Input: one course outcome and a real lesson moment
Output: a 15-20 second promo
Aspect: 9:16
Build log
1. Set the rule
I opened the a lesson plan, slides, or course outline, locked one concept and one presenter reference, and wrote down what could not change.
2. Run one minimum version
I made one small a 15-20 second promo first. The promise felt vague, so I replaced it with a concrete learner result.
3. Build the focused variant
I reused the same reference and changed only one outcome promise and one short lesson sample.
4. Review and export
I checked audio, captions, slide readability, and pacing at full size, then exported the final 16:9 and 9:16.
Complete execution record
Task boundary and source gate
Treat AI course promo video as a deliverable, not a definition. The job receives one course outcome and a real lesson moment and owes a 15-20 second promo in 9:16. Write the rejection reason first, then turn it into the first hard constraint.
Require the objective verb, terminology sheet, worked example, assessment item, and the source of any claim. If a slide contains two ideas, split the node before spending credits. Do not let a caption introduce a claim that the approved source does not contain. A pretty diagram cannot override wrong terminology.
The imgmov route
Break the outline into Canvas nodes before generating media: one observable objective, one example, one check. Slides, notation, and approved screenshots are references. Subtitles, review questions, and progress markers go to the external editor after picture lock.
Keep platform requirements outside the model prompt. The prompt can describe a scene; a named checklist records the crop, claim, consent, and console check that make it deliverable.
Lock the variables that cannot drift
Lock terminology, notation, example numbers, instructor identity, clothing, background light, diagram style, and caption format. A course should feel like one teacher with one method.
Put the locked fields at the top of the prompt, not at the end: The practical move is to lock one concept and one presenter reference and change only one outcome promise and one short lesson sample. Save that prompt with the reference, ratio, and model so the next run starts from a decision instead of a guess.
Step-by-step build path
1) Set the rule: I opened the a lesson plan, slides, or course outline, locked one concept and one presenter reference, and wrote down what could not change. Leave one checkable artifact from this step; do not start the next until it exists. 2) Run one minimum version: I made one small a 15-20 second promo first. The promise felt vague, so I replaced it with a concrete learner result. Leave one checkable artifact from this step; do not start the next until it exists. 3) Build the focused variant: I reused the same reference and changed only one outcome promise and one short lesson sample. Leave one checkable artifact from this step; do not start the next until it exists. 4) Review and export: I checked audio, captions, slide readability, and pacing at full size, then exported the final 16:9 and 9:16. Leave one checkable artifact from this step; do not start the next until it exists.
Change one named variable per rerun: prompt, reference, camera, duration, model, ratio, or export crop. If two variables change together, a better output cannot be reused because nobody knows which fix worked.
Review gates and evidence
Mute the video and read the captions. Then play it once and ask whether the learner can complete the ending check without rewinding.
Keep source, approved wording, rejected version, correction, credit cost, and final crop next to the asset. The record exists so the next reviewer can reproduce the decision without a chat thread.
Failure diagnosis and retry ladder
If learners fail the check, cut a concept instead of adding a recap. If notation drifts, return to the approved slide. If captions lag, split the sentence at a phrase boundary.
Do not upgrade on hope. A 5-second Agnes Video probe is 10 credits. If the concept passes, Seedance 2.5 costs 90 credits at 480p or 195 at 720p for 5 seconds; Kling is 195 without audio or 245 with audio. Veo enters only when a 4/6/8-second cinematic bucket is genuinely worth 80/125/165 credits. Prove hook, subject, and rhythm first, then pay to clean up motion that already worked.
Versioning and handoff
Deliver classroom, mobile, and captioned versions separately. Save the objective, reference set, prompt, and check as the next module's starting point.
The folder is boring on purpose: source, approved reference, generation settings, caption file, platform cuts, QA screenshots, and one correction line. A reusable asset is boring in the right way.
Why this is not a generic answer
The imgmov advantage is the chain: Asset Library preserves the subject reference, Canvas fixes shot order and first/last frames, the workspace routes Agnes to Seedance, Kling, or Veo only after proof, and your external editor adds captions and CTA without regenerating the media.
Do not let a caption introduce a claim that the approved source does not contain. A pretty diagram cannot override wrong terminology. If the request only asks what ai course promo video means, a search page is faster. This page is useful when someone must deliver a 15-20 second promo under real constraints.
Frequently asked questions
What is the one rule I keep repeating?
The practical move is to lock one concept and one presenter reference and change only one outcome promise and one short lesson sample.
The workflow is specific about its stopping point: it starts with one course outcome and a real lesson moment and stops at a 15-20 second promo. In imgmov, upload the source, lock it as a reference, set 9:16, run one proof, and save the passing settings. If a reviewer cannot tell which version is approved, the workflow has failed even when the media looks good.
What do I check before export?
One objective, captions, symbols, framing, and final crop.
Do not upgrade on hope. A 5-second Agnes Video probe is 10 credits. If the concept passes, Seedance 2.5 costs 90 credits at 480p or 195 at 720p for 5 seconds; Kling is 195 without audio or 245 with audio. Veo enters only when a 4/6/8-second cinematic bucket is genuinely worth 80/125/165 credits. Prove hook, subject, and rhythm first, then pay to clean up motion that already worked.
What makes this different from a generic generator?
I start from one course outcome and a real lesson moment and keep a 15-20 second promo consistent, instead of inventing a new scene each run.
The check is not “does it look AI-nice?” It is: Mute the video and read the captions. Then play it once and ask whether the learner can complete the ending check without rewinding. Then the source, approved copy, rejected version, correction, and final crop stay in the same handoff folder.
What should I do if the first render drifts?
Cut the motion, lock the reference, and rerun one small version before batching.
If learners fail the check, cut a concept instead of adding a recap. If notation drifts, return to the approved slide. If captions lag, split the sentence at a phrase boundary. Change one named variable, keep the old version, and record the credit cost. If the same failure repeats, fix the reference or scope rather than asking the prompt for forgiveness.