Skip to content

Compress image

Make images smaller without visible loss.

Drop images here

JPG, PNG, WebP, AVIF, HEIC, SVG and GIF

How to use Compress image

  1. Add images. Drop one or more JPG, PNG, WebP, AVIF, HEIC, or SVG files onto the drop zone.
  2. Choose Quality and Largest side. Pick Balanced (0.75), Smaller file (0.6), or Best quality (0.88), and optionally cap Largest side down to 1280 px or 1920 px.
  3. Pick Format. Keep Same as the original for JPG or WebP, or switch PNG screenshots to WebP or JPG for dramatic size reductions.
  4. Compress and save. Select Compress image (or Compress N images) to download your file or batch ZIP.

Your files are processed on this device and never uploaded. They stay private, and there's nothing to delete afterwards.

How much smaller can a photo get without visible quality loss?

Phone photos and camera JPEGs commonly shrink by 60% to 90% at the default Balanced (0.75) preset, and even further when you cap Largest side to 1920 px or 1280 px to reduce image file size. As noted in the MDN HTMLCanvasElement.toBlob() Reference, toBlob() accepts a 0.0 to 1.0 quality argument for lossy image/jpeg and image/webp encoding when you compress photos online.

If you need exact custom pixel dimensions rather than a longest-side cap, you can scale images to exact pixel widths and heights before or after compressing.

Choosing Smaller file, Balanced, or Best quality presets

The Quality dropdown offers three tuned presets: Smaller file (0.6), Balanced (0.75), and Best quality (0.88). Balanced removes imperceptible high-frequency color noise while keeping textures, skin tones, and text edges clean on retina screens.

For JPG files, lossy encoders apply ITU-T T.81 DCT quantization matrices across 8×8 pixel blocks to discard high-frequency noise that human eyes barely notice.

Selecting Smaller file (0.6) pushes quantization slightly further for strict email or portal limits, while Best quality (0.88) preserves subtle gradients in product photography and portraits. Most users leave Balanced (0.75) active for everyday web uploads.

How the automatic 0.62 retry and no-bloat guard work

If a lossy export at 0.75 or 0.88 comes out larger than the input file (common with already-optimized web JPEGs), the compressor automatically retries once at 0.62. If the output at Keep size and Same as the original is still larger or equal, the no-bloat guard returns your original file bytes untouched.

Comparing WebP, JPG, and lossless PNG compression

Choosing the right output Format matters as much as the Quality slider in any image compression format comparison. Under IETF RFC 9649 WebP Image Format Specification, WebP combines VP8 lossy compression with optional alpha transparency, making it ideal for both photos and transparent UI screenshots.

Format setting Compression type Alpha transparency Quality slider active Best use case
Same as the original (JPG/WebP) Lossy WebP only Yes (0.6, 0.75, 0.88) Quick shrinking without changing extensions
Same as the original (PNG) Lossless Deflate Preserved (32-bit RGBA) Disabled (lossless) Pixel-exact icons and lossless archives
WebP (smallest for websites) Lossy VP8 + alpha Preserved Yes (0.6, 0.75, 0.88) Fast web pages and transparent screenshots
JPG (works everywhere) Lossy DCT Filled with #ffffff Yes (0.6, 0.75, 0.88) Email attachments and strict upload forms

When converting transparent PNGs to JPG, see-through areas are filled with solid white (#ffffff) because JPEG has no alpha channel. If you ever need to reverse a WebP back into an uncompressed graphic for desktop editing, use a lossless WebP to PNG converter.

Choosing WebP (smallest for websites) keeps alpha transparency intact while applying lossy VP8 compression to RGB channels, making it ideal for UI screenshots with drop shadows or product cutouts that would otherwise balloon in lossless PNG.

Halving downscale steps with the Largest side selector

Resizing oversized camera originals is the fastest way to slash megabytes without pushing lossy compression into blocky artifacts. The Largest side dropdown lets you choose Keep size, 3840 px (4K), 2560 px, 1920 px (Full HD), or 1280 px (web, email).

Picking 3840 px, 2560 px, 1920 px, or 1280 px caps

For email attachments, CMS blog posts, and portal uploads, 1280 px or 1920 px provides sharp full-screen clarity at a fraction of a 4000 px raw photo’s byte size. Keep 2560 px or 3840 px for high-resolution hero banners or digital displays.

Images whose longest edge is already smaller than your chosen Largest side cap are never upscaled (Math.min(1, ...)), so mixed batches of large camera photos and small thumbnails can share one setting safely.

Why progressive halving keeps resized edges crisp

Instead of shrinking a 4000 px photo down to 1280 px in one coarse jump, drawResized() steps down in halves (cw / 2 >= targetW) with imageSmoothingQuality = "high" before the final draw. That progressive halving prevents jagged moire patterns on fine fabrics, brickwork, and UI text.

According to the W3C Portable Network Graphics (PNG) Specification Third Edition, PNG always uses lossless Deflate compression, so lowering Largest side is the only way to shrink a PNG without switching to WebP or JPG. If you are starting from modern .avif files for an older portal, you can turn modern AVIF shots into universal JPGs directly.

Local browser encoding and EXIF metadata privacy

Every image is decoded, resized, and encoded inside your browser tab using local memory, never uploaded to an external compression server. Images larger than 4096 px on a side or 16,777,216 total pixels are clamped proportionally to stay within browser 2D canvas limits.

Orientation correction without remote uploads

As specified in the MDN createImageBitmap() Web API Reference, decoding with imageOrientation: "from-image" rotates portrait phone photos upright before resizing or encoding.

Tip: If you drop a folder of PNG screenshots and see the Quality dropdown greyed out, switch Format to WebP (smallest for websites) to shrink them by 70% or more while keeping sharp text.

When re-encoding strips EXIF versus keeping original bytes

Re-encoding or resizing via canvas automatically strips camera EXIF tags and GPS coordinates from the -compressed output. When the no-bloat guard hands back your untouched original file, you can strip EXIF and GPS tags without re-encoding separately.

Questions

Why is the Quality dropdown disabled when I add PNG images?

PNG uses lossless Deflate compression, so a lossy quality slider has no effect on PNG encoding. Switching Format to WebP (smallest for websites) or JPG (works everywhere) immediately unlocks the Quality presets and shrinks large PNG screenshots.

Should I choose WebP or JPG for smaller website images?

Choose WebP (smallest for websites) when targeting modern browsers, because WebP produces smaller files than JPG at matching visual quality and keeps transparent PNG backgrounds intact. Choose JPG when you need universal compatibility with older email clients or upload portals.

What happens if an image is already smaller than the compressed output?

The compressor includes a built-in no-bloat guard. When Largest side is set to Keep size and Format stays the same as the original, it returns your original file bytes untouched if re-encoding would produce a larger file.

How does lowering the Largest side option reduce file size?

A 4000 x 3000 phone photo holds 12 million pixels, whereas capping Largest side to 1920 px (Full HD) or 1280 px (web, email) cuts total pixel count by 75% to 90% before compression even starts.

Does browser compression strip EXIF and GPS metadata?

Yes, whenever an image is re-encoded or resized onto the canvas, embedded camera EXIF tags and GPS coordinates are dropped. Only files returned untouched by the no-bloat guard retain their original metadata.

Can I compress multiple photos at once and download a ZIP?

Yes. Drop an entire folder or selection of images, pick your Quality, Largest side, and Format settings, and select Compress N images to process every file locally and download one ZIP archive.

Updated