What compression changes
Image compression stores visual information using fewer bytes. It may encode repeated patterns more efficiently, discard detail the encoder predicts people will notice less, reduce pixel dimensions, or combine these techniques. Compression is therefore not one operation with one predictable percentage.
Lossy and lossless methods
Lossless compression can reconstruct the encoded pixels exactly. PNG commonly uses this approach. Lossy compression approximates the source to save more space; JPEG and lossy WebP are examples. Lossy artifacts tend to appear around hard edges, fine text, high-contrast patterns, and smooth gradients.
A quality slider is an encoder instruction, not a universal score. “80” in two browsers or formats does not guarantee identical bytes or appearance. The source content also matters: a smooth sky, a detailed tree, and a screenshot compress differently at the same nominal setting.
Dimensions versus quality
A 4000 × 3000 image contains twelve million pixels. If it will be displayed at 1000 × 750, reducing dimensions removes three quarters of the width and height samples before encoding. That often saves more than repeatedly lowering quality on an unnecessarily large image.
Resize for the actual display need first, then adjust compression. Avoid repeated decode-and-save cycles because every lossy generation starts from pixels already affected by the previous generation.
How target-size compression works
A target such as 200 KB is not a simple setting. An encoder must try a quality, measure the result, and search again. If even very low quality remains too large, dimensions must be reduced. YesFreeTool’s Image Compressor uses an iterative JPEG quality search in target mode and can progressively downscale. The target is an attempt, not a guarantee.
In quality mode, PNG stays PNG, while non-PNG inputs are exported as JPEG. In target mode the result is JPEG with a white background, so transparent source regions do not remain transparent.
Common mistakes
- Judging quality only from the file-size number instead of opening the output.
- Converting an already compressed JPG to PNG and expecting lost detail to return.
- Keeping huge dimensions for a small thumbnail.
- Deleting the original before checking text, gradients, transparency, and colour.
- Assuming metadata will survive a browser canvas export.
Try it carefully
Use Resize Image when pixel dimensions are the main problem. For format selection, compare JPG, PNG, and WebP. Always retain the original when the image has archival or editing value.