
Prepare a science-history explainer that separates a widely repeated legend from what the evidence actually supports. The framework walks you through choosing one claim, building a source-backed claim sheet with dated references, and writing a scene-by-scene brief that shows the popular version first and then the documented version beside it. It includes a fictional storyboard exercise so you can rehearse the structure before committing to a real topic, explicit checks for AI-generated reconstructions and diagrams so nothing invented is presented as a photograph, and guidance on wording uncertainty honestly instead of overstating a result. You also get narration prompts written to be read aloud, a shot list for animated maps and comparison graphics, and a free downloadable PDF creator kit with the full prompt set. Built for documentary-style and short-form vertical video in English or Filipino, and written for creators who would rather be accurate than dramatic.
Build a science history story around evidence
A science explainer should help the viewer follow a question, an observation, and the reasoning that connects them. Dramatic laboratory scenes can support that explanation, but they cannot establish that an experiment happened or that a result is correct. This guide shows how to prepare a research brief and a visual storyboard before using the downloadable Science Legend prompt framework.
1. Narrow the story to one question
Choose a question you can explain without compressing an entire field into one minute. Instead of “the complete history of electricity,” select one documented experiment, one instrument, or one change in scientific understanding. Write a single sentence describing what a viewer should understand by the end. If your draft contains several unrelated discoveries, make them separate videos.
Collect sources before asking AI to write narration. Start with the original publication when available, then use an authoritative educational explanation to understand the context. Museum collection records can help identify instruments; university archives can help establish dates and people. Record the title, author or institution, date, and page or section supporting each important claim. An AI-generated citation is a lead to check, not evidence by itself.
2. Separate the observation from the conclusion
Create a short claim sheet with three fields for each statement: what the source reports, what your narration will say, and what the image will show. This catches a common problem: a visually exciting scene implying more than the evidence supports. An instrument reading, an interpretation of that reading, and a later scientific explanation are different parts of a story.
Keep qualifications that change the meaning. A result obtained under particular conditions should not become a universal rule in the voiceover. Distinguish the date of an experiment from the date it was published, and avoid “first ever” claims unless your sources support that exact scope. If a familiar anecdote cannot be verified, leave it out or identify it as an anecdote rather than staging it as a witnessed event.
3. Plan the visual explanation before the cinematic scene
Begin with a simple diagram or a labelled reference that makes the mechanism understandable. Decide which parts must stay fixed, which part changes, and what the viewer is meant to notice. Use consistent colours and labels across shots. Add text, scale indicators, and arrows in your editing software so you can check them directly instead of relying on generated text.
For historical reconstructions, supply references for the apparatus, room, clothing, and period. Avoid adding modern screens or decorative equipment just because they make a scene look scientific. A reconstruction should be identified as such; it should not be presented as archival footage. If a diagram is schematic or not to scale, say so where that distinction matters to the explanation.
Practice storyboard: reading an invented laboratory record
The following exercise is fictional and demonstrates a production method. It is not a real experiment, a scientific result, or a tested AI output. Imagine a museum display containing an old notebook, a drawing of an instrument, and two recorded observations. The task is to show how a creator would distinguish the record from a later interpretation without inventing a discovery.
Use this planning prompt: “Create a four-scene storyboard from the supplied fictional notebook and instrument drawing. First show the question written in the notebook. Next show the instrument using only features visible in the reference. Then display the two supplied observations side by side, with labels added later in editing. End with the limits of what these records tell us. Do not invent measurements, quotations, dates, scientist names, or a causal conclusion. Mark every visual as an illustrative reconstruction.”
Check the proposed scenes against the supplied material. If the model adds an unrecorded measurement or explains why the observations differ without evidence, remove that claim. This exercise develops a repeatable habit: every factual sentence and every meaningful visual detail should have a source or be clearly identified as an illustration.
Review before publishing
Read the script aloud and compare it with your claim sheet. Check units, labels, dates, names, and the direction of any arrows. Watch the video without sound to see whether the images imply a claim the narration does not make. Then listen without watching to ensure the explanation still makes sense. Keep source credits with the finished work and review permissions for any third-party material.
Use AI drafts as material to review and revise. This framework does not validate scientific claims, replace expert review, or guarantee monetization. For the general workflow, follow the creator instructions; for assembling narration and scenes, see the video editing guide.
Download the Science Legend prompt framework
Share this guide


