PicToWebP bulk-converts your chunky images into lightweight WebP beauties, from a CLI or your browser. In many collections it cuts file sizes by 80%+ with no visible quality loss.
Images are like cats on the internet: everywhere. But unlike cats, heavy images slow down your site, eat storage, and make users wait. WebP files are dramatically smaller than JPEG/PNG at equivalent quality, which means:
- Smaller size, same quality: WebP beats JPEG/PNG on compression.
- Faster websites: fewer bytes, zippier loads.
- Lower storage costs: thousands of saved images add up to real money. 💰
- Happier planet: less server power. 🌍
PicToWebP is private by architecture, not by policy:
- Nothing ever leaves your machine. The browser edition has no server at all; the Python web UI runs on
127.0.0.1only. There is nothing to upload to. - No tracking. No analytics, no cookies, no telemetry, no accounts.
- EXIF/GPS stripped by default: camera info and location tags are removed from every converted file (the browser edition can't preserve them at all).
No install? No problem, use it right now:
https://xqbuilds.com/pictowebp
- Bulk convert: pick a folder and every image inside (including subfolders) is converted, with the directory structure preserved.
- Single-image convert: drop one image, get its WebP instantly, with a draggable before/after comparison slider.
- Shareable stats card: after a conversion, generate an image of your savings for social sharing.
- Detailed failure report: anything that couldn't be converted is listed on screen and saved to
conversion-errors.txt. - Metadata handling: strip EXIF (the default, best for the web) or keep it, per run.
- Graceful cancellation: stop mid-run and keep everything already converted.
And tons of optimizations, edge cases handled, and performance tweaks to give the best experience to you!
The fastest way in: open https://xqbuilds.com/pictowebp and you're converting.
- Bulk: click to select a folder (or drag one in) and the whole tree is converted with structure preserved. Save results back to any folder or download a ZIP.
- Single: drop an image, paste one with
Ctrl+V, or click to browse. You get an instant 1:1 preview with a before/after comparison slider. - Nothing to install, nothing to sign up for.
Prefer to run it yourself? 🏠
cd web-ts
npm install
npm run build # the exact GitHub Pages artifact (browser backend → dist/)
npm run preview # serve it at http://localhost:4173/PicToWebP/(The /PicToWebP/ subpath is intentional: the static build is rooted there to
match the Pages URL.) For live reload while hacking, use npm run dev and open
http://localhost:3000/PicToWebP/.
Honest limitations of doing everything in the browser:
- Metadata (EXIF/GPS) is always stripped: canvas decoding cannot preserve it. For keep-metadata workflows, use the Python or Rust CLI.
- There is no separate lossless mode; use the quality slider at 100 for the best the browser encoder offers.
- Requires a Chromium-based browser (Chrome/Edge) for folder access; single-image mode works anywhere WebP encoding is supported.
Requires Python 3.10+. Using uv (recommended):
uv sync
uv run pictowebp path/to/images -q 85Or with pip (CLI only; install .[web] if you also want the local web UI):
pip install .
pictowebp path/to/images -q 85Run without arguments for an interactive session (it will prompt for the folder,
quality and thread count; just press Enter to accept the defaults). Every run
creates a fresh images_webp_<timestamp> folder next to the source, mirroring
its structure. Quality defaults to 80; add --lossless to keep every pixel.
Prefer converting in your browser but with Python's format support? Serve the local web UI (requires the web extra):
pip install .[web]
pictowebp-web # serves http://127.0.0.1:8000Prefer containers? 🐳
docker build -t pictowebp .
docker run --rm -p 8000:8000 pictowebpRequires Rust 1.85+ (edition 2024):
cargo install --path src_rust
pictowebp # then follow the promptsThe Rust CLI exposes the same flags and behaviour as the Python one (-q, -t,
--lossless, --keep-metadata, ...). Run
pictowebp --help for the full list.
In one real-world run it delivered an 85.79% reduction across 2,598 images
in 46 seconds. Failures are grouped by reason, listed on screen, and saved to
conversion-errors.txt so nothing fails silently.
If you're tired of images taking up more space than they should, put them on the PicToWebP diet: they lose the weight but keep the charm. Feedback and PRs are always welcome, drop by the Issues section! 💌
Want to hack on PicToWebP? See CONTRIBUTING.md. Curious about what the test suite covers and how to run it? See TESTING.md. Wonder how the Python and Rust engines stack up? Read the benchmark.
P.S. No cats were harmed in the making of this tool. They were too busy ruling the internet. 🐈⬛
MIT © 2026 aditya-xq