http://forums.furaffinity.net/threads/reverse-image-searching-within-the-site.113402/#post-2818329
Don't count on something like this ever being implemented...
Comparing images is not even close to easy to code, not to mention setting up a search engine for it.

No offence, but just being realistic here...

Since http://iqdb.harry.lu/ is pretty good at reverse searching e621, I was wondering if the above is still true.
Aside from the fact that a simple MD5 search would not be impossible to implement on Furaffinity.

I've heard that for them to host the search it needs to be through their servers, which aren't strong enough for it, but considering the only thing stopping google from reverse image searching it is the fact that FA hid themselves from Google, I'm sure it's actually possible to implement a way to do it.

FA has made it very clear they do not want you to easily find original artist for stuff. No matter how easy it would be to implement reverse search engine, their regular search is already absurdly bad that they clearly do not want to improve it at all, let alone adding reverse feature.

Because I can kinda understand blocking googles bot on explicit material, as I have seen material removed here for that very reason, but for SFW material what's the reason? Not being able to find source for image you found on tumblr which has signature removed?

just let FA continue its slow, painful death

Mario69 said:
FA has made it very clear they do not want you to easily find original artist for stuff. No matter how easy it would be to implement reverse search engine, their regular search is already absurdly bad that they clearly do not want to improve it at all, let alone adding reverse feature.

Because I can kinda understand blocking googles bot on explicit material, as I have seen material removed here for that very reason, but for SFW material what's the reason? Not being able to find source for image you found on tumblr which has signature removed?

Yeah I use reverse image searches to find artist/source. Not always easy when accounts have been closed and they have 4 names.

lol Knotty Curls' avatar post #940838
FA: "kill me"

pixelPile said:
lol Knotty Curls' avatar post #940838

i still remember something he once said but i fear it might upset him if i say it.

lol ...rave something's...

no, wait...now it'd be rave poofs!

Comparing images is pretty much uniformly done by reducing the image to a very compact 'fingerprint', and comparing that with the fingerprints on file.

Findimagedupes implements this in Perl -- more precisely, it finds duplicate images, but the exact same approach can be used for reverse search. You probably would want something more performant than that, but in any case, the man page for findimagedupes demonstrates how simple an effective reverse search can be:

To calculate an image fingerprint:
1) Read image.
2) Resample to 160x160 to standardize size.
3) Grayscale by reducing saturation.
4) Blur a lot to get rid of noise.
5) Normalize to spread out intensity as much as possible.
6) Equalize to make image as contrasty as possible.
7) Resample again down to 16x16.
8) Reduce to 1bpp.
9) The fingerprint is this raw image data.

To compare two images for similarity:
1) Take fingerprint pairs and xor them.
2) Compute the percentage of 1 bits in the result.
3) If percentage exceeds threshold, declare files to be similar.

Of course, there are always speed concerns (you have to compare the input image to every single image you have on file), but the above algo is both reasonably fast and simple.

I can't comment on any difficulties that may be involved in running this on a server, but since the OP quote wasn't about that..