重要前提
安装AI Skills的关键前提是:必须科学上网,且开启TUN模式,这一点至关重要,直接决定安装能否顺利完成,在此郑重提醒三遍:科学上网,科学上网,科学上网。查看完整安装教程 →
latex-posters by k-dense-ai/claude-scientific-skills
npx skills add https://github.com/k-dense-ai/claude-scientific-skills --skill latex-posters研究海报是会议、研讨会和学术活动中进行科学交流的重要媒介。本技能提供了使用 LaTeX 包创建专业、视觉吸引力强的研究海报的全面指导。生成具有适当布局、排版、配色方案和视觉层次结构的出版质量海报。
在以下情况下应使用此技能:
标准工作流程:在创建 LaTeX 海报之前,使用 AI 生成所有主要视觉元素。
这是创建视觉吸引力强的海报的推荐方法:
目标:海报面积的 60-70% 应为 AI 生成的视觉元素,30-40% 为文本。
⚠️ 海报不得在边缘处有文本或内容被截断。
常见的溢出问题:
预防规则:
广告位招租
在这里展示您的产品或服务
触达数万 AI 开发者,精准高效
✅ 良好 - 5 个部分,留有呼吸空间:
- 标题/页眉
- 引言/问题
- 方法
- 结果(1-2 个关键发现)
- 结论
❌ 糟糕 - 8 个以上部分挤在一起:
- 概述、引言、背景、方法、
- 结果 1、结果 2、讨论、结论、未来工作
2. 在 LaTeX 中设置安全边距:
% tikzposter - 添加充足的边距
\documentclass[25pt, a0paper, portrait, margin=25mm]{tikzposter}
% baposter - 确保内容不接触边缘
\begin{poster}{
columns=3,
colspacing=2em, % 列之间的间距
headerheight=0.1\textheight, % 较小的页眉
% 在底部留出空间
}
3. 图形大小 - 切勿使用 100% 宽度:
% 在图形周围留出边距
\includegraphics[width=0.85\linewidth]{figure.png} % 不要用 1.0\linewidth
4. 在打印前检查溢出:
# 编译并在 100% 缩放下检查 PDF
pdflatex poster.tex
# 检查:
# - 在任何边缘被截断的文本
# - 接触页面边界的内容
# - .log 文件中的 Overfull hbox 警告
grep -i "overfull" poster.log
5. 字数限制:
⚠️ AI 生成的视觉化内容中的所有文本必须具有海报可读性。
为海报生成图形时,您必须在每个提示中包含字体大小规格。海报图形是从 4-6 英尺外观看的,因此文本必须足够大。
⚠️ 常见问题:内容溢出和密度过高
AI 生成的海报图形的首要问题是内容过多。这会导致:
解决方案:生成内容最少化的简单图形。
每个海报图形的强制性提示要求:
海报格式要求(严格执行):
- 每个图形绝对最多 3-4 个元素(3 个为理想)
- 整个图形中绝对最多 10 个单词
- 不要有 5 个以上步骤的复杂工作流程(应拆分为 2-3 个简单图形)
- 不要有多级嵌套图表(应扁平化为单层)
- 不要有包含多个子部分的案例研究(每个案例一个关键点)
- 所有文本为巨型粗体(标签 80pt+,关键数字 120pt+)
- 仅使用高对比度(白底黑字或黑底白字,文本不要使用渐变)
- 强制性要求至少 50% 的空白空间(图形的一半应为空白)
- 仅使用粗线(最小 5px+),大图标(最小 200px+)
- 每个图形传达一个单一信息(而不是 3 个相关信息)
⚠️ 生成前:检查您的提示并计算元素数量
每种图形类型的内容限制(严格):
| 图形类型 | 最多元素数 | 最多单词数 | 如果出现以下情况则拒绝 | 良好示例 |
|---|---|---|---|---|
| 流程图 | 最多 3-4 个框 | 8 个单词 | 5 个以上阶段,嵌套步骤 | "发现 → 验证 → 批准"(3 个单词) |
| 关键发现 | 最多 3 个项目 | 9 个单词 | 4 个以上指标,段落 | "95% 准确率" "2 倍更快" "FDA 就绪"(6 个单词) |
| 比较图 | 最多 3 个条形 | 6 个单词 | 4 个以上方法,图例文本 | "我们的:95%" "最佳:85%"(4 个单词) |
| 案例研究 | 1 个案例,3 个元素 | 6 个单词 | 多个案例,子故事 | 徽标 + "18 个月" + "到发现"(2 个单词) |
| 时间线 | 最多 3-4 个点 | 8 个单词 | 逐年详细说明 | "2020 开始" "2022 试验" "2024 获批"(6 个单词) |
示例 - 错误(7 阶段工作流程 - 过于复杂):
# ❌ 糟糕 - 这会创建像药物发现海报那样微小难读的文本
python scripts/generate_schematic.py "药物发现工作流程显示:阶段 1 靶点识别,阶段 2 分子合成,阶段 3 虚拟筛选,阶段 4 AI 先导化合物优化,阶段 5 临床试验设计,阶段 6 FDA 批准。包括每个阶段的成功指标、时间线和验证步骤。" -o figures/workflow.png
# 结果:7 个以上阶段,文本微小,从 6 英尺外难以阅读 - 海报失败
示例 - 正确(简化为 3 个关键阶段):
# ✅ 良好 - 相同内容,拆分为一个简单的高级图形
python scripts/generate_schematic.py "A0 海报格式。超简单 3 框工作流程:'发现' → '验证' → '批准'。每个单词为巨型粗体(120pt+)。粗箭头(10px)。60% 空白空间。无子步骤,无细节。总共 3 个单词。从 10 英尺外可读。" -o figures/workflow_overview.png
# 结果:干净、有冲击力、可读 - 如果需要,可以单独添加细节图形
示例 - 错误(包含多个部分的复杂案例研究):
# ❌ 糟糕 - 创建拥挤难读的部分
python scripts/generate_schematic.py "案例研究:Insilico Medicine(候选药物、发现时间、临床试验)、Recursion Pharma(平台、方法论、结果)、Exscientia(候选药物、FDA 状态、时间线)。包括公司徽标、指标和结果。" -o figures/cases.png
# 结果:3 个案例研究,每个 4 个以上元素 = 总共 12 个以上元素,文本微小
示例 - 正确(一个案例研究,一个关键指标):
# ✅ 良好 - 显示一个案例,一个关键数字
python scripts/generate_schematic.py "A0 海报格式。一个案例研究卡片:公司徽标(大)、'18 个月' 为巨型文本(150pt)、下方为'到发现'(60pt)。总共 3 个元素:徽标 + 数字 + 说明文字。50% 空白空间。从 10 英尺外可读。" -o figures/case_single.png
# 结果:清晰、可读、有冲击力。如果需要 3 个案例,请制作 3 个单独的图形。
示例 - 错误(关键发现过于复杂):
# 糟糕 - 项目太多,细节太多
python scripts/generate_schematic.py "关键发现显示 8 个指标:准确率 95%,精确率 92%,召回率 94%,F1 0.93,AUC 0.97,训练时间 2.3 小时,推理 50ms,模型大小 145MB,并与 5 个基线方法进行比较" -o figures/findings.png
# 结果:图形拥挤,数字微小
示例 - 正确(关键发现简单):
# 良好 - 仅 3 个关键项目,巨型数字
python scripts/generate_schematic.py "A0 海报格式。关键发现,仅包含 3 个大卡片。卡片 1:'95%' 为巨型文本(120pt),下方为'准确率'(48pt)。卡片 2:'2X' 为巨型文本,下方为'更快'。卡片 3:勾选图标,带有'已验证'的大文本。50% 空白空间。高对比度颜色。无其他文本或细节。" -o figures/findings.png
# 结果:粗体、可读、有冲击力的陈述
海报提示的字体大小参考:
| 元素 | 最小尺寸 | 提示关键词 |
|---|---|---|
| 主要数字/指标 | 72pt+ | "巨大"、"非常大"、"巨型"、"海报尺寸" |
| 部分标题 | 60pt+ | "大粗体"、"突出" |
| 标签/说明文字 | 36pt+ | "从 6 英尺外可读"、"清晰标签" |
