Parsing the IIIF manifest assumes a 600x is available about pix-plot

[cncoleman]

@duhaime Success with parsing the manifests and returning the images as .pngs.

[duhaime]

Ah sorry about that, I thought IIIF would resize an image into any requested size, but I guess there's a max width beyond which IIIF servers will no longer increase an image's size. So e.g.

# 5000 width works
https://images.britishart.yale.edu/iiif/95696475-8858-4b08-9b04-877cbe2f3e32/full/5000,/0/native.jpg

# 50000 width is too big
https://images.britishart.yale.edu/iiif/95696475-8858-4b08-9b04-877cbe2f3e32/full/50000,/0/native.jpg

I just pushed updates to the iiif-downloader package that's used herein. Could you try to pip install iiif_downloader -U and then retry the build command?

[duhaime]

@cncoleman Did you get a chance to try the new iiif downloader release? I wanted to make sure you were all set on that front...

[cncoleman]

@duhaime I did load the new iiif downloader reease. It is still trying to load a jpg that does not exist and saves it with the common filename iiif.png.
It loads the manifest and then loads the image from the URL. It is this URL that is mistakenly parsed from the manifest. For example, it attempts to load:

https://stacks.stanford.edu/image/iiif/pj302jt0460%2FFarallon_1990_8256_23A24/full/600,/0/default.jpg

which does not exist. If, instead, I change '600,' to 'full' I end up with an image path that works:

https://stacks.stanford.edu/image/iiif/pj302jt0460%2FFarallon_1990_8256_23A24/full/full/0/default.jpg

[duhaime]

@cncoleman many thanks for your follow up and the details of this manifest--I've learned more about IIIF because of this thread. I thought all IIIF servers would serve any size of an image <= the width and height for the given image advertised in the image's manifest, but I see now that an image can advertise discrete width and height attributes in its own metadata json. Curious!

I also had assumed the urls were more consistent but I see the prefix part of the scheme has some variability to it.

I just pushed up a new version of the IIIF downloader with these corrections. You could please try pip install iiif_downloader==0.0.8 and rm -rf output to clear the pixplot output. Then could you try to process the collection again? We should be getting closer now!