Two-Camera Documentary with Separate Audio
A class is making a four-minute documentary on a local weaving cooperative for the school's annual day. Two phones are used: Phone A records 1920 × 1080 at 25 fps; Phone B records 1280 × 720 at 50 fps. Sound is recorded separately on a borrowed recorder at 44.1 kHz, and the clap at the start is visible in both video files. The editor's laptop has 4 GB RAM, a 1 TB hard disk and no dedicated GPU. After placing all clips on the timeline, the student notices that the external audio drifts ahead of the picture by about half a second by the end of the film, that Phone B's motion looks smooth but the file is large, and that a test export to 1080p is 1.2 GB for four minutes. The school needs a 480p file under 40 MB for WhatsApp and a 1080p file for YouTube. Internet upload speed is 2 Mbps. The annual day is in five days.
1. What sequence settings and frame rate should be chosen, and how should each source clip be conformed? Justify the choices.
Read the worked answer
Choose a 1920 × 1080, 25 fps, progressive timeline. 25 fps matches Phone A, the higher-resolution and likely primary camera, and 1080p25 is a standard YouTube upload. Phone B's 720p50 footage cannot simply be dropped in: conform it by interpreting the 50 fps clips at 25 fps, which gives 2× slow motion, or by frame-sampling to 25 fps for normal speed. Mixing 25 and 50 fps clips without conforming causes duplicate or dropped frames and jerky motion. Scale 720p up to 1080p, a 2.25× pixel increase; it will look softer, so use Phone A for important shots. On a 4 GB RAM laptop, create 540p proxy files for editing and relink to the originals for export; do not scrub 1080p50 H.264 directly. Set a consistent audio sample rate, preferably 48 kHz, and keep the recorder's 44.1 kHz files until export.
2. Diagnose the half-second audio drift and describe a reliable sync procedure for this edit.
Read the worked answer
Half a second of drift by the end is progressive, so nudging the whole audio clip will not fix it. The usual causes are clock drift between the phone and the separate recorder, and variable frame rate video from the phone being interpreted as constant frame rate. First convert the phone clips to constant frame rate with HandBrake or ffmpeg before editing. Sync the clap at the start, then check the waveform at the middle and near the end. If the audio is progressively ahead or behind, split and re-sync at several points, or time-stretch the audio by the measured percentage, for example 0.1–0.3%, so its duration matches the video. Better still, use a waveform alignment tool on the clap and on another visible hand clap near the end. For future shoots, record scratch audio on the camera and use a clapperboard; a stable recorder clock reduces drift.
3. Recommend export settings for the two deliverables, including resolution, bitrate, codec and audio, and estimate whether the 2 Mbps upload is workable.
Read the worked answer
For WhatsApp 480p: MP4/H.264, 854 × 480, 25 fps, 1.0–1.2 Mbps video, AAC-LC 128 kbps stereo. At 240 s, 1.1 Mbps video is about 264 Mb, or 33 MB, plus about 3.8 MB audio, so around 37 MB, under 40 MB. Use two-pass encoding or CRF 23 and check the final size. For YouTube 1080p: MP4/H.264, 1920 × 1080, 25 fps, 8–10 Mbps video, AAC 192 kbps. At 8 Mbps the file is about 240 MB; at 10 Mbps about 300 MB. Uploading 300 MB at a theoretical 2 Mbps takes 2400 Mb ÷ 2 = 1200 s, or 20 minutes; real overhead makes it 25–35 minutes. That is workable if the connection is stable, so upload overnight and keep the computer awake. Use fast-start for web playback and do not upscale the final above 1080p.