Shrink photos to a file size you choose, resize them, or convert HEIC, PNG and JPG to WebP, AVIF or JPG. Work on several at once and compare the result with the original before you download.
JPEG stays JPEG, PNG stays PNG. HEIC becomes JPEG, because browsers cannot write HEIC.
Applies to JPEG, WebP and AVIF output. Left alone, each format uses its own default (75, or 50 for AVIF). Most photos look the same down to about 60.
Keeps the proportions and never enlarges: a smaller image stays its own size.
JPEG, PNG, WebP, GIF, AVIF, BMP or HEIC, up to 100 MB each. Your images stay on your device. The compressed copies carry no location or camera details.
Runs entirely in your browser. Nothing you type or upload is sent to a server.
How image compression works
A photo straight from a phone is millions of pixels, each with a red, green and blue value. Stored raw, a 12 megapixel picture would take about 36 MB. Image formats shrink that in one of two ways. Lossless formats such as PNG find patterns and store them more compactly, so every pixel comes back exactly; that works well for screenshots and logos with large flat areas, and poorly for photos, where neighbouring pixels are rarely identical. Lossy formats such as JPEG, WebP and AVIF throw away detail the eye is unlikely to miss, which is how a 36 MB photo becomes a 3 MB file that looks the same.
JPEG, the oldest of the three, shows the idea clearly. It first stores color at half the resolution of brightness, because the eye is far more sensitive to light and dark than to hue. It then cuts the picture into 8 × 8 pixel blocks and describes each block as a mix of simple wave patterns, from smooth gradients to fine stripes. The quality setting controls how coarsely those patterns are rounded off. At high quality the fine stripes survive; at low quality they are rounded to nothing, and the blocks begin to show as visible squares, with halos around sharp edges and bands in smooth skies. WebP and AVIF use more flexible block sizes and better prediction from neighbouring blocks, which is why they reach the same look in fewer bytes.
Which format to choose
| Format | Best for | Transparency | Size against JPEG |
|---|---|---|---|
| JPEG | Photos that must open anywhere: email, forms, printing | No | Baseline |
| WebP | Photos and graphics for websites | Yes | About 25 to 35 percent smaller |
| AVIF | The smallest files, when the receiver supports it | Yes | Often around half |
| PNG | Screenshots, logos, diagrams, anything with text | Yes | Larger for photos, smaller for flat graphics |
The encoders here are the ones Google’s Squoosh app uses: MozJPEG, which squeezes more out of JPEG than most cameras and editors do, libwebp, libavif with the AV1 codec, and OxiPNG, which rewrites a PNG as small as it can go without changing a pixel.
Dimensions matter more than quality
File size grows with the number of pixels, so resizing is the strongest lever there is. A phone photo of 4032 × 3024 pixels holds 12.2 million of them. Resized to 1920 pixels on its longest side it holds 2.8 million, less than a quarter, and the file shrinks by roughly the same factor before quality is touched at all. A typical laptop screen is under 2000 pixels wide and a phone screen is narrower, so for anything that will be looked at on a screen, 1600 to 2560 pixels on the long side is plenty. Keep the full size only for printing or for cropping later.
Hitting a target file size
Job portals, visa applications and school forms often set a hard limit such as 200 KB. In target mode the tool encodes the image at its everyday quality, checks the size, then narrows in on the highest quality that fits, usually in three to five attempts. If even the lowest quality is too big, or the only fit is a blurry one, it says so and offers to reduce the dimensions instead, which almost always looks better than a heavily compressed full-size image. Forms disagree about whether a kilobyte is 1,000 or 1,024 bytes, so the tool uses 1,000, the stricter reading.
What re-compressing changes
- Metadata is not carried over. The new file is built from pixels, so GPS location, camera details and dates are gone. To remove those without re-compressing, use the Photo Metadata Remover.
- Colors are converted to sRGB. Recent phones shoot in the wider Display P3 color space. The browser converts to standard sRGB when it draws the image, so the most saturated reds and greens can look very slightly flatter.
- Animation is lost. Animated GIF, WebP and PNG files keep only their first frame.
- Quality loss adds up. Each lossy save discards a little more. Compress from the original file, not from a copy that was already compressed.
HEIC, the iPhone’s default format, is efficient but awkward to share, because many websites and Windows apps cannot open it. Converting to JPEG fixes that for good. If your scans are already PDFs, the PDF Merge & Split tool combines them into one document, also in your browser. To embed a small image directly in HTML or CSS, compress it here first, then turn it into a data URI with the Base64 converter.
