Feature Request:

Name
Email
Subject
Comment *
File	Select/drop/paste files here
* = required field	[▶ Show post options & limits] Confused? See the FAQ.

Embed	(replaces files and can be used instead)
Options	Do not bump (you can also write sage in the email field) Spoiler images (this replaces the thumbnails of your images with question marks)
Password	(For file and post deletion.)
Allowed file types:jpg, jpeg, gif, png, webm, mp4, swf, pdf Max filesize is 12 MB. Max image dimensions are 10000 x 10000. You may upload 5 per post.

File: 3650c818928e900⋯.png (130.53 KB, 818x634, 409:317, Untitled.png)

File: f8fe89fb16938dc⋯.png (118.42 KB, 825x574, 825:574, Untitled2.png)

File: 47a4a243ce3bb85⋯.png (104.6 KB, 636x538, 318:269, Untitled3.png)

Feature Request: Anonymous 08/30/16 (Tue) 04:59:43 cff760 No.3796

dupdetector, but inside hydrus.

if anyone here doesn't know about it, give it a spin. it can actually detect visually similar images that have been altered, reencoded, even cropped, flipped or desaturated. i run it on my hydrus network folder every now and then, and then have hydrus check file integrity to remove the dead entries left behind. i'd love to have something like this integrated in hydrus, it'd be amazingly useful.

it seems to create a database file with some summarized data about each image's color profile, luminance distribution, probably also stores a grid of average color etc. - and then compares that data with adjustably flexible specificity.

Anonymous 08/30/16 (Tue) 22:48:22 32a1fa No.3811

First of all, that program is not open source so it's impossible to integrate it. Second, it doesn't run on linux nor on mac.

If you mean just a general dupe finder for the entire library, the dev stated multiple times that it's in the works and will eventually be done.

Anonymous 08/31/16 (Wed) 03:28:13 cff760 No.3819

>>3811

obviously i don't mean the actual program, just the specific methods that it uses to find dupes - which are very effective, *very* fast, and very clean

hydrus_dev ## Board Owner 09/01/16 (Thu) 00:27:59 79f7e2 No.3835

File: ddff64aa89a81a9⋯.jpg (839.84 KB, 3009x2181, 1003:727, ddff64aa89a81a99f46c97f0ce….jpg)

Thanks for this. The framework for basic visual dupe detection (based on comparing image 'shape') is already in place. You can try it out with system:similar to or by right-clicking a thumbnail and selecting find very similar images, but the search is very slow.

Once I am done with the suggested tags control, I will be writing a faster algorithm for this search. And once that is reliable and fast, I'll be able to expand the framework by adding more algorithms and comparison-judging and -actioning gui. I would appreciate your feedback as this stuff rolls out!