nogataka PDF to HTML Slide Converter
Convert PDF presentations to high-fidelity HTML slides by visually reproducing each slide from screenshots.
Pipeline
PDF → (pdftoppm) → Slide images
↓
Claude reads each image + writes HTML
↓
001.html, 002.html, ... + print.html
Dependencies
brew install poppler
Workflow
Phase 1: Generate Slide Images
python ~/.claude/skills/pdf-to-html/scripts/pdf_to_images.py input.pdf output_dir
This produces output_dir/slide-01.jpg, slide-02.jpg, etc.
Phase 2: Analyze the Deck
Before writing any HTML:
- Read ALL slide images to understand the full deck
- Identify the color palette (3-4 colors: primary dark, accent, secondary)
- Identify the font style (serif/sans-serif, weight patterns)
- Note the header/footer pattern used across content slides
- Note the overall style (creative, elegant, modern, professional, minimalist)
Phase 3: Load Design System
Read these files to load the HTML design system rules:
~/.claude/skills/slidekit-create/SKILL.md— Mandatory constraints, PPTX conversion rules, anti-patterns, design guidelines, HTML boilerplate, pre-delivery checklist~/.claude/skills/slidekit-create/references/patterns.md— 15 layout patterns with DOM trees, component snippets
All rules in slidekit-create apply. Key constraints:
- Slide size: 1280x720px
- Tailwind CSS 2.2.19 + Font Awesome 6.4.0 + Google Fonts via CDN
- No JavaScript
- Root DOM:
<body>→ single wrapper<div> - Text in
<p>/<h*>(not<div>) - No text in
::before/::after
Phase 4: Write HTML Slides
For each slide image, one at a time:
- Read the slide image carefully
- Pick the closest layout pattern from slidekit-create's 15 patterns
- Write the HTML that visually reproduces the slide:
- Match colors precisely (use hex values observed in the image)
- Match text content exactly (all text, numbers, labels)
- Match spatial layout (element positions, proportions)
- Use the consistent header/footer across content slides
- Use Font Awesome icons where the original has icons or bullets
- Save as
{output_dir}/{NNN}.html(zero-padded: 001.html, 002.html, ...)
HTML Boilerplate
<!DOCTYPE html>
<html lang="ja">
<head>
<meta charset="utf-8" />
<meta content="width=device-width, initial-scale=1.0" name="viewport" />
<title>{Slide Title}</title>
<link href="https://cdn.jsdelivr.net/npm/tailwindcss@2.2.19/dist/tailwind.min.css" rel="stylesheet" />
<link href="https://cdn.jsdelivr.net/npm/@fortawesome/fontawesome-free@6.4.0/css/all.min.css" rel="stylesheet" />
<link href="https://fonts.googleapis.com/css2?family={PrimaryFont}:wght@300;400;500;700;900&family={AccentFont}:wght@400;600;700&display=swap" rel="stylesheet" />
<style>
body { margin: 0; padding: 0; font-family: '{PrimaryFont}', sans-serif; overflow: hidden; }
.font-accent { font-family: '{AccentFont}', sans-serif; }
.slide { width: 1280px; height: 720px; position: relative; overflow: hidden; }
/* Custom color classes: .bg-brand-dark, .bg-brand-accent, .bg-brand-sub, etc. */
</style>
</head>
<body>
<div class="slide">
<!-- Content -->
</div>
</body>
</html>
Phase 5: Generate print.html
After all slides are written, generate {output_dir}/print.html:
<!DOCTYPE html>
<html lang="ja">
<head>
<meta charset="utf-8" />
<title>View for Print</title>
<style>
* { margin: 0; padding: 0; box-sizing: border-box; }
body { background: #FFFFFF; }
.slide-frame {
width: 1280px; height: 720px;
margin: 20px auto;
box-shadow: 0 2px 10px rgba(0,0,0,0.15);
border: 1px solid #e2e8f0;
overflow: hidden;
}
.slide-frame iframe { width: 1280px; height: 720px; border: none; }
@media print {
body { background: #FFFFFF; }
.slide-frame {
page-break-after: always; box-shadow: none; border: none;
margin: 0 auto;
transform: scale(0.85); transform-origin: top center;
}
}
</style>
</head>
<body>
<!-- One iframe per slide -->
<div class="slide-frame"><iframe src="001.html"></iframe></div>
...
</body>
</html>
Phase 6: Visual QA
Compare each generated HTML against the original screenshot.
For each slide:
- Read the original slide image (
slide-NN.jpg) - Read the generated HTML file
- Check for:
- Missing or incorrect text
- Wrong colors or color contrast issues
- Layout misalignment (elements shifted, wrong proportions)
- Missing decorative elements (bars, shapes, badges)
- Overflow or clipping issues
- Fix any issues found
- Re-verify after fixes
Do not declare success until at least one full comparison pass reveals no issues.