Scrolling Screenshot OCR on Windows: Free Alternative vs CaptureOCR
Long pages — chat logs, documentation, web articles, terminal output — don’t fit in one screenshot. You need scrolling capture + OCR.
Here’s how the options stack up.
The Scrolling Capture Landscape
| Tool | Scrolling Capture | OCR | Export | Cost |
|---|---|---|---|---|
| ShareX | ✅ Excellent | ❌ None | Image only | Free |
| PicPick | ✅ Good | ❌ None | Image only | Free/Paid |
| Snagit | ✅ Excellent | ✅ Built-in | Text/Image | $62+ |
| Windows Snipping Tool | ❌ No scrolling | ✅ Text Actions | Text only | Free |
| PowerToys Text Extractor | ❌ No scrolling | ✅ | Text only | Free |
| CaptureOCR | ✅ Best-effort v1.0 | ✅ PP-OCRv6 | Markdown/CSV/JSON | $2.99 once |
CaptureOCR’s Scrolling Approach (Technical Details)
Unlike template matching (which fails when neither frame contains the other), CaptureOCR uses row-signature stitching:
- Grayscale + crop right 8% (removes scrollbar noise)
- Compress each row into 32 column buckets (mean pixel value per bucket)
- Shift search: compare
prev[s:]vscur[:H-s]using mean absolute difference - Peak validation: best score must be < threshold AND clearly better than median
- Sticky-header handling: retry skipping 8%/16%/25% of top rows
- Adaptive wheel step: measure actual scroll distance, target ~45% of selection height
Why 32 buckets? Tested 12/24/32/64 on real frame pairs:
- 12 buckets: 7/9 correct
- 24 buckets: 8/9 correct
- 32 buckets: 9/9 correct
- 64 buckets: 9/9 correct (2× compute)
32 is the sweet spot.
CaptureOCR Scrolling Limits (Honest Assessment)
| Scenario | Works? | Notes |
|---|---|---|
| Standard web page (stable scroll) | ✅ | Best case |
| Terminal output (stable) | ✅ | Excellent |
| Chat app (Discord, Slack) | ⚠️ | Virtualized lists break it |
| Browser with lazy-loaded images | ❌ | Content not rendered yet |
| Sticky header/nav bar | ⚠️ | Handled via skip retries |
| Custom scroll animation | ❌ | Breaks frame alignment |
| Horizontal scroll | ❌ | Vertical only |
Hard caps: 999 frames, 512 MB canvas, minimum 20,000px height.
Free Workflow: ShareX + CaptureOCR
If you want free scrolling capture + OCR:
- ShareX (free, open source) for scrolling capture
- Configure: Capture → Scrolling capture → Vertical
- Output: PNG (high quality)
- CaptureOCR (paid, $2.99) for OCR
- Drag the long PNG into CaptureOCR (or capture from screen)
- Runs tiling OCR: 1600px tiles with 64px overlap
- Export Markdown/CSV
Why this combo?
- ShareX has the most robust scrolling capture engine (years of development)
- CaptureOCR has the best offline OCR for structured export
- Total cost: $2.99 vs Snagit’s $62+
CaptureOCR’s Built-in Scrolling (When It Works)
For stable scrollers (terminal, simple web pages, Notepad++), CaptureOCR’s built-in scrolling is convenient:
Ctrl+Shift+X→ holdShift→ drag to select scrollable area- Click “Start Scroll Capture” in the floating panel
- Watch the progress bar (shows frame count, canvas height)
- Stop when it reaches bottom or click “Stop”
- OCR runs automatically (tiling: 1600px tiles, 64px overlap)
- Export Markdown/CSV
Tiling OCR for long images: A 20,000px tall image would be crushed to 2px text if fed whole. CaptureOCR tiles it:
- Tile height: 1600px target
- Overlap: 64px (2 lines at 1080p)
- Core bands: contiguous, non-overlapping
- Line assignment: by center Y coordinate
- Remap to whole-image coordinates → sort by Y then X
Comparison: Built-in vs ShareX + CaptureOCR
| Aspect | CaptureOCR Built-in | ShareX + CaptureOCR |
|---|---|---|
| Convenience | One app | Two apps |
| Scrolling robustness | Best-effort v1.0 | Battle-tested |
| OCR integration | Seamless | Manual drag |
| Cost | $2.99 (included) | $2.99 + free |
| Virtualized lists | ❌ | ❌ (same limitation) |
Recommendation
| Your Need | Recommended Approach |
|---|---|
| Occasional long pages, stable scroll | CaptureOCR built-in (simpler) |
| Frequent scrolling, tricky apps | ShareX capture → CaptureOCR OCR |
| Chat logs, virtualized lists | Manual region capture (reliable fallback) |
| Budget $0 | ShareX (capture only) + Windows OCR (limited) |
The Honest Truth
Scrolling capture + OCR is hard. No tool handles every case. Virtualized lists, lazy loading, custom scroll animations — these break everyone.
CaptureOCR’s v1.0 scrolling is best-effort. Region capture is the reliable fallback. The app tells you this upfront.
If scrolling fails: capture regions manually. It’s slower but 100% reliable.
Get the Tool
CaptureOCR on Microsoft Store — $2.99 one-time. Includes both region and scrolling capture.
ShareX — Free, open source. Best-in-class scrolling capture.
Related reads: