Reinventing the Small Wind Turbine
solar.lowtechmagazine.com
solar.lowtechmagazine.com
(For real, it's running on solar with a battery backup: https://web.archive.org/web/20190714175801/https://solar.low...)
https://www.lowtechmagazine.com/2019/06/small-wooden-wind-tu...
But the folks who run that website are the real deal and not "virtue signalers." They're running experiments and creating great content, not trying to get some social currency you hypothesize.
If you'd can demonstrate data suggesting a better encoding, they'll redo the pictures. They optimized for size.
Dithered images: total 284K (but it's only showing 9 images, not the full 11)
Small images, resized from the big source images using Imagemagick's -distort resize (default settings) operator to 16bit per channel PNG, then encoded to JPEG with mozjpeg -quality 70 and default settings: 273K with all 11 images, or only 236K if I make it a fair comparison by deleting the two that are missing from the dithered version.
My JPEG encodings: https://imgur.com/a/Gi12HRl
There are visible artifacts, but nothing I'd call intolerable. Filesize is significantly smaller than the dithered versions, it's full color, and it preserves details destroyed by the dithering (e.g. the support wires in https://solar.lowtechmagazine.com/dithers/wood-turbines-nepa... ). The dithered versions are sharper, but the website throws away that advantage by resizing them with the browser's default resizing. I think most people would call the re-encoded JPEG versions higher quality than the dithered versions.
[1] "Noise Pollution Prevention in Wind Turbines: Status and Recent Advances", https://www.mdpi.com/2071-1050/4/6/1104
[2] "Reduction of Wind Turbine Noise Using Optimized Airfoils and Trailing-Edge Serrations", https://arc.aiaa.org/doi/abs/10.2514/1.38888
It is very pleasant to listen to (I know, because the huge yacht next to my little boat has one).