obrobka

In-browser image processing

Runs in your browser

Batch processing

Pick more than one file and the widget switches to batch mode. The settings apply to every file and the finished ones are collected into a ZIP. Measured on production: twenty 4032×3024 shots carrying an orientation tag went through in 3.4 s, with a 150 MB JavaScript heap.

drop it here or paste with Ctrl+V · the file never leaves your device

How to fit

Fits and returns the smaller size — no padding

Aspect ratio
Framing
Upscaling
Brush
Layers

Put an image on top — a logo, a sticker, a watermark.

How to do it

  1. Pick several files at once, or drag them into the window
  2. The settings below apply to the whole batch — format, size, background
  3. Take the ZIP. Change a setting and a “Process again” button appears

Frequently asked questions

How many files will it take?
Twenty phone shots is routine: 3.4 s for the whole batch. The limit is not the count but the tab’s memory — every result is held there until the archive is built. The total weight is shown on the ZIP button.
Why is background removal slower?
Because with a model the files go one at a time. Measured: four concurrent sessions eat 1.2 GB and buy only 1.8× — ONNX Runtime already parallelises inside a session, and a second layer of parallelism just takes cores away from it. Without a model, four workers run and it is three times faster.
What happens to a broken file?
It gets marked “failed” and does not stop the rest. One corrupt photo in a batch of twenty is an ordinary thing.
Is the archive compressed?
Depends on the format. For PNG compression buys 15% — our encoder works fast and leaves redundancy behind. For WebP, JPEG and AVIF it buys 0.1% for ten times the CPU, so those are stored as they are.