drag & drop, click, or paste from clipboard
Drop to add
drag & drop, click, or paste from clipboard
Lossless optimization removes bloat that encoders leave behind — redundant metadata blocks, suboptimal compression settings, embedded thumbnails — without touching the pixels themselves. Every pixel in the output is mathematically identical to the input. If no smaller encoding exists, the original file is returned unchanged.
This is the correct tool when quality cannot be compromised: print masters, product photography governed by brand spec, medical images, scientific data, archival files. It is also a sensible first pass on any PNG before deciding whether lossy compression is even needed.
For PNG files, the optimizer tries multiple compression strategies and filter combinations to find the smallest byte representation of the exact same pixel values. PNG's deflate compression has tunable knobs that most exporters leave at default settings — the optimizer exhausts more combinations to find a better one. No pixel value changes. The colorspace, bit depth, and transparency are preserved exactly.
For JPG files, pixel data is untouched in full. The optimizer removes EXIF blocks, XMP tags, embedded preview thumbnails, and comment sections — these are often 20–80 KB of overhead in files produced by cameras and design apps. The DCT coefficients that encode the actual image go nowhere.
The compress tool applies lossy encoding: quality settings below 100 allow the codec to discard image data in exchange for smaller files. Optimization never makes that trade. The two tools serve different jobs. If your target is a specific file size ceiling and pixel accuracy is not critical, compression with a target KB is the faster path. If the pixels must be unchanged — because you are delivering files for print, archiving masters, or feeding a pipeline that will process the images again — optimization is the right choice.
You can use both in sequence across different purposes: optimize the masters, then compress a separate set of web-delivery copies.
Because the optimizer never returns a larger file, it is safe to run on any batch without examining each image first. The worst outcome is that some files come back unchanged because they were already near-optimally compressed. No manual triage needed. Batch up to 200 files, each up to 80 MB, and download the results as a ZIP.
No. If the optimizer finds no improvement, it returns the original file unchanged. You will never receive an output larger than the input.
Both. For PNG, the pixel data is recompressed more efficiently. For JPG, metadata is removed without touching the encoded image data.
Yes. There is an option to preserve the ICC profile so colors render correctly in color-managed applications. Everything else — EXIF, GPS, XMP, comments — is still removed.
Lossless optimization makes zero quality changes. The compress tool applies lossy encoding that changes pixel values to achieve smaller sizes. They are fundamentally different operations.
Anyone who cannot accept pixel changes: print production, brand asset management, scientific imaging, archival workflows, or any pipeline that re-processes images downstream.