Closed arcturus closed 2 years ago
We need to find a way to do this image processing in a way which doesn't completely destroy the stability and performance of the service. We also need to be able to do this for colour extraction from both images and favicons.
@jaredkerim we started an experiment around taking screenshots for websites and extracting the predominant colors:
https://github.com/mozilla-magnet/magnet-metadata-screenshot
At some point we would like to transform it into a media service, separated from the original metadata extraction to avoid hitting performance problems. Maybe we can spend some time on it mid September.
@arcturus Yes I was hoping actually that I can separate all the image processing and proxying and caching into a separate service altogether, so we should talk about that more.
Not sure how common it is, but according to http://ogp.me/#structured, we may be able to extract these from
og:image:width
andog:image:height
properties (although I'd guess seeing those in the wild are pretty rare and we'd need to scrape images, or have Firefox check sizes):