The person brings a slide deck (PDF), the brand (fonts, colours, logos), a presenter (a character sheet or an approved still) and voice guidance. You deliver a 1080p video of the presenter speaking the narration beside slides whose headlines, lists and graphics land on the words, in the layout agreed with the person (core skill).
The presenter's footage is made once, as one continuous lip-synced take of the whole track, and composited into every slide layout (a full-frame shot, a window, a circle bubble). Never render a presenter per layout.
Steps
-
Write the script (
guide-make-a-script-from-source,llm-convert-written-script): the narration asscript.json, one section per slide. Ask the person first whether they already have the script (speaker notes, a voiceover draft, a brief with the words). There is no single answer, so say plainly which case you are in:- They have it: convert it as written; the gate below is a quick confirmation.
- Only the deck: you write the narration per slide, so the person must read it
through before anything is voiced. Put every line to them with
review_approve(the script as a.mdfile) and ask whether they want to read and approve it; tell them it will be voiced exactly as approved.
Gate: the person approves the script word for word; every later word change costs a re-voice and a re-render of that stretch.
Straight after, before casting: the first slide's layout in two or three looks (where the slide sits, the presenter as window, cutout or bubble, how its points build), as the first look (
hn4a-compositor, The first look), with a grey silhouette for the presenter. Gate: the look. It costs nothing and is the first thing the person sees move; the presenter, the lip sync and the other slides all come after. -
Cast and perform the voice (
guide-cast-a-voice,guide-voice-a-script,elevenlabs-tts-v4): pcm_48000, loudness normalized on the finished track, and any word the voice misreads spelled out ("H-R", not "HR" or "H.R."). Transcribe the final track (elevenlabs-speech-to-text) for word and pause times (python-read-word-timings). No gate on the track: the person heard the voice when they picked it, so a track that passes the transcript check goes straight on. -
The presenter and the whole-track take (
guide-present-with-an-avatar, Process): a designed presenter or an avatar of a real person, the 3:4 still locked with any badge removed, then the track lip-synced as one take in chunks (python-plan-lipsync-chunks,h3-max-lipsync-a-still,python-assemble-lipsync-chunks), the first look seen moving with the still in it before the first render. Its gates stand: the still, then the assembled talking head with its lip-sync plot. -
Build the slides (
guide-add-titles-and-motion-graphics,hn4a-compositor-motion-graphics) in the look picked in step 1: one JSON layer spec per layout, laid out against a still of the source slide, the track as the audio. Cue every entrance about 0.1 s before its word from the word times. The presenter as a full-frame window, a circle bubble cropped from the portrait clip, or a cutout over the slide on a real matte (guide-present-with-an-avatar, Layouts). Gate: the live spec (hn4a-compositor, Live spec, then render), the specs asitems; then the render and its checks. -
Deliver (core skill):
deliverthe video.
Revisions
The chunk is the unit of change (guide-present-with-an-avatar, Revisions). After a
re-join, point the slides at the new track, shift every cue after the changed chunk by
its length change, and re-render only the spans that changed.
Choices this workflow makes
- Slides built as motion graphics around the presenter, their points landing on the
words; for the deck's real slides shown as they are,
workflow-pitch-deck-to-video. - Portrait 3:4 at 768P (
guide-present-with-an-avatar, Choosing). - H3 Max is the route end to end. There is no separate draft pass: the still is approved before any render, and every chunk is pinned to it.
Dead ends
- A presenter rendered per layout: the take is made once and composited into every
layout. The presenter's own dead ends (no words in the prompt, finger counting, a
colour-keyed cutout, a full-frame edit to remove a badge) are in
guide-present-with-an-avatar, and the join's (-ttrims that drifted 313 ms over 17 chunks, hard cuts that popped at the seams) inpython-assemble-lipsync-chunks.
Worked example
The Globex title-slides job: Presenter Dana, a 163 s track, 17 chunks.
- All 17 first frames matched the still's layout at 0.999
(
python-measure-layout-drift); the framing held at ≥ 0.984 with ≤ 1.5% zoom. Three chunks drifted 3.3–5.5% slow (08, 10, 13), in sync at the start and 200–500 ms late by the end; the person had spotted two of them, and the check caught the third. The check took about 111 s for the whole track. A VEED re-sync brought chunk 08 back to −4 to −127 ms, inside the band. - Slide
slide_b: 55.8 s at 30 fps (1674 frames). A title over the presenter, then the heading "What to watch for." with three illustrated points building at 22.9, 32.7 and 43.0 s while the presenter shrinks to a corner bubble. The title fades out at 20.4 s as the bubble fades in from 19.9 s: the 0.5 s overlap reads as one move. Rendered in 48 s with 8 workers. A 10 s revision re-renders in about 15 s plus a 0.4 s splice, against about 2.5 min for the whole thing. - Composited into the panel and window layouts, the 3:4 768P footage was approved as is.


