Advertise here with Carbon Ads

This site is made possible by member support. โค๏ธ

Big thanks to Arcustech for hosting the site and offering amazing tech support.

When you buy through links on kottke.org, I may earn an affiliate commission. Thanks for supporting the site!

kottke.org. home of fine hypertext products since 1998.

๐Ÿ”  ๐Ÿ’€  ๐Ÿ“ธ  ๐Ÿ˜ญ  ๐Ÿ•ณ๏ธ  ๐Ÿค   ๐ŸŽฌ  ๐Ÿฅ”

kottke.org posts about NVIDIA

This AI Converts Quick Sketches to Photorealistic Landscapes

NVIDIA has been doing lots of interesting things with deep learning algorithms lately (like AI-Generated Human Faces That Look Amazingly Real). Their most recent effort is the development and training of a program that takes rough sketches and converts them into realistic images.

A novice painter might set brush to canvas aiming to create a stunning sunset landscape โ€” craggy, snow-covered peaks reflected in a glassy lake โ€” only to end up with something that looks more like a multi-colored inkblot.

But a deep learning model developed by NVIDIA Research can do just the opposite: it turns rough doodles into photorealistic masterpieces with breathtaking ease. The tool leverages generative adversarial networks, or GANs, to convert segmentation maps into lifelike images.

Here’s a post I did 10 years ago that shows how far sketch-to-photo technology has come.


AI-Generated Human Faces That Look Amazingly Real

The opening line of Madeline Miller’s Circe is: “When I was born, the name for what I was did not exist.” In Miller’s telling of the mythological story, Circe was the daughter of a Titan and a sea nymph (a lesser deity born of two Titans). Yes, she was an immortal deity but lacked the powers and bearing of a god or a nymph, making her seem unnervingly human. Not knowing what to make of her and for their own safety, the Titans and Olympic gods agreed to banish her forever to an island.

Here’s a photograph of a woman who could also claim “when I was born, the name for what I was did not exist”:

AI Faces

The previous line contains two lies: this is not a photograph and that’s not a real person. It’s an image generated by an AI program developed by researchers at NVIDIA capable of borrowing styles from two actual photographs of real people to produce an infinite number of fake but human-like & photograph-like images.

AI Faces

We propose an alternative generator architecture for generative adversarial networks, borrowing from style transfer literature. The new architecture leads to an automatically learned, unsupervised separation of high-level attributes (e.g., pose and identity when trained on human faces) and stochastic variation in the generated images (e.g., freckles, hair), and it enables intuitive, scale-specific control of the synthesis.

The video offers a good look at how this works, with realistic facial features that you can change with a slider, like adjusting the volume on your stereo.

Photographs that aren’t photographs and people that aren’t people, born of a self-learning machine developed by humans. We’ll want to trust these images because they look so real, especially once they start moving and talking. I wonder…will we soon seek to banish them for our own safety as the gods banished Circe?

Update: This Person Does Not Exist is a single serving site that provides a new portrait of a non-existent person with each reload.


This AI Makes High-Quality Slow Motion Videos from Regular 30 fps Video

NVIDIA trained a deep learning framework to take videos filmed at 30 fps and turn them into slow motion videos at the equivalent of 240 or even 480 fps. Even though the system is guessing on the content in the extra frames, the final results look amazingly sharp and lifelike.

“There are many memorable moments in your life that you might want to record with a camera in slow-motion because they are hard to see clearly with your eyes: the first time a baby walks, a difficult skateboard trick, a dog catching a ball,” the researchers wrote in the research paper. “While it is possible to take 240-frame-per-second videos with a cell phone, recording everything at high frame rates is impractical, as it requires large memories and is power-intensive for mobile devices,” the team explained.

With this new research, users can slow down their recordings after taking them.

Using this technique and what Peter Jackson’s team is doing with WWI footage, it would be interesting to clean up and slow down all sorts of archival footage (like the Zapruder film, just to choose one obvious example).


An AI Can Realistically “Paint In” Missing Areas of Photographs

This video, and the paper it’s based on, is called “Image Inpainting for Irregular Holes Using Partial Convolutions” but it’s actually straight-up witchcraft! Researchers at NVIDIA have developed a deep-learning program that can automagically paint in areas of photographs that are missing. Ok, you’re saying, Photoshop has been able to do something like that for years. And the first couple of examples were like, oh that’s neat. But then the eyes are deleted from a model’s portrait and the program drew new eyes for her. Under close scrutiny, the results are not completely photorealistic, but at a glance it’s remarkably convincing. (via imperica)


Celebrity-ish Faces Generated by an AI Program

Artificial intelligence programs are getting really good at generating high-resolution faces of people who don’t actually exist. In this effort by NVIDIA, they were able to generate hundreds of photos of celebrities that don’t actually exist but look real, even under scrutiny. Here’s a video illustrating the technique…the virtual images begin at 0:38.

And here’s an entire hour of fake celebrity faces, morphing from one to the next:

I’m sure this won’t be too difficult to extend to video in the near future. Combine it with something like Lyrebird and you’ve got yourself, say, a entirely fake Democratic candidate for the House who says racist things or the fake leader of a fake new ISIS splinter group who vows to target only women at abortion clinics around the US. (via interconnected)