Interesting ongoing discussion on @iiif slack channel about level 0 tiles served from zip files over http and range requests.

Interesting things discovered digging in github:
1. old SZI format https://github.com/smartinmedia/SZI-Format/blob/master/SZI%20format%20description%20-%202018-11-24.pdf
2. openseadragon https://github.com/openseadragon/openseadragon/issues/944
3. GeoTIFFTileSource https://github.com/pearcetm/GeoTIFFTileSource

the last one is particularly interesting. my experiment here https://pub-0f1c9e6ddb92456a85802303778fa724.r2.dev/demo/index.html
a ~600M pyramidal TIFF served from cloudflare R2

#IIIF

@edsu @iiif indeed, we started discussing just after wacz was released a couple of years ago :)
https://twitter.com/atomotic/status/1294934128366944256

the GeoTIFFTileSource approach is not efficient as an image server, because it downloads always all tiles, and not just the region viewed, but is definitively something that can be explored.

the pattern of serving data through range requests is becoming widely adopted, as example visidata https://twitter.com/atomotic/status/1562716593289064448 or duckdb reading remote parquet files.

@edsu @iiif the first time I heard about it I think it was about PDFs (linearization?), although I never understood if viewers like pdf.js actually implements this feature

https://en.wikipedia.org/wiki/Byte_serving

@edsu @iiif no, I was wrong. Is not downloading all tiles.
@jbaiter pointed the GeoTIFF spec https://github.com/cogeotiff/cog-spec/blob/master/spec.md

@edsu @jbaiter at page load, without any zoom/panning, it does 105 requests