LighterpixCompress an image

In-browser image optimizer

Compress a PNG and keep the transparency

PNGs are lossless, which is why screenshots and interface graphics look crisp — and why they can be ten times larger than they need to be. Lighterpix re-encodes them with OxiPNG, optionally reduces the palette, and lets you compare the result against WebP and AVIF in the same view.

Choose an imageTry a sample

Compress a PNG in three steps

  1. Open the PNGDrop it on the page or paste it from the clipboard — screenshots go straight in without touching disk first.
  2. Try lossless first, then palette reductionOxiPNG squeezes the file with no quality loss at all. If it is still too big, enable Reduce palette for flat graphics like icons and diagrams.
  3. Check the alpha channelThe preview shows the transparent areas. If they survive, download — and compare against WebP, which supports transparency too and is often smaller.

Lossless, or visually lossless?

Lossless PNG optimisation rewrites the same pixels more cleverly: better filters, tighter deflate streams, smaller chunks. Palette reduction, by contrast, changes the pixels — it stores each colour once, which is perfect for flat art and terrible for photographs. Lighterpix gives you both, and shows the difference before you commit.

When WebP beats PNG

For screenshots and illustrations, WebP in lossless mode is usually 20–35% smaller than an optimised PNG, and it keeps transparency. If your audience is on modern browsers, that is a free win.

What to expect

TechniqueQualityTypical savingBest for
OxiPNG (lossless)Identical pixels10–40%Everything
Reduce paletteFewer colours50–80%Icons, UI, flat art
WebP losslessIdentical pixels20–35% vs PNGScreenshots
AVIF (lossy)Near-identical60–90%Illustrations on modern sites

Questions

Does compressing a PNG remove transparency?

No. OxiPNG and WebP both keep the alpha channel. The preview pane renders transparency as a checkerboard so you can confirm it before downloading.

Is palette reduction safe for screenshots?

For screenshots of interfaces it usually is — they contain few distinct colours. For photos it produces banding, so keep it off there.

Why is my optimised PNG sometimes bigger?

If the source was already optimised, re-encoding can add a few bytes. The size figure next to each pane is the real output size, so check before saving.