Screenshot Stitcher

Joins scrolling screenshots from Uma Musume into one tall image, so you can share a full spark, skill, or inheritance list in a single picture. This version is adapted for the English release of the game.

Live site: https://sandwaves.github.io/Uma-Stitcher/

Everything runs in your browser. Nothing is uploaded, there's no server, and there are no analytics.

How to use

1. Add your screenshots with Choose screenshots, by dragging them onto the page, or by pasting with Ctrl+V. Click a thumbnail to see it full-size, or its × to remove it.
2. Click Stitch screenshots. The stitched image appears below after a few seconds.
3. Save it with Save PNG or Save JPG, or use Copy image.

Options

• Include the top section (name, stats and tabs) keeps the fixed top part of the screen above the stitched list. Untick it to get only the list.
• Add skill icons to sparks (beta) reads skill names with OCR and adds the matching icon next to each one. Names are fuzzy-matched against the skill dictionary, so small OCR mistakes still get the right icon.
• The theme toggle in the top-right switches between light and dark, and remembers your choice.

Getting good results

• Each screenshot should overlap the previous one by at least two rows of content.
• Take every screenshot the same way, with the same device, orientation, and window size.
• On PC/Steam, capture only the game window with Alt+Print Screen, and don't crop the images. The tool needs the panel to be in the same place in every frame.
• Keep mouse cursors, notifications, and overlays out of the shots.
• Screenshots are matched by where they overlap, so their order usually doesn't matter. Any that can't be matched are stacked in the numbered thumbnail order; use the ‹ › buttons (or drag on desktop) to change it.

Running locally

There's no build step. Serve the project folder over HTTP:

python -m http.server 8000

Then open http://localhost:8000. Opening index.html directly works in most browsers, but OpenCV's wasm, clipboard paste, and the Tesseract worker are more reliable over HTTP.

After editing, this catches syntax errors:

node --check js/base.js js/receipt_factor.js

Project layout

.
├── index.html             # main page
├── css/
│   ├── base.css           # UI styles and light/dark themes
│   ├── common.css         # small shared rules (image, .hidden)
│   └── reset.css          # Eric Meyer reset
├── js/
│   ├── base.js            # UI wiring, drag/drop, lightbox, theme toggle, generatePhoto
│   ├── receipt_factor.js  # the stitching pipeline (OpenCV.js)
│   ├── dict_skills.js     # English skill name → icon ID dictionary
│   ├── regexps.js         # OCR text cleanup rules
│   ├── race_factors.js    # JP race names (reference only, not used by the app)
│   ├── shortcut.js        # global keyboard shortcut helper
│   └── opencv.js          # bundled OpenCV.js
└── img/
    ├── tmpl_*.png         # JP UI templates for layout detection
    ├── skill_icons/       # icons for the skill-icon option
    └── ...

What's changed

This is the third project in a chain of forks (see Credits). The page's Update history section has the full versioned list; Ver4.01 onwards covers this line of forks.

In this fork (compared with daftuyda/Stitcher)

• OCR uses the Tesseract English model instead of Japanese.
• dict_skills.js is keyed by English skill names, and the missing English skill icons have been added.
• regexps.js now does light English cleanup: whitespace, quotes, ○ / ◎ / × marker fixes, and common OCR confusions.
• Skill names snap to the closest dictionary entry by Levenshtein distance. The aptitude marker must still match exactly.
• A skill with no icon shows the "unknown" icon instead of throwing an error.
• The upstream Google Analytics tag is removed, and the site metadata and links point to this repo.

In daftuyda/Stitcher (compared with the original)

• All UI text, error messages, and instructions are in English. The race data and UI templates still use Japanese matching keys.
• A new, cleaner UI with a light/dark theme.
• Reset happens in place, without reloading the page, so OpenCV, Tesseract, and the icons stay cached.
• Drag-to-reorder thumbnails and a full-size lightbox preview.
• Memory fixes: several Mat lifecycle leaks closed, a use-after-free fixed, and double-frees removed.
• Speed fixes: bulk pixel reads via mat.data instead of per-pixel ucharAt loops, cached template Mats, and a cached Tesseract worker so the OCR model loads once per session.
• Stricter PC/Steam matching: a tighter panel crop, less noise-sensitive diff masks, stricter overlap checks, and a minimum overlap size that rejects false header-only matches.
• Screenshots that can't be matched keep your order instead of being moved to the end.
• Overlapping content is drawn once; each image only paints the part below what's already on the canvas.

Credits

Nearly all of the stitching pipeline and UI come from these two projects:

1. lt900ed/receipt_factor by lt900ed: the original tool, for the Japanese version of the game.
2. daftuyda/Stitcher: the English UI, rewritten interface, robustness fixes, and PC/Steam support.

Also used:

• Skill icons and data from U-tools.
• OpenCV.js and Tesseract.js, both bundled.
• The loading spinner from SpinKit (sk-fading-circle).

License

The upstream repo has no license file, so neither does this one. Treat it as source-available for personal use.