top of page

The Master Guide / 14 Finishing and delivery · Graphics, captions, versions and delivery

25 fps TV versions by speed-up

A 24 fps film becomes a 25 fps TV version by playing every frame about 4% faster, as European broadcast does. Frame repeat is the fallback.

A 24 fps film becomes a 25 fps TV version by playing every frame about 4% faster, as European broadcast does. Frame repeat is the fallback.

The speed-up (decided): every frame is kept and plays at 25, one to one, with no interpolation, so camera moves stay smooth. Every cut stays on its second, so a 30-second spot becomes 750 frames, and each shot takes about one extra frame per second from its own handles (after its out point first, then before its in point). For example, a 5-second shot is 120 frames at 24 fps and 125 at 25: the 5 extra frames come from its handle after the out point, and from before the in point only when that runs out. The TV version keeps the master's length.

The sound follows three rules (decided):

  1. Each take's own sound and every lip-synced line follow the picture by a plain resample: about 4% faster and about 0.7 of a semitone higher, so sync holds. Nothing is stretched with its pitch kept.

  2. Placed one-shot effects keep their own speed and move to the TV frame of their picture event.

  3. Ambience beds, voice-over, music and any sonic logo are re-laid at their own speed on their own seconds. Because the length is unchanged, the music is not re-edited.

Measure loudness again on the TV file. Graphics are re-rendered at 25 fps.

The TV version is built from copies. The approved 24 fps master is never touched: the TV version is built from speed-up copies, in its own 25 fps sequence. Every resample uses each file's own sample rate; the wrong rate makes a line about 12% short instead of 4% (steps below).

Never convert a finished 24 fps file to 25 on export (steps below). Even the frame-repeat fallback is built as its own 25 fps sequence, so its graphics, captions and frame count are made for 25.

Frame repeat is the fallback (decided): the 24 fps media in a 25 fps sequence (Premiere's default frame sampling), every frame kept, exact length and lip sync, with a slight stutter on camera moves. Use it where a shot has no handle to give, or where you reject the sped-up voice. Optical flow is only for a channel whose sheet requires it, and it needs three or more frames of media after every out point.

The first TV version is made both ways. Watch the two on a 50 Hz TV and listen to one sped-up Arabic line, then write down your verdict. It costs nothing but your time.

Do it yourself

Make the 25 fps TV version by speed-up (ffprobe, ffmpeg, Premiere)

  1. Read each file's own sample rate (usually 48000, often 44100 from a voice engine):

ffprobe -v error -select_streams a:0 -show_entries stream=sample_rate -of csv=p=0 S03_t2.mov

Every resample below uses that number in both places. Writing 48000 for a 44.1 kHz file makes it about 12% short instead of 4% (a 10-second line comes out at 8.8 s).

  1. Each take, picture and its own sound, as one speed-up copy:

ffmpeg -i S03_t2.mov -vf "setpts=PTS*24/25" -r 25 -af "asetrate=48000*25/24,aresample=48000" -c:v prores_ks -profile:v 3 -c:a pcm_s24le S03_t2_25p.mov

Every frame is kept (a 120-frame take stays 120 frames and runs 4.8 s instead of 5), and the sound is resampled with it, so sync holds. prores_ks -profile:v 3 writes ProRes 422 HQ, and pcm_s24le writes 24-bit PCM, uncompressed audio. The original file is left as it was. In ffmpeg, -r 25 is used only after the setpts step: on its own it converts the frame rate, duplicating a frame every second.

In Premiere you can interpret the takes instead (right-click › Modify › Interpret Footage › Assume this frame rate: 25), but only in a copy of the project saved under the TV version's name (File › Save As). Interpret Footage changes every sequence that uses the clip, and the master and the frame-repeat fallback need the 24 fps originals. Mute each clip's own linked sound there and lay its resampled copy (step 3) in its place.

  1. Every lip-synced line (and any take sound kept as its own file) by the same plain resample, never a pitch-kept stretch:

ffmpeg -i line.wav -af "asetrate=48000*25/24,aresample=48000" -c:a pcm_s24le line_25p.wav

or for a 44.1 kHz file asetrate=44100*25/24,aresample=44100. A 10-second line comes out at 9.6 s, about 0.7 of a semitone higher, at either rate. Lay it under its shot at the same picture frame. For the Premiere method, give the take file itself as the input: the WAV then holds its sound only.

  1. Re-cut in a 25 fps sequence: lay each speed-up copy from the same in-point frame and extend it from its handles, so every cut lands on its second. A shot with no handle to give takes the fallback.

  2. Placed one-shot effects keep their own speed: copy each from the 24 fps sequence unchanged, then slide it to the frame where its picture event now plays in the 25 fps cut.

  3. Voice-over, music, ambience beds and any sonic logo are re-laid at their own speed: copy them from the 24 fps sequence unchanged, at the same time in seconds. They keep their seconds, and the film's length is unchanged.

  4. Graphics are re-rendered at 25 fps and laid on their shots; then measure loudness on the TV file.

Export without converting (ffmpeg)

  1. In ffmpeg, never -r 25 alone on 24 fps material.

Sources

bottom of page