Quite surprised I couldn’t find anything like this in the thread:
It’d be nice if the reverse search interface was embedded and integrated on the new image UI, at least for fetched images (if there was a way to cache an uploaded file so it wouldn’t have to upload twice, that’d be cool too.) As it is now, going to a separate page, pasting the URL again, clicking extra times…it could be much more streamlined. Running a reverse search when you click “fetch” on the new image UI and showing results inline would be great.
I think reverse search could be enhanced by also automatically running a source_url non-image-based search. True that for sites like Twitter and Tumblr where one URL can have multiple images this could include false positives (unless the site began saving which one out of a group of images was selected for uploading), or if an image was modified at some point in time, but that’s why it displays results for further human interaction, in any case.
This could further be enhanced for known patterns on major sites: for example, twitter.com/
[any valid username string]
[status ID here]
*(a variety of extra strings could be here)* resolves to the same content.
for the purposes of reverse lookup.
deviantArt and Tumblr have similar partially-static, partially-mutable URLs. I would guess that this is already handled within whatever library(s) are used for fetching, though it might be done internally and not exposed…
I’m no computer scientist so I can’t promise these always work or are safe:
For that matter, the site could resolve redirects, such as through t.co, fav.me, or in the case mentioned above where some other Twitter username is in the URL (aside from just messing around, this also can happen if someone changes their account name - old links will redirect).
Super extra credit for comparing related URLs like the twimg.com direct image URL from Twitter posts.