KB measures the file, not its width
File size is the number of stored bytes. This tool uses 1 KB = 1000 bytes. Pixel dimensions describe width and height, such as 1600 × 1200. Two files with the same dimensions can have very different byte sizes.
Photographs with fine texture usually need more bytes than flat graphics. Changing only a filename or its extension does not compress a file.
Try encoding quality first
JPEG and WebP use lossy encoding. Lower quality can reduce bytes while retaining the same dimensions, but it can blur detail or introduce artifacts. Compare faces, small text and edges at a useful zoom.
The tool performs a bounded search and retains a higher-quality candidate that really fits when one is found. Browser encoders differ, and this is not a proof of the best possible encoding.
Reduce dimensions when the use allows it
For a small on-screen preview, fewer pixels may be enough. Enable resizing only if a smaller width and height are acceptable. For print, documentation or detailed text, preserving dimensions can matter more.
Every smaller candidate is drawn from the original decoded image. The tool does not repeatedly shrink an already-compressed result and never enlarges an image.
PNG uses different controls
PNG does not respond to the Canvas quality value. Palette reduction groups colors to reduce bytes, which may alter gradients or transparent edges. Pixel-lossless mode instead preserves the browser-decoded pixels and keeps dimensions; metadata and color profiles can still change.
Quick compression needs no KB target and retains the smaller original when appropriate. Target KB checks real bytes and only reduces dimensions with permission. PNG palette work is capped at 4 million pixels; native lossless fallback may remain larger than the target.
Choose a format for the background
JPEG is widely accepted for photographs and fills transparent areas with white. WebP supports transparency, but the receiving service may not accept it. The compressor exports PNG, JPEG and WebP. PNG and WebP can preserve transparency.
Animated images are rejected. Re-encoding may remove EXIF and other metadata; orientation is applied once during browser decoding. Keep the original when metadata matters.
When a target cannot be reached
A 1 KB limit may leave too little space for a recognizable image or even the format’s overhead. Quality, resizing, minimum dimensions and device-work limits bound each run.
“Target not met” means exactly that. You can download the smallest candidate with that label, choose a larger target or permit smaller dimensions. The receiving website makes its own acceptance decision.
Keep the original and check the result
If the original already fits, use “Download original” to keep it unchanged. Conversion can sometimes produce a larger file; the tool reports that instead of claiming a saving.
Check the actual bytes, dimensions, format and visual detail before using the result. The target check uses exact bytes rather than the rounded size displayed on screen.
Questions before submitting an image
Why does the receiving website still reject a small file?
It may check pixel dimensions, format or other rules as well as bytes. Check its stated requirements and compare them with the actual downloaded file. This tool cannot decide whether another site will accept it.
Should I type the file size shown by my file manager?
Use the limit specified by the receiving website. File managers can display rounded values or use 1024-byte units. Here, 200 KB means exactly 200,000 bytes; the target check uses the actual output bytes.
How can I keep a copy with its original metadata?
Keep the original file separately. Canvas re-encoding can remove EXIF and other metadata. If the original already meets the limit, the explicitly labeled original download leaves its bytes unchanged.