Skip to main content

How to make an explainer video with AI: script, shots, captions and music in one afternoon

The PrismPoster teamNovember 3, 20265 min read

An explainer video answers one question in about a minute: what this thing is, why the old way fails, how the new way works, what happens next. Its shots are simple by design (a problem, a process, a result), which is why generated stills and short clips suit it better than they suit a drama. This guide makes a 60-second explainer from a script: an eight-beat structure, a still per beat, three animated shots, captions that carry the narration, a score generated to the runtime, and the assembly on a timeline, with the credits it costs.

What is the structure of an explainer video?

Eight beats in a minute: a hook, the old way, the idea, three steps of how it works, the result, and a call to action; the first and seventh beats mirror each other so the change is visible without a word.

Beat Seconds What is on screen What the caption says
1 Hook 6 The situation, wide The question the video answers
2 The old way 8 The frustrating process, medium shot Why it fails
3 The idea 6 The thing that changes, close-up One sentence
4 Step one 8 Insert shot: hands, screen, object The first step
5 Step two 8 Insert shot The second step
6 Step three 8 Insert shot The third step
7 Result 8 Beat 1's framing, now calm The outcome
8 Call to action 8 Logo, URL What to do next

Sixty seconds is roughly 150 spoken words, or about 120 words of on-screen captions read at a comfortable pace, so each caption is one sentence of ten to fifteen words.

Step 1: write the script as captions, not narration

Write every beat as the line that will appear on screen. Captions are how most explainers are watched (muted, in a feed), they remove the need to record a voice, and they force short sentences. If you do record a voice-over later, read the captions aloud and time them; the timecode calculator converts seconds to frames when you place them. Keep the words concrete: a product name, a number, a verb. The old way in beat 2 should be recognisable in three seconds.

Step 2: generate one still per beat

Generate eight stills in the AI image generator at 16:9 (and 9:16 if a vertical cut is planned), 20 credits each, with one style sentence reused across all eight so the video reads as one piece: "clean isometric illustration, soft blue and sand palette, no text". Say "no text" every time; typography goes on the timeline where it can be exact. Beat 7 reuses beat 1's composition with the problem removed, which is the whole story in two frames. Product screens are better as real screenshots than as generated images; generate the world around them instead.

Step 3: animate three beats

Beats 1, 3 and 7 carry the emotion and deserve motion; the three step beats work as stills with a slow scale. Animate with image to video, one camera or motion instruction per clip ("slow push in", "the screen lights up", "papers settle"), in clips of 4, 6, 8, 10 seconds at 16:9 or 9:16, from 25 credits per 360p draft, re-rendering the keepers at 720p. If the explainer is built from a paragraph rather than a storyboard, how to make AI video from text covers writing the shot descriptions from prose.

Step 4: add the captions and the score

Place the captions on the timeline editor as text layers in the lower third (or the upper-middle for a vertical cut, where platform interfaces cover the bottom), one caption per beat, on screen for the beat's full duration. Export an SRT as well so platforms can show the captions natively; how to add captions to a video with AI covers generating and cleaning that file. Generate the score to the video's length in the Music Studio, instrumental, with a brief that names the tone ("light, optimistic, a steady pulse, resolves at the end"), for 25 credits; place it under the captions and duck it if a voice is added later.

Step 5: export and cut the versions

Export 16:9 at 1080p for a website and YouTube. For a vertical cut, regenerate the three moving beats at 9:16 rather than cropping them, crop the stills with the crop image for video preview, and move the captions above the vertical midpoint. A 30-second version for ads keeps beats 1, 3, 7 and 8 and drops the step beats to one; the structure survives the cut because each beat is a self-contained sentence.

What does an explainer video cost in credits?

Eight stills at 20 credits is 160; three 360p drafts at 25 credits is 75; one score is 25 credits. The working total is 260 credits before retries and the 720p renders of the keepers, which a mid-size credit pack covers with room for a vertical cut; a free account's starter credits cover the stills and drafts of a first attempt. Pricing is on /pricing.

How do you pick the visual style for an explainer?

Match the style to the audience and hold it for the whole video: flat isometric illustration for software, warm photographic scenes for services, product-on-surface renders for physical goods. The style sentence in every prompt is what keeps eight separate generations looking like one film; change it and the video falls apart into stock footage. If the explainer needs a recurring character, generate a reference still once and reuse it, as described in the character sheet guide.

Other ways to get from an idea to a finished video: turning product photos into video, How to make AI videos, making a video from photos, How to make a cinemagraph with AI, and How to make a book trailer with AI.

Frequently Asked Questions

How long should an explainer video be?

About 60 seconds, or eight beats of six to eight seconds. A 30-second cut keeps the hook, the idea, the result and the call to action; anything over 90 seconds usually means two videos.

Can AI make an explainer video without a voice-over?

Yes. Write the script as captions, place one caption per beat on the timeline, and export an SRT file as well. Most explainers are watched muted, so captions are the primary track rather than a fallback.

How many shots does a 60-second explainer need?

Eight: one per beat. Three of them animated and five as stills with a slow scale is enough motion; more clips raise the cost without adding clarity.

What does an AI explainer video cost?

Eight stills, three 360p drafts and a score come to 260 credits before retries on PrismPoster, within a single one-off credit pack (packs start at €3.99, no subscription) and within reach of a free account's starter credits for a first version.

AI Video Studio & Timeline Editor

Turn ideas into high-definition AI video clips & timeline cuts

Generate text-to-video and image-to-video in 360p up to 4K, extend scenes per second, and assemble complete multi-track videos with auto-captions and zero desktop install.

Text-to-video & image-to-video from 360p to 4K (16:9 and 9:16)
Multi-track timeline editor with auto-captions and extend-video
150 starter credits, no card — use them within 72 hours
✓ Instant access✓ No credit card required✓ Machine-readable AI provenance✓ No subscription, one-off credit packs

Keep reading