Skip to content

Read Video Timecode Before Giving Notes on a Commercial

Advertising

Read Video Timecode Before Giving Notes on a Commercial

A note that reads "at 0:11 the logo is too small" sends three people to three different frames. The editor scrubs her timeline to 01:00:11, counting from a sequence that starts at 01:00:00:00. The producer opens the review link and lands at 0:11 — elapsed time in a file that begins on a slate the sequence doesn't have. The person who cut the source clip pulls up the raw footage and finds 10:00:03:12, a number that looks like it belongs to another project. Everyone is looking at a real value from a real display. Nobody is looking at the same moment.

That is not a discipline problem. It is four clocks in one room, and the note never said which one it was counting on. They agree only when every zero lines up: a sequence starting at 01:00:00:00, an export whose first frame is the sequence's first frame, a clip whose head sits there too. Shift any one of those and the same instant picks up a different number on each clock.

Four clocks, one frame

Timecode is a label that names a position. It does not name a duration, a recording date, or a moment in the outside world. It names a place inside something, and until you know which something, the number is decoration.

The usual arrangement on a commercial job has at least four:

The clip's own timecode. This label lives in the file, normally stamped by the camera when it recorded, and sometimes rewritten by whatever produced the copy you are watching. It may start anywhere — cameras commonly run continuously for a whole shoot day, so a clip can begin at 10:43:17:06 and just keep going. Call this the source clock.

The sequence clock. In an editing application the timeline has its own start, often 01:00:00:00 by convention, sometimes 00:00:00:00, sometimes whatever the assistant set. Placing a clip into that timeline gives every frame of it a second address.

The player's elapsed time. A review link, a preview window, a scrubbing bar: these usually report seconds since the start of the file you are watching. It is the friendliest number and the least stable, because it depends entirely on where that particular file begins.

The export's own burn-in. When a file is exported with a timecode window in the picture, that window starts wherever the export settings say — which may be the sequence's start, or zero, or something the person who exported it chose.

Two of those are easy to confuse inside a single application. Adobe's documentation for Source and Program Monitors describes separate time controls for each, with a current-position field and a duration field sitting near each other. The same box that reads 00:00:11:12 as a position can read 00:00:30:00 as the length of the whole thing. Before reading any number aloud, check that the field you are quoting is the position field and the monitor is the one you meant. Menu names and label positions move between versions, so confirm the labels on your own screen rather than trusting a guide's screenshot.

The last pair is frames, not a fraction

Timecode follows hours, minutes, seconds, frames: HH:MM:SS:FF. The final pair counts frames inside the current second and runs from 00 up to one less than the frame rate. At 25 frames per second the field goes 00 through 24, then rolls the seconds.

Which means 00:00:03:12 is three seconds and twelve twenty-fifths of a second — 3.48 seconds. It is not 3.12 seconds, and treating the last two digits as decimals is the most common way a correct-looking note points at the wrong frame.

Some displays skip the arithmetic and show a frame count instead, which is genuinely useful: 3.48 seconds at 25 fps is 87 frames from the start, so a frame-count display reads 87 where the timecode reads 00:00:03:12. Adobe's timecode display settings include a frame-count choice alongside media source timecode and a start-at-zero option. The caution is the indexing base. A display that calls the first frame 0 and one that calls it 1 differ by one, which is invisible until someone counts.

Ranges carry their own trap. If the in point is 00:00:03:12 and the out point is 00:00:04:00, the range is thirteen frames long. If you write "cut 3:12 to 4:00," the recipient cannot tell whether the second number is the last frame you want gone or the first frame you want kept. Those differ by one frame, which at 25 fps is forty milliseconds, which is the difference between a clean cut and a visible one. Say which convention you mean.

Check the rate and the counting convention

Before you treat a label as elapsed time, establish two things: the frame rate, and whether the counting convention is drop-frame.

Whole numbers are easier. At 25 fps, every timecode second contains exactly 25 frames, and the labels track real elapsed time closely enough for review notes.

Some rates are not whole numbers. 29.97 and 23.976 are common in material that originated in other systems, and at those rates the relationship between the label and the wall clock changes. At 29.97 with a non-drop count, the numbers drift against real time by roughly 3.6 seconds over an hour. Drop-frame numbering exists to correct that drift: it skips frame numbers — typically 00 and 01 at the top of most minutes — 108 of them across an hour. Apple's archived Technical Note TN2310 makes the narrow point worth keeping: drop-frame notation skips timecode values, not recorded pictures. No frame of your commercial disappears because the count skipped a label.

Two consequences for notes. First, name the rate. "03:12" at 25 fps and "03:12" at 29.97 are not the same instant, and if the two of you are on different rates, every number in the thread is off by a percentage nobody can see in the first ten seconds. Second, establish whether the count is drop-frame or non-drop, because a job can legitimately use either, and the two disagree by a few seconds an hour. At 25 fps there is no drop-frame option to worry about — the convention simply doesn't apply, and anyone offering you one is confused.

Worth saying once: timecode does not tell you when something was recorded, and it does not measure how long the wall clock ran. A camera can be set to any starting value. A transcoder can rewrite one. The label is a reliable address and an unreliable diary.

Mapping source to sequence, when the relationship is known

Now the useful part. Take a real relationship and see what falls out of it.

