Image Compressor

How to compress an image without uploading it

1. Choose or drop JPG, PNG or WebP files. You can add up to 25 images to one batch; unsupported or oversized files receive an error message.
2. Choose Output format. Keep original type uses each image's input format. For a photograph, compare JPEG or WebP; for sharp text or transparency, begin with PNG.
3. Start with the default 80% quality for JPEG or WebP. Set Max width if the original has more pixels than you need. Leaving it empty keeps the original pixel dimensions.
4. Click Compress images, inspect the previews and reported sizes, then use Download on each result you want to keep.

The tool creates separate files named with -compressed before the output extension. It never overwrites your originals and does not offer a combined ZIP download. The preview is scaled to fit its card, so also inspect the downloaded image at the size where it will be used.

Worked example: resize a phone photo for a web page

A 4000 × 3000 photo contains 12,000,000 pixels. Set Max width to 1920: the tool computes a proportional height of 1440, producing 1920 × 1440 pixels. That is 2,764,800 pixels, or 23.04% of the original pixel count. It removes about 77% of the pixels before encoding, so reducing dimensions can have a larger effect than repeatedly lowering quality.

File bytes do not fall in a fixed ratio to pixel count. Detail, noise, transparency, format and the browser's encoder all matter. For a separate hypothetical size example, if a 3 MiB original becomes 300 KiB, the saving is 90.2%: (1 − 307,200 ÷ 3,145,728) × 100. These are arithmetic examples, not promised compression results.

The width control only scales down. An 800 × 600 image with a 1920 maximum stays 800 × 600. Height is rounded to a whole pixel, with a minimum of one pixel; the tool does not crop or stretch the image to a separate height.

Get an image below a form's file-size limit

1. Read the destination's rules first. Confirm its allowed file types, maximum byte size and minimum pixel dimensions. Renaming a file extension does not convert the image.
2. Reduce excess dimensions. For example, try a 1920-pixel maximum for a large photo intended for ordinary screen viewing. Keep any minimum size required by the destination.
3. Compare a suitable format. Try JPEG or WebP for a photo, provided the form accepts it. Keep PNG when crisp text or transparency is essential.
4. Adjust quality gradually. For a photo, compare 80%, then 70% or 60%. Check faces, edges and fine text before choosing the smaller result.
5. Check the actual downloaded file. If it is still too large, reduce width further or adjust quality and run the batch again. Keep the original if the new file looks worse without a useful saving.

There is no “compress to exactly 200 KB” setting or guaranteed target size. The result cards use KiB and MiB, where 1 KiB = 1,024 bytes and 1 MiB = 1,048,576 bytes. A destination using decimal KB or MB may show a different number; compare actual bytes when you are close to its limit.

Choose quality, format and transparency deliberately

The quality slider sets the browser encoder's parameter from 0.10 to 1.00 for JPEG and WebP; “80%” does not mean 80% of the original size or a measured 80% visual quality. Results can differ between browsers. Lower settings often create smaller files, but can introduce blocks, smearing or halos.

PNG encoding ignores that slider. It is lossless relative to the pixels on the canvas, but resizing, browser decoding and color conversion can still change the image. Saving a JPEG as PNG cannot recover detail previously lost to compression and may create a much larger file. This tool does not encode lossless WebP.

JPEG has no alpha channel, so the compressor fills transparent areas with white before drawing. PNG and WebP can retain alpha transparency. If a logo needs a transparent background, inspect a PNG or WebP output rather than selecting JPEG solely for file size. WebP output is offered only when the browser's support check succeeds; an unexpected encoder fallback is reported as an error instead of silently delivering the wrong format.

Limits, memory and unsupported files

Each batch accepts up to 25 files, each no larger than 40 MiB (41,943,040 bytes). Decoded images must have at most 32 million pixels and no side longer than 16,384 pixels. A supplied maximum width must be a whole number from 1 to 16,384.

Those are tool limits, not a promise that every device can process the largest allowed image. A 32-million-pixel RGBA canvas alone needs about 128 million bytes, before the source image, encoded copy and previews. A small phone can run out of memory earlier. Start with fewer files, close other memory-heavy tabs or choose a smaller source when necessary. Images are processed sequentially, while completed results remain available on the page.

GIF, SVG, HEIC/HEIF, AVIF and RAW camera files are not supported inputs here. Animated PNG or WebP is flattened to a single frame. This is a raster re-encoder, not an animation optimizer or an archival format-preservation tool.

Troubleshooting compression and downloads

• The result is larger. The original may already be optimized, or the chosen format may be inefficient for its contents. The card marks the result as “larger.” Compare another format or smaller width, or keep the original.
• The file cannot be decoded. It may be damaged, unsupported or beyond the device's available memory. Opening it in another image editor can help identify the problem; changing the filename alone will not repair it.
• The browser cannot encode the requested format. Choose PNG or JPEG and retry. The tool does not disguise a fallback file with a misleading extension.
• Text or faces look blurred. Increase quality or preserve more pixels. For screenshots, try PNG. The small preview cannot show every compression artifact.
• Downloads are missing. Use the Download link on each completed result and check your browser's download list and permission prompts. Save what you need before clearing results or leaving the page.

Settings and selection are disabled during a batch. There is no separate Cancel button for the compressor. Results are temporary browser objects, not an online gallery; Clear all removes the queue and results once processing has finished.

What the new file preserves and removes

The output is a newly encoded raster image. Original camera metadata, GPS tags and animation are not copied across. Your original stays unchanged. Visible identifying content, such as faces, addresses or text within the image, remains visible; re-encoding is not a redaction tool or a forensic privacy guarantee.

Browser encoders can write their own format metadata, and color appearance may change during decoding or conversion. Retain the original for print work, editing or archival use. For a document photo with specific dimensions, use the Passport Photo Maker and check the destination's official requirements.

Method reference: MDN's Canvas toBlob documentation explains browser encoding, quality parameters and format fallback. The limits and fallback checks described above are this tool's own implementation.