Caption animation should clarify timing, not compete with reading. The ordinary caption layer can be animated through supported transform and appearance properties, while Typewriter Reveal exposes a text-aware progress track.
Layer-level motion
Use position, scale, rotation, dimensions, or opacity keyframes for the whole caption box. A short opacity and vertical-position entrance can establish the caption region without animating every word.
Typewriter Reveal
Add Typewriter Reveal from Layer Effects. Its Progress parameter ranges from 0–100% and determines how much eligible caption text is revealed; Softness controls the reveal edge. Add keyframes to Progress for explicit reveal timing.
Visual fades
Caption layers can display visual fade handles, but the current handle-finalization path does not reliably accept the caption type. Do not depend on those handles for a deliverable. Use opacity keyframes for a caption-layer entrance or exit and review the rendered frames.
Timing priorities
- Token timing matches speech.
- Pages remain on screen long enough to read.
- Layer motion settles before attention must move to the words.
- Effect animation does not hide a spoken token.
Avoid combining a slow typewriter progress track with word timings that already reveal text sequentially unless you have reviewed every page. Two reveal systems can multiply into missing or late words.
Test protocol
Play the full caption section at normal speed, once while reading and once while watching the picture. Then frame-step the first page, a middle page boundary, and the final page. Check the H.264 output if browser text rendering or effect performance differs under load.
See keyframe overview and easing. 
