Course Black Friday offer clip: lesson excerpt under a price card

To make a course Black Friday offer video, cut 20 seconds of your best lesson with video trim ($0.02), stack a price card above it with timeline compose ($0.02), then burn the lesson's own speech as captions ($0.20), which transcribes automatically when you send no cues. Total at listed rates: $0.24. The buyer sees the real teaching and the offer in the same frame, and you add no extra voice-over.

Why show a lesson instead of an ad?

A price card says what the offer is. A taste of the lesson shows what the course is. A half-banner layout, still on top and video below, is exactly what the compose stack operation produces by default (the docs call it half-banner), with the still taking a share between 0.1 and 0.9 of the frame. For a 16:9 lesson in a 9:16 frame, give the card 0.3 and let the video fill the rest with the cover fit.

What are the three steps?

First, import the lesson recording, then trim a start and a duration. Exact precision, the default, is a frame-accurate re-encode that keeps the audio. Second, send the trimmed clip and your price-card still to compose with operation stack, a horizontal split and the image at the top. Third, send the composed MP4 to video captions with no cues. The captions run speech-to-text and align to the clip's speech, which suits a talking instructor.

What should the price card say?

The card is a still you make anywhere, because compose does not draw text. Put the offer, the ends-on date and the audience on it, and keep the numbers identical to your sales page. Do not rely on the spoken line for the price: captions show what was said, and a discount may change before the video does. Pick the date in your own terms rather than a platform calendar.

What if the lesson is longer or the teacher is quiet?

Trim handles sources up to 1800 seconds, and anything longer should be split before import. If the lesson has long silences or background music over the speech, captions may mistime. In that case send script_text so the burned text follows your script while Sume keeps the speech timings, or send cues to author the text yourself. The captions docs list script_alignment_mismatch and script_alignment_failed as the typed errors for a script that cannot be aligned.