Current two example pictures don't do justice to this program.

[fidergo-stephane-gourichon]

This program is pretty good. Slower than regular decoding but pretty good.

Both examples shown on README.md are relevant, but less extreme examples would me more convincing.

• The Lena example is too compressed, okay the result is smoother but not pretty.
• The drawing example indeed shows reduced mosquito noise and chroma noise, yet feels not fully convincing because it's not a photograph.

Perhaps putting examples with typical real-life photographs with typical (not extremed) (over)compression would make a better case.

Keep up the good work!

[MatthiasWinkelmann]

Also: https://en.wikipedia.org/wiki/Lenna#Criticism

[zvezdochiot]

Hi @fidergo-stephane-gourichon .

See also:

• https://github.com/ilyakurdyukov/jpeg-quantsmooth
• https://github.com/ImageProcessing-ElectronicPublications/libjpegqs

[ghost]

I could see infographics as .jpegs being the perfect candidate here.