# Midjourney AI: A League Above DALL-E 2! ## Метаданные - **Канал:** Two Minute Papers - **YouTube:** https://www.youtube.com/watch?v=mKbssV5hTLE - **Дата:** 30.03.2023 - **Длительность:** 7:14 - **Просмотры:** 136,109 ## Описание ❤️ Check out Weights & Biases and sign up for a free demo here: https://wandb.com/papers More about Midjourney (update: as of the day of publishing the video, the free tier appears to have been discontinued): https://www.midjourney.com/ Icy manipulator: https://twitter.com/DonaldM38768041/status/1636832520296869896 Alien cocktails, robot sushi: https://twitter.com/DonaldM38768041/status/1622363496695615488/photo/4 People, holograms: https://twitter.com/bilawalsidhu/status/1636382452657848331 Interior design: https://twitter.com/nickfloats/status/1635116672054079488 Aging: https://twitter.com/nickfloats/status/1637438460297629697 Hands, fingers (previously): https://twitter.com/graycrawford/status/1521755209667555328 GPT-4 creating prompts: https://twitter.com/LinusEkenstam/status/1635270385137770496 Burger shots: https://twitter.com/LinusEkenstam/status/1635270398630830080/photo/1 and https://twitter.com/LinusEkenstam/status/1635270411054350336/photo/1 Hands and fingers: https://twitter.com/TheCartelDel/status/1636262759947210753/photo/2 Dragon’s eye macro: https://twitter.com/RyszardRiccardo/status/1636296638745509890/photo/1 Apple pie macro: https://twitter.com/AIATMYF/status/1636745162075381760/photo/1 Spider, frog macro: https://twitter.com/Sitegains/status/1639308852712685594 My latest paper on simulations that look almost like reality is available for free here: https://rdcu.be/cWPfD Or this is the orig. Nature Physics link with clickable citations: https://www.nature.com/articles/s41567-022-01788-5 🙏 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, Edward Unthank, Eric Martel, Geronimo Moralez, Gordon Child, Jace O'Brien, Jack Lukic, John Le, Jonas, Jonathan, Kenneth Davis, Klaus Busse, Kyle Davis, Lorin Atzberger, Lukas Biewald, Matthew Allen Fisher, Matthew Valle, Michael Albrecht, Michael Tedder, Nevin Spoljaric, Nikhil Velpanur, Owen Campbell-Moore, Owen Skarpness, Rajarshi Nigam, Ramsey Elbasheer, Richard Sundvall, Steef, Taras Bobrovytsky, Ted Johnson, Thomas Krcmar, Timothy Sum Hon Mun, Torsten Reil, Tybie Fitzhugh, Ueli Gallizzi. If you wish to appear here or pick up other perks, click here: https://www.patreon.com/TwoMinutePapers Thumbnail background design: Felícia Zsolnai-Fehér - http://felicia.hu Károly Zsolnai-Fehér's links: Twitter: https://twitter.com/twominutepapers Web: https://cg.tuwien.ac.at/~zsolnai/ ## Содержание ### [0:00](https://www.youtube.com/watch?v=mKbssV5hTLE) Dear Fellow Scholars, this is Two Minute  Papers with Dr. Károly Zsolnai-Fehér. Today we are going to have a look at the  incredible capabilities of Midjourney,   a text to image AI that creates fantasy  images that are so good that it’s frankly   almost unbelievable. It is  a league above any other AI   I have seen so far. I’ll show you 5  of my favorite examples in a moment. Now, what is this? Text to image means exactly  what it says, we write a short description,   and the AI paints our image. But  things are changing now. We may   not even need to be writing our own  prompts. What? Why is that? Well,   the GPT-4 AI also just came out, this can be  used with ChatGPT, OpenAI’s amazing assistant. And, one, hold on to your papers, because we  can ask it to break down the key elements of   food photography, what kinds of food should  we use, what should the background look like,   what about lighting, composition, and so much  more. Photography is an art after all. So,   what did we get here? Did we get some sort of  tutorial, or some ideas? Well, don’t forget,   this is a chat assistant, so we can ask it to  tabulate all these ideas, and, oh yes! We get a   table with detailed instructions for each photo.   