Workflow

Ten image edits worth doing in a browser (and three that are not)

Browsers handle geometry and encoding — resize, crop, rotate, convert, compress, watermark — as well as any desktop application. They are a poor fit for detailed retouching, colour-managed print work and layered composition, which need pixel-level control and a colour profile pipeline.

Hands at a trackpad beside a tablet showing the same photograph cropped several ways, with printed contact sheets alongside.

Published Updated 8 min readEditorial policy

The edits a browser does properly

These are deterministic operations. There is no quality difference between doing them in a browser and doing them in expensive software, because the maths is the same and the encoder is the same class of encoder.

  • Resizing to exact pixel dimensions.
  • Cropping to a composition or a fixed ratio.
  • Rotating by 90° steps, which reorders pixels and loses nothing.
  • Flipping and mirroring.
  • Converting between JPG, PNG and WebP.
  • Compressing to hit a size budget.
  • Converting to greyscale with luminance weighting.
  • Applying a uniform blur.
  • Adding a text watermark.
  • Basic brightness, contrast and saturation correction.

The order matters more than the tools

Each lossy save discards a little more. Doing four operations as four separate download-and-reupload steps means four encodes. Sequence the work so the destructive step happens once, at the end.

A reliable order: crop first to decide the composition, then resize to the final pixel dimensions, then apply tonal or effect changes, and compress or convert last.

Three things to do elsewhere

Detailed retouching — removing objects, healing skin, masking hair — needs layers, brushes and history that a single-purpose web tool does not have.

Colour-managed print work needs CMYK conversion and an ICC profile pipeline. Browsers are sRGB-centric and will quietly discard profiles.

Batch work across hundreds of files with different rules is faster with a command-line tool such as ImageMagick, or a desktop batch processor.

A worked example: preparing a product photo for a shop listing

Start from the camera original, not from a copy already sent through a messaging app. Crop square so the product fills the frame with a small margin. Resize to 1600×1600, which is enough for zoom on most storefronts. Correct brightness if the shot is flat. Convert to WebP at quality 80, or JPG at 85 if the platform rejects WebP. The result is typically 120–250 KB and indistinguishable from the original at display size.

Spotted something inaccurate? Send a correction — technical corrections are applied and the updated date changed.