一个对比型图形摘要提示词：画“有 vs 没有方法”的 GA
来源：GitHub 仓库 XYD-GIS/paper-toolkits 文件约 80 行，约 3.5 KB
一个对比型图形摘要提示词。
它解决的是：要证明"有 vs 没有方法"差异显著时，画图形摘要。
它的结构：
- 上半"Without {方法名}"：浅灰背景 #F2F2F2，左侧起始场景，右侧负面结果，中间连接箭头，下方斜体灰文案
- 下半"With {方法名}"：白底 + 温暖黄强调，左侧相同起始场景，右侧正面结果，中间连接箭头，下方斜体深色文案
- 底部居中加粗 Arial 结论一句话
- 风格：扁平矢量、白底、Arial、期刊级、4:3 或 1:1
- 负向约束：不要照片、不要 3D、不要投影、不要卡通、不要 emoji、caption 不要讽刺语气、不要长句
填空示例（提速 + 提精度）：
- 方法名 = AdaptiveAttention
- without 文案 = Existing approach: 87% accuracy, 120 minutes per epoch.
- with 文案 = Our method: 93% accuracy, 50 minutes per epoch.
- 结论一句话 = AdaptiveAttention achieves higher accuracy in less time by routing inputs adaptively.
调优提示：
- 左右对比 instead of 上下 → 把 horizontal divider 改为 vertical divider
- 多个指标对比 → 在 caption 中并列列出
- 想强调时间维度 → 把 starting scenario 改为不同 epoch，做时间序列对比
适合谁：
- 写论文要画"有 vs 没有方法"图形摘要的科研小组成员
- 想强调方法差异显著的人
- 想要期刊级 GA 的人
用法： 把这段提示词复制到 AI 工具里，把占位符填上就行。
提示词
A graphical abstract for an academic paper, designed as a vertical comparison split by a horizontal thin gray line in the middle.
UPPER HALF — "Without {方法名}"
- Light gray background tint #F2F2F2
- Left side: render a starting scenario (a small flat-vector illustration)
- Right side: render the negative outcome (problem, error, slow result)
- Connecting arrow in the middle, labeled with the problematic process
- Below, in italic gray: "{without 文案：现有方法的指标}"
LOWER HALF — "With {方法名}"
- White background, with subtle warm yellow accent
- Left side: same starting scenario
- Right side: render the positive outcome (clear result, fast finish, correct answer)
- Connecting arrow in the middle, labeled with our method's name
- Below, in italic dark: "{with 文案：本方法的指标}"
At the bottom center, in bold Arial: "{结论一句话}"
Style: flat vector, white background, Arial sans-serif, journal-quality. Aspect ratio 4:3 or 1:1.
Negative constraints: NO photorealistic, NO 3D, NO drop shadows, NO cartoon, NO emoji, NO sarcastic tone in captions, NO long sentences.
# 填空示例
{方法名} = AdaptiveAttention
{without 文案} = Existing approach: 87% accuracy, 120 minutes per epoch.
{with 文案} = Our method: 93% accuracy, 50 minutes per epoch.
{结论一句话} = AdaptiveAttention achieves higher accuracy in less time by routing inputs adaptively.评论 0
更多
登录后可点赞、收藏、评论和举报。
还没有评论，先发起一个具体问题。