What is entropy coding in image compression?

It's the final lossless stage of most image codecs, where frequently occurring values are encoded with short codes and rare values with long ones. PNG uses DEFLATE, JPEG uses Huffman tables, and AV1/WebP use arithmetic or ANS coding.

Is entropy coding lossless?

Yes, always. Entropy coding only changes how data is represented, never the data itself. That's why PNG optimizers like OxiPNG can shrink files with zero quality loss — they only redo this stage more efficiently.

Entropy coding

The final, lossless stage of image compression — squeezing out the last bits by representing common values more compactly.

Definition

Entropy coding is a lossless compression step that represents frequently occurring values with short codes and rare values with long ones. It's the last stage of most image codecs: PNG's DEFLATE, JPEG's Huffman tables, and the arithmetic/ANS coding in AV1 and WebP. Because it's lossless, re-encoding it more aggressively can only help.

In practice

This is what a PNG optimizer like OxiPNG attacks: it re-runs entropy coding with stronger settings and better filters to produce a mathematically identical but smaller file. It's why \"lossless optimization\" is real and why two PNGs of the same pixels can have different file sizes.

PNG compressor What is lossless compression?

Compress your images now — free and private

Drop your files and ImageForge compresses them in your browser. No uploads, no accounts, no limits. Batch process hundreds at once and download a ZIP.

Open the compressor

100% private No signup