture lock should happen at least 10 to 12 weeks before the final mix start date. For a short film with under 15 minutes of music, 6 to 8 weeks is workable. Documentary films using library music can compress to 4 to 6 weeks if licensing moves efficiently. The most common mistake is locking picture too late, then either rushing the composer or starting the mix with an incomplete score.
What does "picture lock" actually mean on a low-budget film?
Picture lock is the declaration by the director and editor that no further editorial changes will be made. Every frame is frozen. On low-budget productions, "picture lock" often means "mostly done, might trim a frame here and there" - which is operationally different from true lock and must be communicated clearly to the composer. True lock means the editor has exported a master file, everyone with approval authority has signed off, and no changes will be made short of a broadcast standards compliance correction.
Can I send commercial temp tracks to the composer as reference without licensing them?
Yes, in a private creative reference context. Sending a commercial track to a composer as a reference for mood, tempo, or instrumentation is internal creative communication. What is not permitted is including that commercial track in any publicly screened or distributed cut without a sync license. The legal exposure begins when the temp music appears in a festival submission, a distributor screening, or any public-facing version of the film.
Related Tools
The Tempo to Frame Sync Calculator converts between BPM and frames-per-beat at any standard frame rate, confirming whether a given tempo places accents on clean frame boundaries. For the loudness standards the final mix must meet, LUFS and Loudness Normalization explains the integrated LUFS targets your re-recording mixer needs to hit for each delivery platform. The Timecode Calculator handles conversions between timecode formats when working with a composer whose DAW uses a different display standard.
For the SMPTE timecode standard that governs how timecode is embedded and read across DAWs and NLEs, the Society of Motion Picture and Television Engineers publishes the reference documentation for SMPTE 12M and related standards.
A Precision Handoff Between Two Creative Departments
The music-to-picture workflow is procedural, not creative. The failure points - frame rate mismatches, post-lock trims, imprecise hit point feedback, wrong delivery formats - all have the same cause: unclear communication about what "locked" means and what format each side of the handoff is working in. Clear a spotting document, confirm the frame rate in writing, and build in a revision protocol before the first cue is delivered.
This post covers the standard workflow for a single composer and single score. Multi-composer projects, music supervisors working with a library catalog, and Atmos music delivery involve additional coordination layers beyond this framework.
If you've worked with a composer on a scored project, what was the communication method that worked best for hit point feedback - timecode notes, annotated screenshots, or something else?
