Delivery: Fit the frame and the words

A finely paced edit can still fail in the final export pass if the subject's hands are cropped off the edge of a vertical screen or if burned-in captions sit directly behind a platform's translucent caption description bar. Treating aspect ratios and subtitles as deliberate compositional choices—rather than automated export presets—protects your story on any screen size.

1. Choosing and reframing aspect ratios without motion sickness

Visible problem in the clip: A horizontal 16:9 interview or demonstration was converted to 9:16 using an AI auto-reframe filter, causing the vertical frame to drift left and right every time the speaker nods or gestures.

Purpose of the edit: A camera frame should feel anchored unless the subject actually walks across the room. When repurposing horizontal 4K (`3840×2160`) footage into a vertical `1080×1920` sequence, you have generous horizontal pixel headroom, so lock the X-axis position per cut rather than letting software chase micro-movements.

• Cut instead of panning: When two people converse at a table or a craftsperson moves from a vise to a saw, duplicate the clip at the moment of transition and snap the X-position to the new focal point on a hard cut (or a 6-frame ease-in/ease-out keyframe pair).
• Preserve nose room (lead room): If a subject looks 30 degrees toward screen-right, place their eyes slightly left of the vertical centreline so their gaze opens into the wider half of the 9:16 frame.
• Use Split-Screen Stacks when horizontal width is essential: If a shot compares two wide objects or shows a speaker reacting to a wide monitor, stack two 16:9 or 4:3 frames vertically inside the 9:16 canvas rather than cropping away half the visual evidence.

2. Designing captions for effortless reading speed

Visible problem in the clip: Single-word "pop-up" captions flash at 240 words per minute in neon colors, forcing the viewer to stare at the text like a strobe light instead of watching the visual demonstration above it.

Purpose of the edit: Good captions allow a viewer in a quiet train carriage—or a viewer learning in a second language—to absorb the spoken phrase in a single peripheral glance and immediately return their eyes to the action.

1. Break lines by linguistic phrase, not character limit

   Keep between 3 and 7 words on screen at a time (roughly 1.2 to 2.2 seconds per card). Never leave prepositions ("of", "to", "with") dangling at the end of Line 1 while their object starts on Line 2.

2. Ensure luminance contrast against bright footage

   White text disappears the moment a presenter wears a white shirt or steps in front of a window. Add a subtle dark semi-opaque background pill (`rgba(15, 25, 22, 0.75)` with `12px` corner radius) or a crisp dark outer stroke plus tight drop shadow.

3. Lead the spoken audio by 1 to 2 frames

   When a subtitle card snaps onto the screen one or two frames before the speaker voices the first syllable of that phrase, the viewer perceives the text as perfectly responsive rather than lagging behind the mouth movement.

3. The Pre-Export Technical Checklist

Before queuing your render, run a 15-second technical verification pass across your sequence settings:

• Color Space Tagging: Verify your timeline working color space and output color space are both set to Rec.709 (Scene) or Rec.709-A (if cutting on macOS QuickTime pipelines) so skin tones do not wash out on upload.
• Frame Rate Consistency: Confirm your export matches the native cadence of your primary camera footage (`23.976`, `25`, `29.97`, or `30` fps). Avoid mixing 24 fps and 30 fps clips in the same sequence without optical retiming, or motion will stutter every fifth frame.
• Sample Rate: Export stereo AAC audio at `48,000 Hz` (`48 kHz`) and `320 kbps` to prevent high-frequency sibilance distortion during platform re-encoding.