Acceptable Thumbor Usage

[brunohaid]

Building Yet Another Biodiversity Portal, we link to GBIF occurrences and show a preview of the first image in an occurrence. The current 500px is a bit too blurry for our UI (it zooms 2x on hover), and the raw source images, especially at the tail end when a species only has very high res scans of preserved specimens, sometimes crashes the browser.

1. Is it generally OK with you to use your thumbnail setup that way? We'd of course set up our own preview server when and if we get more serious usage, and are attributing and linking directly in every instance.
2. Are you using any higher resolution besides the 500px anywhere that we could use without gumming up any CPU/disc?

[MattBlissett]

Hi,

Thanks for asking. It's fine to use the Thumbor service in this way, so long as you aren't accessing tremendous numbers of images (e.g. all images, or all images from the USA etc). Ordinary use by users is fine, downloading 100,000s of images for use in machine learning is not.

500px is the highest I see that we're using, other than the unaltered image. I think it's fine to pick another resolution, unless @MortenHofft has something else in mind for the new GBIF website.

[MortenHofft]

It is probably a bit early to say we have settled for a format in the new stack, but we currently use 800px height on occurrence detail pages. e.g.
https://api.gbif.org/v1/image/cache/x800/occurrence/5139800301/media/e0e21212e9aeb8554e4eacefc311ff61

[brunohaid]

x800 is great and truly appreciated. Definitely no scraping, we plan a section on every species page with visual backlinks to GBIF, iNat, Wiki Commons, and Flickr.

There are some features that might lead to high usage, eg vernacular name search in 300+ languages could make it an educational resource in populous regions, but once that's the case we'd add our own caching layer.

For now grateful that it's possible to stand on the shoulder of giants, and I'll email you a link once we have it deployed somewhere.

[MortenHofft]

Please do. I like looking at other sites that have similar data.