Filmed footage and interviews

Read this only when the film has real filmed footage. Set the film master's frame rate first, clean the location sound, and cut interviews on paper before any assembly.

The film master's frame rate, set in your shot plan before any edit [best available, not yet validated]:

• 24 fps by default, Seedance's own rate.

• 25 fps when filmed footage with sync sound (interviews, filmed dialogue) carries the film, or when the main delivery is 25 fps TV. The generated 24 fps shots then sit in the 25 fps sequence: beside filmed sync footage by frame repeat, and in a TV-led film master as speed-up copies, every frame kept and about 4% faster, with frame repeat as the fallback (decided; see Graphics, captions, versions and delivery → 25 fps TV versions by speed-up).

• Filmed 25p footage in a 24 fps film master is used only where it carries no sync sound, and you choose its conform in Premiere. Speech is never slowed or sped.

Clean the location sound before the cut [proven] (standard post-production practice):

• Keep the location sound as received, with a cleaned copy per clip.

• Choose one mic per speaker per scene (the lav for clarity, the boom for room) and keep to it across the scene's cuts.

• Measure first: the noise floor in the speech-free parts, clipping, true peak, and each speaker's loudness.

• Clean gently, in this order:

  1. a high-pass at about 80 Hz;

  2. de-click, where clicks are measured;

  3. light broadband noise reduction (the starting value is below; measure the noise floor before and after);

  4. one gain per speaker per scene, to about −20 LUFS (loudness units relative to full scale, the unit loudness is measured in).

  Heavy restoration (wind, hum, reverb) is a sound engineer's work.

• Lay the location's own room tone under every cut in the scene, on the AMBIENCE track.

• The cleaned clip plays on DIALOGUE. Keep the raw clip in the timeline beneath it, muted, on a DIALOGUE ALT track from DIALOGUE ALT 2 down, so the cleaning can be undone. DIALOGUE ALT 1 stays for the dry takes of placed lines (see Editing → The editable timeline).

Cut interviews from a paper edit. Transcribe with speakers and word timestamps (ElevenLabs Scribe). Mark the strongest answers against the treatment's story, in story order, each with its in and out points from the word times, and settle the paper edit before any assembly. Cuts fall in natural pauses (at least 0.3 s long), never inside a word; a cut inside an answer is covered by a cutaway or a generated shot. Quotes stay letter for letter in the speaker's dialect; nothing is reworded.

Do it yourself

Reduce location noise lightly (ffmpeg)

1. Use ffmpeg's afftdn=nr=6:nf=-50 as the starting value.

2. Measure the noise floor in the speech-free parts before and after.

Sources

• From standard documentary post-production practice.