Originally published at: https://boingboing.net/2020/02/01/new-machine-learning-algorithm.html
…
It can also generate never-before-seen cars, churches, and animals.
Wow, it’s “near perfect”…
Perhaps it could generate a new POTUS? That would be a nice upgrade.
How long before we find out they faked their results?
Faking the eyes is the hard part.
one ‘e’ away from a muted topic
I know there are some issues here (e.g. being able to create unsearchable images for sock-puppet social media accounts), but from an art perspective this still seems really cool to me. (I mean, I love that, for say a video game, one can now create a photo-realistic image that fits a specific set of criteria - you can create a character background and generate an image to match. I’ve been playing with some pre-generated faces for a project, and it’s useful that the faces are all lined up exactly the same way - if I want to overlay glasses, say, onto faces, the eyes are all in exactly the same place in the pictures, making it easy.)
Now I’d like to see the combined output of those models. Let’s see some car-people, some cat-churches!
I wonder if the eyes are all the same distance apart like I noticed another algorithm produces.
Is there non evil things you can do with that ?
So, can they use the mixing algorithm to mix human faces with cats? Probably a better result than that movie, and probably also horrifying.
After thiscatdoesnotexist.com I’m not willing to look.
I’m still recovering from those nightmare fuel felines.
Use it to spam facial recognition databases?
Seems like they’ve made some eye improvements.
The family has been playing which face is real for a while. It’s easy to tell when you know what to look for. This is a new algorithm and looks like it will be much, much harder.
Misread the title. I couldn’t understand why someone would want to make a computer algorithm create near perfect human feces.
Now you no longer have to embarrass a real model as the hemorrhoid person, or the person with STDs in ads.
Okay that’s a great one.
Meh, you just can use drawings or just not show a face.
We use them for both print and digital as placeholder images for mockups and demos that have portraits - like company About pages or conference information. No issue with rights management, especially when it comes to faces of kids (one of our teams works with high-school sports photos, which are a copy/rights nightmare to deal with).
…which is now powered by the improved algorithm described in the video.
Previously:
This topic was automatically closed after 5 days. New replies are no longer allowed.