Here is an illustration worked on paper, not a test of any particular application. An original clip shot at 25 fps carries source timecode beginning at 10:00:00:00. The clip is placed into a sequence, unchanged, with its first frame sitting at sequence timecode 01:00:08:00. The sequence itself starts at 01:00:00:00.

An event at source 10:00:03:12 is 3 seconds and 12 frames into the clip — 3.48 seconds past the clip's first frame. The clip's first frame sits 8 seconds into the sequence. So the event sits 8 + 3.48 = 11.48 seconds into the sequence, which is 01:00:11:12.

The same frame, read four ways:

Where you are reading What it shows
Source monitor, clip's own clock 10:00:03:12
Sequence / Program monitor 01:00:11:12
Export with a zero start, burn-in 00:00:11:12
Player elapsed time about 11.48 s, commonly displayed as 0:11

The arithmetic is a subtraction and an addition: take the event's source label, subtract the clip's source start, add the result to the sequence position where the clip begins. Notice what else it depends on. The sequence start being 01:00:00:00 is what lets 11.48 seconds and 01:00:11:12 describe the same frame. Change the sequence start and the sequence column changes while the source column doesn't.

And notice how easily the whole relationship breaks:

  • The clip is retimed, so a source second no longer equals a sequence second.
  • A different version of the clip was used, or a cutaway was inserted before it, moving everything downstream.
  • The clip was trimmed at its head, so its in point inside the sequence is no longer the clip's first frame.
  • The export was trimmed, or exported with a start timecode other than the sequence's, so the player's zero is somewhere new.

When you can see the clip's source start, the sequence placement, and the export settings, and nothing is retimed, the offset is sound. When you can't, do not perform the conversion anyway. Ask which version is under review, or name the event and let the editor find the frame:

"The wide shot where the delivery van backs out of frame — the third cut after the kitchen. The logo on the van's rear door is soft."

That sentence is recoverable by anyone holding the right file, and it costs nothing. A confident number attached to the wrong export costs a round trip.

Writing the note so someone else can recover it

A review note that survives contact with four clocks has four parts: the version, the clock, the position or range, and the visible event. Then the change, kept separate.

Version: pitch_v4_review.mp4 (the export with burn-in starting at zero) Clock: sequence timecode Where: 01:00:11:12 — the frame where the van's rear door logo fills the left third Change: hold two frames longer before the whip pan

The version has to be something the recipient can actually identify — a filename, or a link with a date attached, or "the cut you sent Tuesday." "The latest" is not a version. Numbers and versions belong in the same breath, not split between the note and a technical appendix nobody opens, because the number is meaningless without the version and takes the reader three seconds to lose.

Where an issue appears and what should change are separate statements. "The logo is too small at 01:00:11:12" locates a problem. "Push in two frames at 01:00:11:12" proposes a fix. Run them together and the editor can't tell whether the push-in at that point was meant to solve the small logo or something else entirely.

Then let the recipient confirm. If they open the file and find a different picture at your number, the useful response is not to argue the arithmetic — it's to check the version and the clock. If those turn out to be unavailable, an honest approximate player position plus a recognizable event beats a precise-looking mistake:

"About eleven seconds in on the review link — the van shot. The logo's soft. I don't know which cut I'm on, so tell me if you need the exact frame."

That note is weaker than a correct one and much stronger than a wrong one.

Here's the short version to carry into a review: open the file, confirm which clock the number comes from, confirm the version, read the last field as frames, name the event. Then write the note so that anyone with the same file arrives at the same picture. The number is not the point. Agreement about what the number counts is.

Frequently asked questions

Why can a note like 'at 0:11' point to different frames?

Timecode is a label that names a position inside something, and a commercial job usually has at least four clocks: the clip's own source timecode, the sequence clock, the player's elapsed time and an export's burn-in. They agree only when the zeros line up; shift the sequence start, export start or clip placement and the same instant gets a different number.

How should the last pair of digits in timecode be read?

As frames, not a decimal fraction. At 25 fps, 00:00:03:12 means 3 seconds plus 12 frames, or 3.48 seconds, not 3.12 seconds. A frame-count display at 25 fps would show 87 frames from the start for that same point, though a display that starts at 0 versus 1 differs by one.

What should a review note include to be recoverable?

Name an identifiable version, the clock the number comes from, the position or range, and the visible event, then state the change separately. If the version or clock is unavailable, an honest approximate player position plus a recognizable event is stronger than a precise-looking number attached to the wrong file. 'The latest' is not a version.

How do frame rate and drop-frame affect a note?

Name the frame rate, because 03:12 at 25 fps and at 29.97 are not the same instant. At 29.97 with a non-drop count, labels drift against real time by roughly 3.6 seconds an hour; drop-frame numbering skips timecode values, not recorded pictures. At 25 fps there is no drop-frame option, and timecode does not tell you when something was recorded.

When can source timecode be converted to sequence timecode?

Only when the relationship is known and nothing is retimed: subtract the clip's source start from the event's source label, then add the sequence position where the clip begins. For example, source 10:00:03:12 with a clip start of 10:00:00:00 placed at sequence 01:00:08:00 on a sequence starting at 01:00:00:00 becomes 01:00:11:12. If the clip was retimed, trimmed, replaced by another version or exported with a different start, do not perform the conversion; ask which version is under review or name the event.

More in Advertising Browse all articles