Where quality is actually lost
Four places, and this preset handles each:
- Aliasing on reduction — prevented by widening the filter to the full footprint of each output pixel.
- Darkening and colour shifts — prevented by resampling in linear light instead of sRGB values.
- Dark or light fringes around transparent edges — prevented by weighting colour by alpha (premultiplication).
- Re-compression — avoided by saving PNG; if you need JPEG or WebP, quality 90+ keeps losses invisible.
What cannot be promised
Enlarging cannot conjure detail the original never had. A classical filter keeps the enlargement clean; the AI engine adds reconstructed detail that looks right but is the model’s best guess. For documents or evidence, prefer classical methods.
Questions
Is PNG always lossless?
Yes. PNG stores every pixel exactly. Files are larger than JPEG for photos; use JPEG or WebP at high quality when size matters.
Does scaling down then up lose quality?
Yes — each reduction discards pixels for good. Always scale from the original, never from a previously scaled copy.
What does “linear light” mean, and why is it on by default?
Pixel values in a JPEG or PNG are gamma-encoded: value 128 is not half as bright as 255, it is about a fifth. Averaging those numbers directly — what most resizers do — darkens fine bright detail and muddies colour edges. With linear light on, every pixel is converted to physical light before it is weighted and converted back afterwards. The test chart’s one-pixel checkerboard shows the difference: reduced in linear light it matches the light grey patch beside it; reduced the naive way it matches the dark one.
Is my picture uploaded anywhere?
No. Decoding, scaling and encoding all run in this browser tab, and the scaling engine runs in a Web Worker with no network code in it. The only download the site can make for you is an AI model file, and only after you press the button that names its size — that request contains no picture.