Log in | Register
Forum > Site Discussion > Thread

Request: Blacklist for specific pictures

Feb 20, 2021 - edited Feb 20, 2021 - permalink

Unfortunately Korean/Japanese/Chinese "cute" culture leads to so many of these athletes applying face shrinking filters on many of their selfies.

Kim Ji-Hyun doesn't do it to all her pictures, but damn, the ones she does do it to she sets it to 11. Just look at the distortion of the column.


I know I'm not the first person to bring this up. But is there really no other way to deal with this besides blacklisting the person? It's a shame to see great models be banned because of this.

Honestly I think a way to ban specific pictures (via image recognition technology using a computer vision library and not hashing) would be best. This can't possibly be that hard to do. That way, known morphs, pictures from paysites, specific pictures models wanted removed, and more can reliably be kept off the site for good. Also this would help screen for duplicates better, or help with merging to ensure the best image is kept.

Pictures that were rejected by moderation can also be blacklisted so as to prevent reuploads (and complaints), and giving a reason for the rejection upon someone attempting to reupload would probably help improve upload quality.

Feb 24, 2021 - edited Feb 24, 2021 - permalink

So far we blacklist models who self-morph images too frequently. One this is because we don’t want to encourage any sort of morphing and two because from a moderation perspective it becomes a nightmare. For the last point, it is because you get to a point you don’t know what is a morph anymore and it becomes a whole debate and time waster.

The simplest solution: for the model to stop morphing herself.

Feb 24, 2021 - permalink

Thanks for the feedback tamarok. I agree that models manipulating their own images isn't OK for this site. Actually it's a trend that I'd love to see stop on social media.

I looked at Kim Ji-Hyun's pictures some more, started reporting the morphs, then gave up. It's like 1/5 of her 1000+ pictures here.

I still think backend image recognition technology would help in general, with moderation and for users.

« first < prev Page 1 of 1 next > last »