| 正文文本 | 24pt+ | "海报可读"、"大文本" |
始终在提示中包含:
⚠️ 每个 AI 生成的图形应专注于一个概念,且内容最少化。
问题:生成包含许多元素的复杂图表会导致文本过小。
解决方案:生成元素少、文本大的简单图形。
示例 - 错误(过于复杂,文本会很小):
# 糟糕 - 一个图形中元素太多
python scripts/generate_schematic.py "完整的 ML 流水线,显示数据收集、
包含 5 个步骤的预处理、包含 8 种技术的特征工程、包含超参数调优的模型训练、
包含交叉验证的验证,以及包含监控的部署。
包括所有标签和描述。" -o figures/pipeline.png
示例 - 正确(简单、专注、文本大):
# 良好 - 拆分为多个带有大文本的简单图形
# 图形 1:高级概述(最多 3-4 个元素)
python scripts/generate_schematic.py "A0 海报格式:简单的 4 步流水线。
四个大框:数据 → 处理 → 模型 → 结果。
巨型标签(80pt+),粗箭头,大量空白空间。
总共仅 4 个单词。从 8 英尺外可读。" -o figures/overview.png
# 图形 2:关键结果(突出显示 1 个指标)
python scripts/generate_schematic.py "A0 海报格式:单个关键指标显示。
巨型 '95%' 文本(150pt+),下方为'准确率'(60pt+)。
勾选图标。极简设计,高对比度。
从 10 英尺外可读。" -o figures/accuracy.png
AI 生成海报图形的规则:
| 规则 | 限制 | 原因 |
|---|---|---|
| 每个图形的元素数 | 最多 3-5 个 | 元素越多,文本越小 |
| 每个图形的单词数 | 最多 10-15 个 | 文本越少,字体越大 |
| 流程图步骤数 | 最多 4-5 个 | 保持标签可读 |
| 图表类别数 | 最多 3-4 个 | 防止拥挤 |
| 嵌套层级数 | 最多 1-2 级 | 避免复杂 |
将复杂内容拆分为多个简单图形:
不要使用 1 个包含 12 个元素的复杂图表:
→ 创建 3 个包含 4 个元素的简单图表
→ 每个图形可以有更大的文本
→ 以清晰的视觉流排列在海报中
⚠️ 在生成任何图形之前,请检查您的内容计划:
对于每个计划的图形,询问以下问题:
元素数量:我能否用 3-4 个项目或更少来描述这个?
复杂性检查:这是一个多阶段工作流程(5 个以上步骤)还是嵌套图表?
单词数量:我能否用 10 个或更少的单词描述所有文本?
信息清晰度:这个图形是否传达了一个清晰的信息?
总是失败的常见模式(拒绝这些):
通过生成前审查后,确定所需的视觉元素:
⚠️ 关键:在继续之前检查步骤 0 的清单。
为每种元素类型使用适当的工具:
用于示意图和图表(scientific-schematics):
# 创建 figures 目录
mkdir -p figures
# 药物发现工作流程 - 仅高级视图,3 个阶段
# 糟糕:"阶段 1:靶点识别,阶段 2:分子合成,阶段 3:虚拟筛选,阶段 4:AI 先导化合物优化..."
# 良好:压缩为 3 个超级阶段
python scripts/generate_schematic.py "A0 海报格式。超简单 3 框工作流程:'发现'(120pt 粗体) → '验证'(120pt 粗体) → '批准'(120pt 粗体)。粗箭头(10px)。60% 空白空间。仅这 3 个单词。无子步骤。从 12 英尺外可读。" -o figures/workflow_simple.png
# 系统架构 - 最多 3 个组件
python scripts/generate_schematic.py "A0 海报格式。超简单 3 组件堆栈:'数据'框(120pt) → 'AI 模型'框(120pt) → '预测'框(120pt)。粗垂直箭头。60% 空白空间。仅 3 个单词。从 12 英尺外可读。" -o figures/architecture.png
# 时间线 - 仅 3 个关键里程碑(非逐年)
# 糟糕:"2018、2019、2020、2021、2022、2023、2024 及事件"
# 良好:仅 3 个突破性时刻
python scripts/generate_schematic.py "A0 海报格式。时间线,仅包含 3 个点:'2018' + 图标,'2021' + 图标,'2024' + 图标。巨型年份(120pt)。大图标。60% 空白空间。无连接线或细节。从 12 英尺外可读。" -o figures/timeline.png
# 案例研究 - 一个案例,一个关键指标
# 糟糕:"3 个案例研究:Insilico(细节)、Recursion(细节)、Exscientia(细节)"
# 良好:一个案例,一个数字
python scripts/generate_schematic.py "A0 海报格式。一个案例研究:大徽标 + '18 个月'(150pt 粗体) + '到发现'(60pt)。总共 3 个元素。60% 空白空间。从 12 英尺外可读。" -o figures/case1.png
# 如果需要 3 个案例 → 制作 3 个单独的简单图形(而不是一个复杂图形)
用于风格化块和图形(Nano Banana Pro):
# 标题块 - 简单
python scripts/generate_schematic.py "A0 海报格式。标题块:'用于药物发现的 ML' 为巨大粗体文本(120pt+)。深蓝色背景。一个微妙图标。无其他文本。40% 空白空间。从 15 英尺外可读。" -o figures/title_block.png
# 引言视觉 - 简单,仅 3 个元素
python scripts/generate_schematic.py "A0 海报格式。简单问题视觉,仅包含 3 个图标:药物图标、箭头、靶标图标。每个图标一个标签(80pt+)。50% 空白空间。无详细文本。从 8 英尺外可读。" -o figures/intro_visual.png
# 结论/总结 - 仅 3 个项目,巨型数字
python scripts/generate_schematic.py "A0 海报格式。关键发现,仅包含恰好 3 个卡片。卡片 1:'95%'(150pt 字体)和'准确率'(60pt)。卡片 2:'2X'(150pt)和'更快'(60pt)。卡片 3:勾选图标和'就绪'(60pt)。50% 空白空间。无其他文本。从 10 英尺外可读。" -o figures/conclusions_graphic.png
# 背景视觉 - 简单,仅 3 个图标
python scripts/generate_schematic.py "A0 海报格式。简单视觉,仅包含 3 个大图标排成一行:问题图标 → 挑战图标 → 影响图标。每个图标一个单词标签(80pt+)。50% 空白空间。无详细文本。从 8 英尺外可读。" -o figures/background_visual.png
用于数据可视化 - 简单,最多 3 个条形:
# 简单图表,仅包含 3 个条形,巨型标签
python scripts/generate_schematic.py "A0 海报格式。简单条形图,仅包含 3 个条形:基线(70%)、现有(85%)、我们的(95%)。条形上的巨型百分比标签(100pt+)。无轴标签,无图例,无网格线。我们的条形以不同颜色突出显示。40% 空白空间。从 8 英尺外可读。" -o figures/comparison_chart.png
⚠️ 关键:在将每个生成的图形添加到海报之前进行检查。
对于每个生成的图形,以 25% 缩放打开并检查:
✅ 通过标准(必须全部为真):
❌ 失败标准(如果任何一项为真,则重新生成):
常见失败及修复方法:
如果任何图形未通过上述检查,请勿继续组装。
所有图形通过生成后审查后,将它们包含在您的海报模板中:
tikzposter 示例:
\documentclass[25pt, a0paper, portrait]{tikzposter}
\begin{document}
\maketitle
\begin{columns}
\column{0.5}
\block{引言}{
\centering
\includegraphics[width=0.85\linewidth]{figures/intro_visual.png}
\vspace{0.5em}
此处为简要上下文文本(最多 2-3 句)。
}
\block{方法}{
\centering
\includegraphics[width=0.9\linewidth]{figures/methods_flowchart.png}
