New AI Makes The Mona Lisa Come Alive!
❤️ Check out Lambda here and sign up for their GPU Cloud:
📝 The paper "LivePortrait: Efficient Portrait Animation with Stitching and Retargeting Control" is available here:
Try it now! –
📝 My paper on simulations that look almost like reality is available for free here:
Or this is the orig. Nature Physics link with clickable citations:
🙏 We would like to thank our generous Patreon supporters who make Two Minute Papers possible:
Alex Balfanz, Alex Haro, B Shang, Benji Rabhan, Gaston Ingaramo, Gordon Child, John Le, Kyle Davis, Loyal Alchemist, Lukas Biewald, Martin, Michael Albrecht, Michael Tedder, Owen Skarpness, Richard Sundvall, Taras Bobrovytsky, Ted Johnson, Thomas Krcmar, Tybie Fitzhugh, Ueli Gallizzi.
If you wish to appear here or pick up other perks, click here:
My research:
X/Twitter:
Thumbnail design: Felícia Zsolnai-Fehér –
Hello fellow scholars!
Its destiny, watching 2 min paper.. 2 minute right after it got published
if someone started watching it right when it came out must have just finished it
Vtubers are about to look a lot better.
This is a type of AI that will boost human creativity, going to use it in in-game film making to animate faces when characters talk!
Consider the possibility of translating facial animations and someday even full-body movement 1:1 to any VR avatar in real time.
5:02 I have more fingers and toes than I have teeth, so if I could do job interviews and appear to have teeth, that would be the Holy Grail to me.
You just need a photo with a closed mouth. The AI will imagine teeth for you.
@@reinerheiner1148 I wonder if any job interviewers will recognize young Brad Pitt?
😂 1:26 Didn’t expect to see that quak YouTube dieter Frank Turfano on here.
Now it’s time to never trust any footage, new or historical.
Exactly, I don’t trust this video anyway, technology isn’t moving that fast.
@@DezorianGuy LMAO, go to school.
@@Also_sprach_Zarathustra. I know right, someone hasn’t been paying attention the past year.
@@DezorianGuy bruh
Wow! What a time to be alive!
I think the cosmetic industry is screwed, lol. Imagine every woman on earth getting her hair and make-up done professionally, just once for a photoshoot.
From then on, every Instagram post, every zoom call… she looks exactly how she wants to with zero effort. A little image touchup can change the make-up and hair color, add or hide tattoos, she can sit around in Yoga pants and a stained t-shirt, but will appear to be wearing designer clothes, even if they’re just photoshopped and she never even tried them on, lol.
Just look good in one picture and you’re set for life, lol.
Who needs even a single make-up session when you can use AI to generate your own perfect photo.
Case in point: my 2-years-old YouTube user image.
“What a time to be alive!” (TM)
It’s all good, till its used by scammers
it’s all bad, till its used for curing diseases
@@itskittyme Your life is a scam.
Go to school, and search how to cure diseases, but don’t give a sermon to everyone for something you’re not doing. Especially in a society where half the state budget is spent on military purposes.
Immediately I’m thinking of using it as a tool to help one “get into character” when writing. With better-than-real-time results, a virtual mirror of a fictional character would let me see my own emotions reflected in a character I have only a single image of, helping me to feel like I’ve gotten into their head.
What a time to be Mona Lisa
I tried it—it’s much worse compared to Roop (which uses InsightFace swapper, which is more like single shot DeepFake) and Thin Plane Spline Motion Model. It CANNOT handle any head movements to the left or right; it breaks down completely. Example driver video: https://youtube.com/shorts/wyz3vuDuJGI?feature=share
This solves the viewing angle problem when the webcam is not at the center of the point of focus on the screen when we have a video call. It can transfer over slow connections and present a stereoscopic representation too. It’s the holy grail of video telephony.
The idea of using this AI for facetime-like calls with just an image and a low-res video is genius. It makes me wonder about the implications for remote work and personal communication.
It’s no longer two papers down the line. He can’t keep up with what’s out there now.
I think character animation for cartoons and anime will become so much easier with these techniques.
No, it won’t
For the movie industry:
1) you can make dead actors act
2) you can make old actors young again
3) you just need to purchase rights from an actor to use his face and voice, rest can be done by your team and he’ll be your protagonist without spending a minute on set or doing anything for you.
4) a single actor could “work” in countless movies at the same time. With significantly less fees and without having to actually work!
The problem is, in the world we live in :
Greed might use your voice once dead, couple it with this AI & bam, make a profit without your consent…
(Or someone reports a crime & uses your face this way to put you in jail)
No. An actor is more than a model.
@@brexitgreens what of him can an AI not replicate? I’m talking 5/7 years from now…
@@brexitgreens I’m actually thinking about AI making bad actors/non actors like myself look like professionals in the future…
So lawyers will now have an even harder time convincing judges they aren’t a cat?