Scientific Visualization Prompts

These prompts treat a visualization as an argument with visible evidence and controls. Use them to explain a relationship, model or dataset; use a different art brief when the actual goal is atmosphere or decoration.

Intended use

The master prompt develops a truth-constrained visual design. Three compact variations specialize it for interactive simulations, explanatory diagrams and handwritten visual notes.

Suitable inputs

Supply the scientific question, source data or equations, variable definitions, units, model assumptions, uncertainty information, audience and delivery medium. If the underlying relationship is only hypothesized, say so explicitly.

Scientific-visualization master prompt

Design an explanatory scientific visualization for [SCIENTIFIC QUESTION OR CLAIM].

Evidence and audience

• Audience: [AUDIENCE AND PRIOR KNOWLEDGE]
• Intended learning outcome: [WHAT THE VIEWER SHOULD BE ABLE TO EXPLAIN, COMPARE OR PREDICT]
• Source data, equations or reference material: [SOURCE MATERIAL]
• Variables and definitions: [VARIABLES]
• Units and dimensional conventions: [UNITS]
• Known uncertainty or error: [UNCERTAINTY INFORMATION]
• Model assumptions and valid domain: [ASSUMPTIONS AND DOMAIN]
• Output medium and dimensions: [WEB, PRINT, SLIDE, NOTEBOOK OR OTHER MEDIUM]

First decide whether the request is genuinely explanatory. If it is only a decorative illustration, label it as such and do not borrow scientific authority through ornamental axes, equations, molecular shapes or data-like marks.

Truth contract

1. Use only variables, relationships and mechanisms supported by the supplied evidence or clearly identified standard knowledge.
2. Do not invent missing values, equations, causal links, confidence intervals, categories or precision.
3. Separate observed data, derived quantities, model output and hypothetical scenarios visually and in the legend.
4. Treat association as association unless a causal design or mechanism justifies stronger language.
5. If the evidence is insufficient for a requested encoding, name the gap and propose a safer alternative.
6. Keep units dimensionally consistent. Show conversions and normalization choices rather than hiding them.

Design the explanation

Produce:

1. One-sentence visual thesis — the exact relationship or process the visual will help the viewer understand.
2. Evidence table — for each mark or channel, list the variable, type, unit, source, uncertainty and whether it is observed, derived, simulated or hypothetical.
3. Visual grammar — specify position, scale, axes, colour, shape, line, annotation, facets and interaction. Explain why each channel fits the variable.
4. Scale and legend plan — state linear, logarithmic, categorical, diverging or other scale choices; domain, zero handling, tick logic, units, reference values and legend wording.
5. Uncertainty plan — show measurement error, intervals, ensembles, sensitivity or unknown regions in a form the audience can interpret. Do not reduce uncertainty to a vague footnote.
6. Narrative sequence — define what appears first, which comparison follows and where annotation should interrupt likely misunderstanding.
7. Implementation specification — identify the rendering form, data transformations, coordinate system, responsive behaviour and deterministic fallback.
8. Caption and alt text — write a concise caption and a substantive text description that communicates the main pattern, axes, units and important caveat.
9. Quality-control checklist — list concrete tests for calculations, units, scale domains, labels, legend consistency, responsive layout and representative values.

Accessibility and motion

• Do not encode meaning by colour alone. Pair colour with position, shape, pattern, line style or direct labels.
• Choose palettes with sufficient light/dark contrast and avoid misleading rainbow scales.
• Make interactive controls keyboard-operable with visible focus and explicit labels.
• Provide the same essential values or conclusion in text, a compact table or an accessible summary.
• Define a reduced-motion path in which transitions become instant or brief and no understanding depends on animation.
• Pause or stop nonessential motion; avoid flashing and uncontrolled continuous animation.
• At a narrow viewport, preserve labels, controls and the key comparison without horizontal page overflow.

Scientific review

Before finalizing, audit:

• Are every variable, unit and sign convention defined?
• Do axes and legends say exactly what was measured or computed?
• Could truncation, area, volume, perspective or colour exaggerate the effect?
• Is uncertainty visible at the level where conclusions are drawn?
• Does the visual imply causality, continuity or precision that the evidence does not support?
• Are baseline, control, boundary conditions and parameter defaults explicit?
• Can a domain expert trace every relationship back to a supplied source or equation?
• Can a nonvisual reader obtain the core result?

Report unresolved scientific questions separately from design choices. A polished visual is not permission to conceal an unresolved model.

Interactive-simulation variation

Apply the master prompt with these additional requirements:

• System state: [STATE VARIABLES]
• Parameters the learner may change: [PARAMETERS, UNITS AND SAFE RANGES]
• Initial and boundary conditions: [INITIAL AND BOUNDARY CONDITIONS]
• Numerical method and time step: [METHOD OR UNKNOWN]
• Comparison or reference solution: [REFERENCE]

Separate the scientific model from rendering. Define which quantities are conserved, bounded or expected to converge; identify numerical artifacts that could be mistaken for science. Use seeded randomness when comparison or replay matters. Provide pause, reset, step, keyboard control, preset and reduced-motion behaviour. Alongside the animation, show current values, units, model limitations and a static representative state. Never add an interaction whose scientific consequence cannot be explained.

Explanatory-diagram variation

Create a diagram for [PROCESS, SYSTEM OR MECHANISM].

Identify actors or components, system boundary, direction of flow, time ordering, feedback, quantities and exceptions. Use arrows only when their meaning is explicit: material flow, information flow, force, causation, sequence and association are not interchangeable. Distinguish known mechanism from hypothesis, and distinguish scale inset from literal adjacency. Supply a legend, numbered reading order, labels that remain readable at [TARGET SIZE], alt text and a monochrome-safe version.

Handwritten-visual-note variation

Turn [SOURCE MATERIAL] into a compact handwritten visual note for [AUDIENCE].

Preserve scientific accuracy while using a human notebook vocabulary: a clear H2-level title, small labelled sketches, short equations, arrows with declared meanings, margin cautions and one worked example. Keep decorative texture subordinate to the explanation. Write variables and units legibly, align fractions and indices, and do not fabricate quotations or observational details to make the page feel authentic. Include a typed transcription and a plain-text summary so handwriting is not the only route to the content.

Usage notes

Begin with the smallest defensible claim. A visualization becomes clearer when it refuses irrelevant variables rather than when it adds another panel. For a live simulation, ask a domain reviewer to inspect both the model and representative outputs; a screenshot can verify layout but cannot verify the equations behind it.