}
\column{0.5}
\block{结果}{
\begin{minipage}{0.48\linewidth}
\centering
\includegraphics[width=\linewidth]{figures/result_1.png}
\end{minipage}
\hfill
\begin{minipage}{0.48\linewidth}
\centering
\includegraphics[width=\linewidth]{figures/result_2.png}
\end{minipage}
\vspace{0.5em}
关键发现,以 3-4 个要点列出。
}
\block{结论}{
\centering
\includegraphics[width=0.8\linewidth]{figures/conclusions_graphic.png}
}
\end{columns}
\end{document}
baposter 示例:
\headerbox{方法}{name=methods,column=0,row=0}{
\centering
\includegraphics[width=0.95\linewidth]{figures/methods_flowchart.png}
}
\headerbox{结果}{name=results,column=1,row=0}{
\includegraphics[width=\linewidth]{figures/comparison_chart.png}
\vspace{0.3em}
关键发现:我们的方法达到 92% 准确率。
}
包含所有质量检查的完整工作流程:
# 步骤 0:生成前审查(强制)
# 内容计划:药物发现海报
# - 工作流程:7 个阶段 → ❌ 太多 → 减少到 3 个超级阶段 ✅
# - 3 个案例研究 → ❌ 太多 → 每个图形一个案例(制作 3 个图形) ✅
# - 时间线 2018-2024 → ❌ 太详细 → 仅 3 个关键年份 ✅
# 步骤 1:创建 figures 目录
mkdir -p figures
# 步骤 2:生成具有严格限制的超简单图形
# 工作流程 - 仅高级视图(从 7 个阶段压缩为 3 个)
python scripts/generate_schematic.py "A0 海报格式。超简单 3 框工作流程:'发现' → '验证' → '批准'。每个单词 120pt+ 粗体。粗箭头(10px)。60% 空白空间。总共仅 3 个单词。从 12 英尺外可读。" -o figures/workflow.png
# 案例研究 1 - 一个案例,一个指标(将制作 3 个单独的图形)
python scripts/generate_schematic.py "A0 海报格式。一个案例:公司徽标 + '18 个月'(150pt 粗体) + '到药物发现'(60pt)。仅 3 个元素。60% 空白空间。从 12 英尺外可读。" -o figures/case1.png
python scripts/generate_schematic.py "A0 海报格式。一个案例:公司徽标 + '95% 成功'(150pt 粗体) + '在试验中'(60pt)。仅 3 个元素。60% 空白空间。" -o figures/case2.png
python scripts/generate_schematic.py "A0 海报格式。一个案例:公司徽标 + 'FDA 批准'(150pt 粗体) + '2024'(60pt)。仅 3 个元素。60% 空白空间。" -o figures/case3.png
# 时间线 - 仅 3 个关键年份(非 7 年)
python scripts/generate_schematic.py "A0 海报格式。仅 3 年:'2018'(150pt)+ 图标,'2021'(150pt)+ 图标,'2024'(150pt)+ 图标。大图标。60% 空白空间。无线条或细节。从 12 英尺外可读。" -o figures/timeline.png
# 结果 - 仅 2 个条形(我们的方法 vs 最佳基线,非 5 种方法)
python scripts/generate_schematic.py "A0 海报格式。仅两个条形:'基线 70%' 和 '我们的 95%'(突出显示)。条形上的巨型百分比(150pt)。无轴,无图例。60% 空白空间。从 12 英尺外可读。" -o figures/results.png
# 步骤 2b:生成后审查(强制)
# 以 25% 缩放打开每个图形:
# ✅ workflow.png:3 个元素,文本可读,60% 空白 - 通过
# ✅ case1.png:3 个元素,巨型数字,干净 - 通过
# ✅ case2.png:3 个元素,巨型数字,干净 - 通过
# ✅ case3.png:3 个元素,巨型数字,干净 - 通过
# ✅ timeline.png:3 个元素,可读,简单 - 通过
# ✅ results.png:2 个条形,巨型百分比,清晰 - 通过
# 全部通过 → 继续组装
# 步骤 3:编译 LaTeX 海报
pdflatex poster.tex
# 步骤 4:PDF 溢出检查(参见第 11 节)
grep "Overfull" poster.log
# 以 100% 打开并检查所有 4 个边缘
如果任何图形未通过步骤 2b 审查:
⚠️ 关键:每个图形必须有一个信息,且最多 3-4 个元素。
绝对限制 - 这些不是指南,而是硬性限制:
对于每个海报部分 - 严格要求:
| 部分 | 最多元素数 | 最多单词数 | 示例提示(必需模式) |
|---|---|---|---|
| 引言 | 3 个图标 | 6 个单词 | "A0 海报格式:超简单 3 个图标:[图标1] [图标2] [图标3]。单字标签(100pt 粗体)。60% 空白空间。总共 3 个单词。" |
| 方法 | 3 个框 | 6 个单词 | "A0 海报格式:超简单 3 框工作流程:'步骤1' → '步骤2' → '步骤3'。巨型标签(120pt+)。60% 空白空间。仅 3 个单词。" |
| 结果 | 2-3 个条形 | 6 个单词 | "A0 海报格式:两个条形:'基线 70%' '我们的 95%'。条形上的巨型百分比(150pt+)。无轴。60% 空白空间。" |
| 结论 | 3 个卡片 | 9 个单词 | "A0 海报格式:三个卡片:'95%'(150pt)'准确率','2X'(150pt)'更快',勾选'就绪'。60% 空白空间。" |
| 案例研究 | 3 个元素 | 5 个单词 | "A0 海报格式:一个案例:徽标 + '18 个月'(150pt)+ '到发现'(60pt)。60% 空白空间。" |
| 时间线 | 3 个点 | 3 个单词 | "A0 海报格式:仅三年:'2018' '2021' '2024'(每个 150pt)。大图标。60% 空白空间。无细节。" |
强制性提示元素(全部必需,无例外):
总是失败的模式(立即拒绝):
有效的模式:
有关创建示意图的详细指导,请参阅 scientific-schematics 技能文档。
关键能力:
支持三个主要的 LaTeX 海报包,每个都有独特的优势。有关详细比较和包特定指导,请参阅 references/latex_poster_packages.md。
beamerposter:
tikzposter:
baposter:
遵循视觉传达原则创建有效的海报布局
Research posters are a critical medium for scientific communication at conferences, symposia, and academic events. This skill provides comprehensive guidance for creating professional, visually appealing research posters using LaTeX packages. Generate publication-quality posters with proper layout, typography, color schemes, and visual hierarchy.
This skill should be used when:
STANDARD WORKFLOW: Generate ALL major visual elements using AI before creating the LaTeX poster.
