OpenAI’s DALL-E 2 - AI-Based Art Is Here! 🧑‍🎨

📝 The paper "Hierarchical Text-Conditional Image Generation with CLIP Latents" is available here:

🧑‍🎨 Check out Felícia Zsolnai-Fehér's works:

🧑‍🎨 Judit Somogyvári's works:

Credits:

❤️ Watch these videos in early access on our Patreon page or join us here on YouTube:

🙏 We would like to thank our generous Patreon supporters who make Two Minute Papers possible:
Aleksandr Mashrabov, Alex Balfanz, Alex Haro, Andrew Melnychuk, Benji Rabhan, Bryan Learn, B Shang, Christian Ahlin, Eric Martel, Geronimo Moralez, Gordon Child, Ivo Galic, Jace O'Brien, Jack Lukic, John Le, Jonas, Jonathan, Kenneth Davis, Klaus Busse, Kyle Davis, Lorin Atzberger, Lukas Biewald, Matthew Allen Fisher, Michael Albrecht, Michael Tedder, Nevin Spoljaric, Nikhil Velpanur, Owen Campbell-Moore, Owen Skarpness, Rajarshi Nigam, Ramsey Elbasheer, Steef, Taras Bobrovytsky, Ted Johnson, Thomas Krcmar, Timothy Sum Hon Mun, Torsten Reil, Tybie Fitzhugh, Ueli Gallizzi.

Chapters:
0:00 Teaser
0:10 What is DALL-E 2?
1:22 Is it as good as human artists?
2:10 1 - Texture synthesis
4:53 2 - Photorealistic faces!
5:36 3 - Cosmopolitan
6:00 4 - Product design
6:18 5 - 3D information
6:58 Building virtual worlds
7:54 +1 - Product design, but better!

Károly Zsolnai-Fehér's links:

[magalareuben]

you sparked the love of AI in me thanks two minute papers you are influencing lives down here in Uganda ❤️❤️❤️

[Twoneight]

Can you make a video about Stable Diffusion? I think it's really interesting since it's Open Source.

[mikopiko]

I love the progress OpenAi is doing within the fields of Ai but I cannot stop thinking about how misleading their name is, they should be named ClosedAi.

If you are looking for results very close to Dall-e 2 but is truly open, then check out Stable diffusion.

[Ulkerer]

We... have already... seen... all this footage.... before... many times. WHAT A TIME TO BE ALIVE!

[chasingdaydreams]

What face restoration program did you use at 5:09 ?

[iau]

The zoom out concept is incredibly fun to watch. I'd love to see multiple runs on the same image to see what crazy creative ideas the AI can come up with.

[brd]

Seriously, I have a feeling that we are seeing the historical key moment. Since last couple of months, we've been seeing some truly magnificent developments happening in the field of artificial intelligence. People have fantasized, imaged, talked about AI on this kind of level for so many years. And, we are seeing these kinds of AI with our own eyes right in the moment! Maybe the breaking point is finally coming... Here we go, age of AI is coming right now🤯! Brace yourselves!

[juliandarley]

being able to change the costume of an actor is one of the key things needed for micro-budget AI-assisted filmmaking. it looks like it is getting closer.

[mm-rjvo]

I need more of that 3D modeling aspect, with VR integration. I want a voice activated holodeck

[TheDefirion]

I've been subscribed for more than 3 years now, thank you very much for bringing all these updates on AI I find it absolutely invaluable!

[onedeadsaint]

when this gets opened up to everyone, we're going to see an absolute explosion of amazing art

[erase]

I can't believe I'm alive to witness this, technology is advancing at an incredible pace, great video thanks for posting!

[goatpepperherbaltea]

I only started paying attention to this stuff when gpt3 came out and I remember a lot of comments about “imagine when it can do pictures” and it’s crazy to me how fast this was able to happen what a time to be alive

[phenomagator]

Using Stable Diffusion I've been able to make some amazing artwork at home on a gaming PC. The image generation time is incredibly short for the amount of detail it's producing. I can't imagine where this technology is going to be two papers down the line. This channel has gotten me very fascinated with AI.

[pandoraeeris]

Two minute papers is turning into ten minute papers!

[Triggerx]

(7:30) Why is the virtual assistant basically Baby Sinclair from "Dinosaurs"? "Not the mama!" lol.
(5:20) Animating a portrait is neat though. It would be really awesome to take old family photos from past generations, then colorize and animate them. Seeing your ancestors face moving, in color, would really bring then to life and help you imagine them more vividly.
(6:03) The "toilet car" would probably actually be great, for use in healthcare to provide better accessibility and dignity. Might even be useful in humanitarian aid too, since they could be more easily transported than a traditional portapotty.

[surajvkothari]

When Dall-E 2 was initially released it seemed like a specialised AI for generating images. But the progress since its launch makes me feel we might have our first glimpse at an AGI!

[IAmNumber]

This will be a huge deal for photorealistic texture generation.

[Zebred]

I imagine that in a couple of years I won't have to keep screwing around with graphics programmes like Blender I'll just have an increasingly specific dialogue with the AI and I'll just tweak everything down to how I want it!

[Calthecool]

I’ve been experimenting with the out painting, and it is truly amazing.