# Image generation prompt — B-research

Use case: scientific-educational
Asset type: one 16:9 landscape research-presentation content slide, not a title cover
Primary request: Create an elegant, publication-quality scientific data page about a fictional water-treatment experiment. The values are a made-up single-run teaching example, not experimental evidence.
Scene/backdrop: quiet warm-white research-paper field with faint grain and a very subtle plotting grid.
Subject: one precise, single-series line chart as the dominant visual. Plot exactly four points in this order and connect them with one descending line: (0, 1.00), (20, 0.93), (40, 0.85), (60, 0.78). Do not add another line, series, confidence interval, error bar, statistical test, or any other values. Label every point with the exact corresponding value.
Style/medium: contemporary scientific-journal editorial design; restrained, clear, mathematically careful; finely drawn axes and point markers; clean Chinese sans-serif typography; one vivid teal or vermilion data line against ink-blue labels and an off-white ground. Keep the chart legible and visually beautiful without dashboard cards.
Composition/framing: 16:9 horizontal content slide with a strong readable headline, large chart area, two short interpretation notes in a clear secondary column, and a small evidence-status line. Let spacing and typographic hierarchy carry the page; optional tiny water/membrane line illustration only if it does not compete with the chart.
Lighting/mood: bright, thoughtful, measured, research-oriented.
Color palette: warm white, deep ink blue, and one restrained bright teal or vermilion chart accent; no extra series colors.
Text (verbatim): Use only these exact visible strings, preserving punctuation and case:

先看趋势，再讨论机理。
归一化通量（示例数据）
时间 (min)
0
20
40
60
J/J0
1.00
0.93
0.85
0.78
单次示例曲线，不能推断显著性。
机理判断仍需重复与对照。
演示数据，非实验结果

Constraints: one four-point series only; horizontal axis is 时间 (min) with ticks 0, 20, 40, 60; vertical axis is J/J0; point labels must match the four given values exactly and remain paired with their x positions; make the title and notes readable at presentation distance; prominently mark the data as a demo, not a real experiment.
Avoid: a second line, duplicated or extra numbers, invented sample sizes, units other than min, p-values, significance claims, confidence bands, error bars, extra conclusions, fabricated laboratory results, dense dashboard panels, stock photography, pseudo-text, and watermark.