This is the recommended approach for creating visually compelling posters:
Target: 60-70% of poster area should be AI-generated visuals, 30-40% text.
⚠️ POSTERS MUST NOT HAVE TEXT OR CONTENT CUT OFF AT EDGES.
Common Overflow Problems:
Prevention Rules:
1. Limit Content Sections (MAXIMUM 5-6 sections for A0):
✅ GOOD - 5 sections with room to breathe:
- Title/Header
- Introduction/Problem
- Methods
- Results (1-2 key findings)
- Conclusions
❌ BAD - 8+ sections crammed together:
- Overview, Introduction, Background, Methods,
- Results 1, Results 2, Discussion, Conclusions, Future Work
2. Set Safe Margins in LaTeX:
% tikzposter - add generous margins
\documentclass[25pt, a0paper, portrait, margin=25mm]{tikzposter}
% baposter - ensure content doesn't touch edges
\begin{poster}{
columns=3,
colspacing=2em, % Space between columns
headerheight=0.1\textheight, % Smaller header
% Leave space at bottom
}
3. Figure Sizing - Never 100% Width:
% Leave margins around figures
\includegraphics[width=0.85\linewidth]{figure.png} % NOT 1.0\linewidth
4. Check for Overflow Before Printing:
# Compile and check PDF at 100% zoom
pdflatex poster.tex
# Look for:
# - Text cut off at any edge
# - Content touching page boundaries
# - Overfull hbox warnings in .log file
grep -i "overfull" poster.log
5. Word Count Limits:
⚠️ ALL text within AI-generated visualizations MUST be poster-readable.
When generating graphics for posters, you MUST include font size specifications in EVERY prompt. Poster graphics are viewed from 4-6 feet away, so text must be LARGE.
⚠️ COMMON PROBLEM: Content Overflow and Density
The #1 issue with AI-generated poster graphics is TOO MUCH CONTENT. This causes:
SOLUTION: Generate SIMPLE graphics with MINIMAL content.
MANDATORY prompt requirements for EVERY poster graphic:
POSTER FORMAT REQUIREMENTS (STRICTLY ENFORCE):
- ABSOLUTE MAXIMUM 3-4 elements per graphic (3 is ideal)
- ABSOLUTE MAXIMUM 10 words total in the entire graphic
- NO complex workflows with 5+ steps (split into 2-3 simple graphics instead)
- NO multi-level nested diagrams (flatten to single level)
- NO case studies with multiple sub-sections (one key point per case)
- ALL text GIANT BOLD (80pt+ for labels, 120pt+ for key numbers)
- High contrast ONLY (dark on white OR white on dark, NO gradients with text)
- MANDATORY 50% white space minimum (half the graphic should be empty)
- Thick lines only (5px+ minimum), large icons (200px+ minimum)
- ONE SINGLE MESSAGE per graphic (not 3 related messages)
⚠️ BEFORE GENERATING: Review your prompt and count elements
Content limits per graphic type (STRICT):
| Graphic Type | Max Elements | Max Words | Reject If | Good Example |
|---|---|---|---|---|
| Flowchart | 3-4 boxes MAX | 8 words | 5+ stages, nested steps | "DISCOVER → VALIDATE → APPROVE" (3 words) |
| Key findings | 3 items MAX | 9 words | 4+ metrics, paragraphs | "95% ACCURATE" "2X FASTER" "FDA READY" (6 words) |
| Comparison chart | 3 bars MAX | 6 words | 4+ methods, legend text | "OURS: 95%" "BEST: 85%" (4 words) |
| Case study | 1 case, 3 elements | 6 words | Multiple cases, substories | Logo + "18 MONTHS" + "to discovery" (2 words) |
Example - WRONG (7-stage workflow - TOO COMPLEX):
# ❌ BAD - This creates tiny unreadable text like the drug discovery poster
python scripts/generate_schematic.py "Drug discovery workflow showing: Stage 1 Target Identification, Stage 2 Molecular Synthesis, Stage 3 Virtual Screening, Stage 4 AI Lead Optimization, Stage 5 Clinical Trial Design, Stage 6 FDA Approval. Include success metrics, timelines, and validation steps for each stage." -o figures/workflow.png
# Result: 7+ stages with tiny text, unreadable from 6 feet - POSTER FAILURE
Example - CORRECT (simplified to 3 key stages):
# ✅ GOOD - Same content, split into ONE simple high-level graphic
python scripts/generate_schematic.py "POSTER FORMAT for A0. ULTRA-SIMPLE 3-box workflow: 'DISCOVER' → 'VALIDATE' → 'APPROVE'. Each word in GIANT bold (120pt+). Thick arrows (10px). 60% white space. NO substeps, NO details. 3 words total. Readable from 10 feet." -o figures/workflow_overview.png
# Result: Clean, impactful, readable - can add detail graphics separately if needed
Example - WRONG (complex case studies with multiple sections):
# ❌ BAD - Creates cramped unreadable sections
python scripts/generate_schematic.py "Case studies: Insilico Medicine (drug candidate, discovery time, clinical trials), Recursion Pharma (platform, methodology, results), Exscientia (drug candidates, FDA status, timeline). Include company logos, metrics, and outcomes." -o figures/cases.png
# Result: 3 case studies with 4+ elements each = 12+ total elements, tiny text
Example - CORRECT (one case study, one key metric):
# ✅ GOOD - Show ONE case with ONE key number
python scripts/generate_schematic.py "POSTER FORMAT for A0. ONE case study card: Company logo (large), '18 MONTHS' in GIANT text (150pt), 'to discovery' below (60pt). 3 elements total: logo + number + caption. 50% white space. Readable from 10 feet." -o figures/case_single.png
# Result: Clear, readable, impactful. Make 3 separate graphics if you need 3 cases.
Example - WRONG (key findings too complex):
# BAD - too many items, too much detail
python scripts/generate_schematic.py "Key findings showing 8 metrics: accuracy 95%, precision 92%, recall 94%, F1 0.93, AUC 0.97, training time 2.3 hours, inference 50ms, model size 145MB with comparison to 5 baseline methods" -o figures/findings.png
# Result: Cramped graphic with tiny numbers
Example - CORRECT (key findings simple):
# GOOD - only 3 key items, giant numbers
python scripts/generate_schematic.py "POSTER FORMAT for A0. KEY FINDINGS with ONLY 3 large cards. Card 1: '95%' in GIANT text (120pt) with 'ACCURACY' below (48pt). Card 2: '2X' in GIANT text with 'FASTER' below. Card 3: checkmark icon with 'VALIDATED' in large text. 50% white space. High contrast colors. NO other text or details." -o figures/findings.png
# Result: Bold, readable impact statement
Font size reference for poster prompts:
| Element | Minimum Size | Prompt Keywords |
|---|---|---|
| Main numbers/metrics | 72pt+ | "huge", "very large", "giant", "poster-size" |
| Section titles | 60pt+ | "large bold", "prominent" |
| Labels/captions | 36pt+ | "readable from 6 feet", "clear labels" |
| Body text | 24pt+ | "poster-readable", "large text" |
Always include in prompts:
⚠️ Each AI-generated graphic should focus on ONE concept with MINIMAL content.
Problem : Generating complex diagrams with many elements leads to small text.
Solution : Generate SIMPLE graphics with FEW elements and LARGE text.
Example - WRONG (too complex, text will be small):
# BAD - too many elements in one graphic
python scripts/generate_schematic.py "Complete ML pipeline showing data collection,
preprocessing with 5 steps, feature engineering with 8 techniques, model training
with hyperparameter tuning, validation with cross-validation, and deployment with
monitoring. Include all labels and descriptions." -o figures/pipeline.png
Example - CORRECT (simple, focused, large text):
# GOOD - split into multiple simple graphics with large text
# Graphic 1: High-level overview (3-4 elements max)
python scripts/generate_schematic.py "POSTER FORMAT for A0: Simple 4-step pipeline.
Four large boxes: DATA → PROCESS → MODEL → RESULTS.
GIANT labels (80pt+), thick arrows, lots of white space.
Only 4 words total. Readable from 8 feet." -o figures/overview.png
# Graphic 2: Key result (1 metric highlighted)
python scripts/generate_schematic.py "POSTER FORMAT for A0: Single key metric display.
Giant '95%' text (150pt+) with 'ACCURACY' below (60pt+).
Checkmark icon. Minimal design, high contrast.
Readable from 10 feet." -o figures/accuracy.png
Rules for AI-generated poster graphics:
| Rule | Limit | Reason |
|---|---|---|
| Elements per graphic | 3-5 maximum | More elements = smaller text |
| Words per graphic | 10-15 maximum | Minimal text = larger fonts |
| Flowchart steps | 4-5 maximum | Keeps labels readable |
| Chart categories | 3-4 maximum | Prevents crowding |
| Nested levels | 1-2 maximum | Avoids complexity |
Split complex content into multiple simple graphics:
Instead of 1 complex diagram with 12 elements:
→ Create 3 simple diagrams with 4 elements each
→ Each graphic can have LARGER text
→ Arrange in poster with clear visual flow
⚠️ BEFORE generating ANY graphics, review your content plan:
For EACH planned graphic, ask these questions:
Element count : Can I describe this in 3-4 items or less?
Complexity check : Is this a multi-stage workflow (5+ steps) or nested diagram?
Word count : Can I describe all text in 10 words or less?
Message clarity : Does this graphic convey ONE clear message?
Common patterns that ALWAYS fail (reject these):
After passing the pre-generation review, identify visual elements needed:
⚠️ CRITICAL: Review Step 0 checklist before proceeding.
Use the appropriate tool for each element type:
For Schematics and Diagrams (scientific-schematics):
# Create figures directory
mkdir -p figures
# Drug discovery workflow - HIGH-LEVEL ONLY, 3 stages
# BAD: "Stage 1: Target ID, Stage 2: Molecular Synthesis, Stage 3: Virtual Screening, Stage 4: AI Lead Opt..."
# GOOD: Collapse to 3 mega-stages
python scripts/generate_schematic.py "POSTER FORMAT for A0. ULTRA-SIMPLE 3-box workflow: 'DISCOVER' (120pt bold) → 'VALIDATE' (120pt bold) → 'APPROVE' (120pt bold). Thick arrows (10px). 60% white space. ONLY these 3 words. NO substeps. Readable from 12 feet." -o figures/workflow_simple.png
# System architecture - MAXIMUM 3 components
python scripts/generate_schematic.py "POSTER FORMAT for A0. ULTRA-SIMPLE 3-component stack: 'DATA' box (120pt) → 'AI MODEL' box (120pt) → 'PREDICTION' box (120pt). Thick vertical arrows. 60% white space. 3 words only. Readable from 12 feet." -o figures/architecture.png
# Timeline - ONLY 3 key milestones (not year-by-year)
# BAD: "2018, 2019, 2020, 2021, 2022, 2023, 2024 with events"
# GOOD: Only 3 breakthrough moments
python scripts/generate_schematic.py "POSTER FORMAT for A0. Timeline with ONLY 3 points: '2018' + icon, '2021' + icon, '2024' + icon. GIANT years (120pt). Large icons. 60% white space. NO connecting lines or details. Readable from 12 feet." -o figures/timeline.png
# Case study - ONE case, ONE key metric
# BAD: "3 case studies: Insilico (details), Recursion (details), Exscientia (details)"
# GOOD: ONE case with ONE number
python scripts/generate_schematic.py "POSTER FORMAT for A0. ONE case study: Large logo + '18 MONTHS' (150pt bold) + 'to discovery' (60pt). 3 elements total. 60% white space. Readable from 12 feet." -o figures/case1.png
# If you need 3 cases → make 3 separate simple graphics (not one complex graphic)
For Stylized Blocks and Graphics (Nano Banana Pro):
# Title block - SIMPLE
python scripts/generate_schematic.py "POSTER FORMAT for A0. Title block: 'ML FOR DRUG DISCOVERY' in HUGE bold text (120pt+). Dark blue background. ONE subtle icon. NO other text. 40% white space. Readable from 15 feet." -o figures/title_block.png
# Introduction visual - SIMPLE, 3 elements only
python scripts/generate_schematic.py "POSTER FORMAT for A0. SIMPLE problem visual with ONLY 3 icons: drug icon, arrow, target icon. ONE label per icon (80pt+). 50% white space. NO detailed text. Readable from 8 feet." -o figures/intro_visual.png
# Conclusion/summary - ONLY 3 items, GIANT numbers
python scripts/generate_schematic.py "POSTER FORMAT for A0. KEY FINDINGS with EXACTLY 3 cards only. Card 1: '95%' (150pt font) with 'ACCURACY' (60pt). Card 2: '2X' (150pt) with 'FASTER' (60pt). Card 3: checkmark icon with 'READY' (60pt). 50% white space. NO other text. Readable from 10 feet." -o figures/conclusions_graphic.png
# Background visual - SIMPLE, 3 icons only
python scripts/generate_schematic.py "POSTER FORMAT for A0. SIMPLE visual with ONLY 3 large icons in a row: problem icon → challenge icon → impact icon. ONE word label each (80pt+). 50% white space. NO detailed text. Readable from 8 feet." -o figures/background_visual.png
For Data Visualizations - SIMPLE, 3 bars max:
# SIMPLE chart with ONLY 3 bars, GIANT labels
python scripts/generate_schematic.py "POSTER FORMAT for A0. SIMPLE bar chart with ONLY 3 bars: BASELINE (70%), EXISTING (85%), OURS (95%). GIANT percentage labels ON the bars (100pt+). NO axis labels, NO legend, NO gridlines. Our bar highlighted in different color. 40% white space. Readable from 8 feet." -o figures/comparison_chart.png
⚠️ CRITICAL: Review EVERY generated graphic before adding to poster.
For each generated figure, open at 25% zoom and check:
✅ PASS criteria (all must be true):
❌ FAIL criteria (regenerate if ANY are true):
Common failures and fixes:
DO NOT PROCEED to assembly if ANY graphic fails the checks above.
After all figures pass the post-generation review, include them in your poster template:
tikzposter example:
\documentclass[25pt, a0paper, portrait]{tikzposter}
\begin{document}
\maketitle
\begin{columns}
\column{0.5}
\block{Introduction}{
\centering
\includegraphics[width=0.85\linewidth]{figures/intro_visual.png}
\vspace{0.5em}
Brief context text here (2-3 sentences max).
}
\block{Methods}{
\centering
\includegraphics[width=0.9\linewidth]{figures/methods_flowchart.png}
}
\column{0.5}
\block{Results}{
\begin{minipage}{0.48\linewidth}
\centering
\includegraphics[width=\linewidth]{figures/result_1.png}
\end{minipage}
\hfill
\begin{minipage}{0.48\linewidth}
\centering
\includegraphics[width=\linewidth]{figures/result_2.png}
\end{minipage}
\vspace{0.5em}
Key findings in 3-4 bullet points.
}
\block{Conclusions}{
\centering
\includegraphics[width=0.8\linewidth]{figures/conclusions_graphic.png}
}
\end{columns}
\end{document}
baposter example:
\headerbox{Methods}{name=methods,column=0,row=0}{
\centering
\includegraphics[width=0.95\linewidth]{figures/methods_flowchart.png}
}
\headerbox{Results}{name=results,column=1,row=0}{
\includegraphics[width=\linewidth]{figures/comparison_chart.png}
\vspace{0.3em}
Key finding: Our method achieves 92% accuracy.
}
Full workflow with ALL quality checks:
# STEP 0: Pre-Generation Review (MANDATORY)
# Content plan: Drug discovery poster
# - Workflow: 7 stages → ❌ TOO MANY → Reduce to 3 mega-stages ✅
# - 3 case studies → ❌ TOO MANY → One case per graphic (make 3 graphics) ✅
# - Timeline 2018-2024 → ❌ TOO DETAILED → Only 3 key years ✅
# STEP 1: Create figures directory
mkdir -p figures
# STEP 2: Generate ULTRA-SIMPLE graphics with strict limits
# Workflow - HIGH-LEVEL ONLY (collapsed from 7 stages to 3)
python scripts/generate_schematic.py "POSTER FORMAT for A0. ULTRA-SIMPLE 3-box workflow: 'DISCOVER' → 'VALIDATE' → 'APPROVE'. Each word 120pt+ bold. Thick arrows (10px). 60% white space. ONLY 3 words total. Readable from 12 feet." -o figures/workflow.png
# Case study 1 - ONE case, ONE metric (will make 3 separate graphics)
python scripts/generate_schematic.py "POSTER FORMAT for A0. ONE case: Company logo + '18 MONTHS' (150pt bold) + 'to drug discovery' (60pt). 3 elements only. 60% white space. Readable from 12 feet." -o figures/case1.png
python scripts/generate_schematic.py "POSTER FORMAT for A0. ONE case: Company logo + '95% SUCCESS' (150pt bold) + 'in trials' (60pt). 3 elements only. 60% white space." -o figures/case2.png
python scripts/generate_schematic.py "POSTER FORMAT for A0. ONE case: Company logo + 'FDA APPROVED' (150pt bold) + '2024' (60pt). 3 elements only. 60% white space." -o figures/case3.png
# Timeline - ONLY 3 key years (not 7 years)
python scripts/generate_schematic.py "POSTER FORMAT for A0. ONLY 3 years: '2018' (150pt) + icon, '2021' (150pt) + icon, '2024' (150pt) + icon. Large icons. 60% white space. NO lines or details. Readable from 12 feet." -o figures/timeline.png
# Results - ONLY 2 bars (our method vs best baseline, not 5 methods)
python scripts/generate_schematic.py "POSTER FORMAT for A0. TWO bars only: 'BASELINE 70%' and 'OURS 95%' (highlighted). GIANT percentages (150pt) ON bars. NO axis, NO legend. 60% white space. Readable from 12 feet." -o figures/results.png
# STEP 2b: Post-Generation Review (MANDATORY)
# Open each figure at 25% zoom:
# ✅ workflow.png: 3 elements, text readable, 60% white - PASS
# ✅ case1.png: 3 elements, giant numbers, clean - PASS
# ✅ case2.png: 3 elements, giant numbers, clean - PASS
# ✅ case3.png: 3 elements, giant numbers, clean - PASS
# ✅ timeline.png: 3 elements, readable, simple - PASS
# ✅ results.png: 2 bars, giant percentages, clear - PASS
# ALL PASS → Proceed to assembly
# STEP 3: Compile LaTeX poster
pdflatex poster.tex
# STEP 4: PDF Overflow Check (see Section 11)
grep "Overfull" poster.log
# Open at 100% and check all 4 edges
If ANY graphic fails Step 2b review:
⚠️ CRITICAL: Each graphic must have ONE message and MAXIMUM 3-4 elements.
ABSOLUTE LIMITS - These are NOT guidelines, these are HARD LIMITS:
For each poster section - STRICT requirements:
| Section | Max Elements | Max Words | Example Prompt (REQUIRED PATTERN) |
|---|---|---|---|
| Introduction | 3 icons | 6 words | "POSTER FORMAT for A0: ULTRA-SIMPLE 3 icons: [icon1] [icon2] [icon3]. ONE WORD labels (100pt bold). 60% white space. 3 words total." |
| Methods | 3 boxes | 6 words | "POSTER FORMAT for A0: ULTRA-SIMPLE 3-box workflow: 'STEP1' → 'STEP2' → 'STEP3'. GIANT labels (120pt+). 60% white space. 3 words only." |
| Results | 2-3 bars | 6 words | "POSTER FORMAT for A0: TWO bars: 'BASELINE 70%' 'OURS 95%'. GIANT percentages (150pt+) ON bars. NO axis. 60% white space." |
| Conclusions | 3 cards | 9 words | "POSTER FORMAT for A0: THREE cards: '95%' (150pt) 'ACCURATE', '2X' (150pt) 'FASTER', checkmark 'READY'. 60% white space." |
| Case Study | 3 elements | 5 words | "POSTER FORMAT for A0: ONE case: logo + '18 MONTHS' (150pt) + 'to discovery' (60pt). 60% white space." |
MANDATORY prompt elements (ALL required, NO exceptions):
PATTERNS THAT ALWAYS FAIL (REJECT IMMEDIATELY):
PATTERNS THAT WORK:
For detailed guidance on creating schematics, refer to the scientific-schematics skill documentation.
Key capabilities:
Support for three major LaTeX poster packages, each with distinct advantages. For detailed comparison and package-specific guidance, refer to references/latex_poster_packages.md.
beamerposter :
tikzposter :
baposter :
Create effective poster layouts following visual communication principles. For comprehensive layout guidance, refer to references/poster_layout_design.md.
Common Poster Sections :
Layout Strategies :
Apply evidence-based design principles for maximum impact. For detailed design guidance, refer to references/poster_design_principles.md.
Typography :
Color and Contrast :
Visual Elements :
Content Guidelines :
Support for international and conference-specific poster dimensions:
International Standards :
North American Standards :
Orientation :
Provide ready-to-use templates for each major package. Templates available in assets/ directory.
beamerposter Templates :
beamerposter_classic.tex - Traditional academic stylebeamerposter_modern.tex - Clean, minimal designbeamerposter_colorful.tex - Vibrant theme with blockstikzposter Templates :
tikzposter_default.tex - Standard tikzposter layouttikzposter_rays.tex - Modern design with ray themetikzposter_wave.tex - Professional wave-style themebaposter Templates :
baposter_portrait.tex - Classic portrait layoutbaposter_landscape.tex - Landscape multi-columnbaposter_minimal.tex - Minimalist designOptimize visual content for poster presentations:
Best Practices :
LaTeX Figure Commands :
% Include graphics package
\usepackage{graphicx}
% Simple figure
\includegraphics[width=0.8\linewidth]{figure.pdf}
% Figure with caption in tikzposter
\block{Results}{
\begin{tikzfigure}
\includegraphics[width=0.9\linewidth]{results.png}
\end{tikzfigure}
}
% Multiple subfigures
\usepackage{subcaption}
\begin{figure}
\begin{subfigure}{0.48\linewidth}
\includegraphics[width=\linewidth]{fig1.pdf}
\caption{Condition A}
\end{subfigure}
\begin{subfigure}{0.48\linewidth}
\includegraphics[width=\linewidth]{fig2.pdf}
\caption{Condition B}
\end{subfigure}
\end{figure}
Provide professional color palettes for various contexts:
Academic Institution Colors :
Scientific Color Palettes (color-blind friendly):
Package-Specific Theme Selection :
beamerposter :
\usetheme{Berlin}
\usecolortheme{beaver}
tikzposter :
\usetheme{Rays}
\usecolorstyle{Denmark}
baposter :
\begin{poster}{
background=plain,
bgColorOne=white,
headerColorOne=blue!70,
textborder=rounded
}
Ensure readability and visual appeal:
Font Selection :
% Sans-serif fonts recommended for posters
\usepackage{helvet} % Helvetica
\usepackage{avant} % Avant Garde
\usepackage{sfmath} % Sans-serif math fonts
% Set default to sans-serif
\renewcommand{\familydefault}{\sfdefault}
Text Sizing :
% Adjust text sizes for visibility
\setbeamerfont{title}{size=\VeryHuge}
\setbeamerfont{author}{size=\Large}
\setbeamerfont{institute}{size=\normalsize}
Emphasis and Highlighting :
\textbf{important}\textcolor{blue}{highlight}Enhance poster interactivity for modern conferences:
QR Code Integration :
\usepackage{qrcode}
% Link to paper, code repository, or supplementary materials
\qrcode[height=2cm]{https://github.com/username/project}
% QR code with caption
\begin{center}
\qrcode[height=3cm]{https://doi.org/10.1234/paper}\\
\small Scan for full paper
\end{center}
Digital Enhancements :
Generate high-quality PDF output for printing or digital display:
Compilation Commands :
# Basic compilation
pdflatex poster.tex
# With bibliography
pdflatex poster.tex
bibtex poster
pdflatex poster.tex
pdflatex poster.tex
# For beamer-based posters
lualatex poster.tex # Better font support
xelatex poster.tex # Unicode and modern fonts
Ensuring Full Page Coverage :
Posters should use the entire page without excessive margins. Configure packages correctly:
beamerposter - Full Page Setup :
\documentclass[final,t]{beamer}
\usepackage[size=a0,scale=1.4,orientation=portrait]{beamerposter}
% Remove default beamer margins
\setbeamersize{text margin left=0mm, text margin right=0mm}
% Use geometry for precise control
\usepackage[margin=10mm]{geometry} % 10mm margins all around
% Remove navigation symbols
\setbeamertemplate{navigation symbols}{}
% Remove footline and headline if not needed
\setbeamertemplate{footline}{}
\setbeamertemplate{headline}{}
tikzposter - Full Page Setup :
\documentclass[
25pt, % Font scaling
a0paper, % Paper size
portrait, % Orientation
margin=10mm, % Outer margins (minimal)
innermargin=15mm, % Space inside blocks
blockverticalspace=15mm, % Space between blocks
colspace=15mm, % Space between columns
subcolspace=8mm % Space between subcolumns
]{tikzposter}
% This ensures content fills the page
baposter - Full Page Setup :
\documentclass[a0paper,portrait,fontscale=0.285]{baposter}
\begin{poster}{
grid=false,
columns=3,
colspacing=1.5em, % Space between columns
eyecatcher=true,
background=plain,
bgColorOne=white,
borderColor=blue!50,
headerheight=0.12\textheight, % 12% for header
textborder=roundedleft,
headerborder=closed,
boxheaderheight=2em % Consistent box header heights
}
% Content here
\end{poster}
Common Issues and Fixes :
Problem : Large white margins around poster
% Fix for beamerposter
\setbeamersize{text margin left=5mm, text margin right=5mm}
% Fix for tikzposter
\documentclass[..., margin=5mm, innermargin=10mm]{tikzposter}
% Fix for baposter - adjust in document class
\documentclass[a0paper, margin=5mm]{baposter}
Problem : Content doesn't fill vertical space
% Use \vfill between sections to distribute space
\block{Introduction}{...}
\vfill
\block{Methods}{...}
\vfill
\block{Results}{...}
% Or manually adjust block spacing
\vspace{1cm} % Add space between specific blocks
Problem : Poster extends beyond page boundaries
% Check total width calculation
% For 3 columns with spacing:
% Total = 3×columnwidth + 2×colspace + 2×margins
% Ensure this equals \paperwidth
% Debug by adding visible page boundary
\usepackage{eso-pic}
\AddToShipoutPictureBG{
\AtPageLowerLeft{
\put(0,0){\framebox(\LenToUnit{\paperwidth},\LenToUnit{\paperheight}){}}
}
}
Print Preparation :
Digital Display :
CRITICAL : Always review the generated PDF before printing or presenting. Use this systematic checklist:
Step 1: Page Size Verification
# Check PDF dimensions (should match poster size exactly)
pdfinfo poster.pdf | grep "Page size"
# Expected outputs:
# A0: 2384 x 3370 points (841 x 1189 mm)
# 36x48": 2592 x 3456 points
# A1: 1684 x 2384 points (594 x 841 mm)
Step 2: OVERFLOW CHECK (CRITICAL) - DO THIS IMMEDIATELY AFTER COMPILATION
⚠️ THIS IS THE #1 CAUSE OF POSTER FAILURES. Check BEFORE proceeding.
Step 2a: Check LaTeX Log File
# Check for overflow warnings (these are ERRORS, not suggestions)
grep -i "overfull\|underfull\|badbox" poster.log
# ANY "Overfull" warning = content is cut off or extending beyond boundaries
# FIX ALL OF THESE before proceeding
Common overflow warnings and what they mean:
Overfull \hbox (15.2pt too wide) → Text or graphic is 15.2pt wider than columnOverfull \vbox (23.5pt too high) → Content is 23.5pt taller than available spaceBadbox → LaTeX struggling to fit content within boundariesStep 2b: Visual Edge Inspection (100% zoom in PDF viewer)
Check ALL FOUR EDGES systematically:
TOP EDGE:
BOTTOM EDGE:
LEFT EDGE:
RIGHT EDGE:
BETWEEN COLUMNS:
If ANY check fails, you have overflow. FIX IMMEDIATELY before continuing:
Fix hierarchy (try in order):
Check AI-generated graphics first:
Reduce sections:
Cut text content:
Adjust figure sizing:
width=\linewidth? → Change to width=0.85\linewidthwidth=1.0\columnwidth? → Change to width=0.9\columnwidthIncrease margins (last resort):
DO NOT proceed to Step 3 if ANY overflow exists.
Step 3: Visual Inspection Checklist
Open PDF at 100% zoom and check:
Layout and Spacing :
Typography :
Visual Elements :
Content Completeness :
Technical Quality :
Step 4: Reduced-Scale Print Test
Essential Pre-Printing Test :
# Create reduced-size test print (25% of final size)
# This simulates viewing full poster from ~8-10 feet
# For A0 poster, print on A4 paper (24.7% scale)
# For 36x48" poster, print on letter paper (~25% scale)
Print Test Checklist :
Step 5: Digital Quality Checks
Font Embedding Verification :
# Check that all fonts are embedded (required for printing)
pdffonts poster.pdf
# All fonts should show "yes" in "emb" column
# If any show "no", recompile with:
pdflatex -dEmbedAllFonts=true poster.tex
Image Resolution Check :
# Extract image information
pdfimages -list poster.pdf
# Check that all images are at least 300 DPI
# Formula: DPI = pixels / (inches in poster)
# For A0 width (33.1"): 300 DPI = 9930 pixels minimum
File Size Optimization :
# For email/web, compress if needed (>50MB)
gs -sDEVICE=pdfwrite -dCompatibilityLevel=1.4 \
-dPDFSETTINGS=/printer -dNOPAUSE -dQUIET -dBATCH \
-sOutputFile=poster_compressed.pdf poster.pdf
# For printing, keep original (no compression)
Step 6: Accessibility Check
Color Contrast Verification :
Color Blindness Simulation :
Step 7: Content Proofreading
Systematic Review :
Peer Review :
Step 8: Technical Validation
LaTeX Compilation Log Review :
# Check for warnings in .log file
grep -i "warning\|error\|overfull\|underfull" poster.log
# Common issues to fix:
# - Overfull hbox: Text extending beyond margins
# - Underfull hbox: Excessive spacing
# - Missing references: Citations not resolved
# - Missing figures: Image files not found
Fix Common Warnings :
% Overfull hbox (text too wide)
\usepackage{microtype} % Better spacing
\sloppy % Allow slightly looser spacing
\hyphenation{long-word} % Manual hyphenation
% Missing fonts
\usepackage[T1]{fontenc} % Better font encoding
% Image not found
% Ensure paths are correct and files exist
\graphicspath{{./figures/}{./images/}}
Step 9: Final Pre-Print Checklist
Before Sending to Printer :
Printing Specifications to Confirm :
Digital Presentation Checklist :
Review Script (Available in scripts/review_poster.sh):
#!/bin/bash
# Automated poster PDF review script
echo "Poster PDF Quality Check"
echo "======================="
# Check file exists
if [ ! -f "$1" ]; then
echo "Error: File not found"
exit 1
fi
echo "File: $1"
echo ""
# Check page size
echo "1. Page Dimensions:"
pdfinfo "$1" | grep "Page size"
echo ""
# Check fonts
echo "2. Font Embedding:"
pdffonts "$1" | head -20
echo ""
# Check file size
echo "3. File Size:"
ls -lh "$1" | awk '{print $5}'
echo ""
# Count pages (should be 1 for poster)
echo "4. Page Count:"
pdfinfo "$1" | grep "Pages"
echo ""
echo "Manual checks required:"
echo "- Visual inspection at 100% zoom"
echo "- Reduced-scale print test (25%)"
echo "- Color contrast verification"
echo "- Proofreading for typos"
Common PDF Issues and Solutions :
| Issue | Cause | Solution |
|---|---|---|
| Large white margins | Incorrect margin settings | Reduce margin in documentclass |
| Content cut off | Exceeds page boundaries | Check total width/height calculations |
| Blurry images | Low resolution (<300 DPI) | Replace with higher resolution images |
| Missing fonts | Fonts not embedded | Compile with -dEmbedAllFonts=true |
| Wrong page size | Incorrect paper size setting | Verify documentclass paper size |
| Colors look wrong | RGB vs CMYK mismatch | Convert color space for print |
| File too large (>50MB) | Uncompressed images | Optimize images or compress PDF |
| QR codes don't work | Too small or low resolution | Minimum 2×2cm, high contrast |
Effective content organization for different research types:
Experimental Research Poster :
Computational/Modeling Poster :
Review/Survey Poster :
Design posters that are accessible to diverse audiences:
Color Blindness Considerations :
Visual Impairment Accommodations :
Language and Content :
Guidance beyond LaTeX for effective poster sessions:
Content Strategy :
Physical Presentation Tips :
Digital Backups :
Determine poster requirements :
Develop content outline :
Choose LaTeX package :
CRITICAL: Generate SIMPLE figures with MINIMAL content. Each graphic = ONE message.
Content limits:
Create figures directory :
mkdir -p figures
Generate SIMPLE visual elements :
python scripts/generate_schematic.py "POSTER FORMAT for A0. SIMPLE visual with ONLY 3 elements: [icon1] [icon2] [icon3]. ONE word labels (80pt+). 50% white space. Readable from 8 feet." -o figures/intro.png
# Methods - ONLY 4 steps maximum
python scripts/generate_schematic.py "POSTER FORMAT for A0. SIMPLE flowchart with ONLY 4 boxes: STEP1 → STEP2 → STEP3 → STEP4. GIANT labels (100pt+). 50% white space. NO sub-steps." -o figures/methods.png
# Results - ONLY 3 bars/comparisons
python scripts/generate_schematic.py "POSTER FORMAT for A0. SIMPLE chart with ONLY 3 bars. GIANT percentages ON bars (120pt+). NO axis, NO legend. 50% white space." -o figures/results.png
# Conclusions - EXACTLY 3 items with GIANT numbers
python scripts/generate_schematic.py "POSTER FORMAT for A0. EXACTLY 3 key findings: '[NUMBER]' (150pt) '[LABEL]' (60pt) for each. 50% white space. NO other text." -o figures/conclusions.png
3. Review generated figures - check for overflow:
* **View at 25% zoom** : All text still readable?
* **Count elements** : More than 5? → Regenerate simpler
* **Check white space** : Less than 40%? → Add "60% white space" to prompt
* **Font too small?** : Add "EVEN LARGER" or increase pt sizes
* **Still overflowing?** : Reduce to 3 elements instead of 4-5
Select or create template :
assets/Design layout structure :
Set typography :
Create poster header :
Integrate AI-generated figures :
\includegraphics with proper sizingAdd minimal supporting text :
Add supplementary elements :
Review and iterate :
Test readability :
Optimize for printing :
Compile final PDF :
pdflatex poster.tex # Or for better font support: lualatex poster.tex
Verify output quality :
Prepare for printing :
Create supplementary materials :
This skill works effectively with:
Recommended workflow : Always use scientific-schematics and generate-image skills BEFORE creating the LaTeX poster to generate all visual elements.
AI-Generated Graphics Mistakes (MOST COMMON):
Fixing Overflow in AI Graphics: If your AI-generated graphics are overflowing or have small text:
Design Mistakes :
Content Mistakes :
Technical Mistakes :
Best Practices :
Ensure required LaTeX packages are installed:
# For TeX Live (Linux/Mac)
tlmgr install beamerposter tikzposter baposter
# For MiKTeX (Windows)
# Packages typically auto-install on first use
# Additional recommended packages
tlmgr install qrcode graphics xcolor tcolorbox subcaption
Helper scripts available in scripts/ directory:
compile_poster.sh: Automated compilation with error handlinggenerate_template.py: Interactive template generatorresize_images.py: Batch image optimization for postersposter_checklist.py: Pre-submission validation toolComprehensive reference files for detailed guidance:
references/latex_poster_packages.md: Detailed comparison of beamerposter, tikzposter, and baposter with examplesreferences/poster_layout_design.md: Layout principles, grid systems, and visual flowreferences/poster_design_principles.md: Typography, color theory, visual hierarchy, and accessibilityreferences/poster_content_guide.md: Content organization, writing style, and section-specific guidanceReady-to-use poster templates in assets/ directory:
Load these templates and customize for your specific research and conference requirements.
Weekly Installs
63
Repository
GitHub Stars
17.3K
First Seen
Jan 20, 2026
Security Audits
Gen Agent Trust HubPassSocketPassSnykPass
Installed on
opencode56
gemini-cli54
codex54
cursor52
claude-code51
github-copilot50
MCP图像生成技能:使用Gemini AI为营销、UI设计、演示文稿创建高质量图像
7,600 周安装
| Timeline | 3-4 points MAX | 8 words | Year-by-year detail | "2020 START" "2022 TRIAL" "2024 APPROVED" (6 words) |
| Timeline | 3 points | 3 words | "POSTER FORMAT for A0: THREE years only: '2018' '2021' '2024' (150pt each). Large icons. 60% white space. NO details." |
\documentclass[25pt, a0paper, portrait, margin=25mm]{tikzposter}