These instructions can be dumped into a prompt,   which ChatGPT will happily do for us, and  then, what do we do with this? Of course,   we give this to MidJourney and create these  beautiful hamburgers. Look, these instructions   were truly amazing. We can go for a retro-style  top-view, or, if this made us super hungry,   and we wish to get over the top, we can do this  too. All with the help of two AIs combined. Wow. The same concept can be applied  to interior design as well. The   lighting and the mood on this one results  in a stunning picture. I don’t know about   you but I still cannot believe that an  AI made this. Or, two AIs if you will. Now, two we can also create not only  images, but videos with these methods.    Look. If we have made a character that has  the right look, but not quite the right age,   or we wish to age this character over time, we can  do that too. This video is not to be used as-is,   so it should not be confused by techniques  that are specifically tailored for that. But,   if we stop it at individual frames,  these can be used as images. Three, it is now able to do something that  none of the previous text to image AIs were   able to. Interestingly, none of the previous  previous techniques were unable to synthesize   human hands with 5 fingers. Also, they were  often not quite the right shape. And…can that   really be? Is it finally solved? Look.   So, let’s see this one. Wow. Now those ### [3:18](https://www.youtube.com/watch?v=mKbssV5hTLE&t=198s) Fingers (new method) are not only 5 fingers everywhere, but the  detail and the variety of these images are   absolutely incredible. I will now officially  allow it to give a big thumbs up to itself. Four, MidJourney is currently on version 5, and  I found it really surprising that improvement has   still not plateaued. For instance, in version  4, it imagined holograms and people to look ### [3:45](https://www.youtube.com/watch?v=mKbssV5hTLE&t=225s) Version 5 like this, which is not too great. But the same  prompts on version 5. Now that’s what I am talking   about. And just imagine what version 7 will  be capable of two more versions down the line. But we are not done yet. Not even close!   Five, we have a few more amazing results,   and look who this one is coming from. Yes, that’s  right, the legendary computer graphics researcher,   Donald Mitchell. Among many other things, he is  the author of the legendary Mitchell-Netravali   filter, which contributes to creating beautiful  high-quality images in light transport simulations   and more. He asked for some robot sushi, and  I love how imaginative the results were. These ### [4:33](https://www.youtube.com/watch?v=mKbssV5hTLE&t=273s) Robot sushi works both pass for robots, but they also seem  like edible lunch at the same time. Bravo. And,   what better way to wash it down then asking  for a cocktail from an alien planet. One again, ### [4:51](https://www.youtube.com/watch?v=mKbssV5hTLE&t=291s) Alien cocktails they certainly have a strong alien  feel, and I have to say, some of these   also look drinkable. But, be careful,  who knows what they contain. Loving it. And plus one because I cannot resist. My  favorite. Perhaps everyone’s favorite.    Macro photos. Close-up photography of typically  small subjects. I have found this apple pie and   this dragon eye, not only the best AI-made  macro images I’ve ever seen, but these are   so good they might even rival a journeyman  photographer’s images. What a time to be alive! I will note that here is the point where you  usually get some deeper analysis in the video,   however, it is not possible here. This  work likely builds on previous papers,   but it does not have a paper itself that I know  of. However, if a technical document describing   it appears, I will be very excited to  have a look and show it to you. Also,   I expect that you are excited to  try this. If you wish to try it,   the first 25 images are free. After that,  they are asking for a paid subscription. So, which image from these was  your favorite? Have you done your   own experiments? Let me know in the comments  below! And for now, let the experiments begin! Thanks for watching and for your generous  support, and I'll see you next time! --- *Источник: https://ekstraktznaniy.ru/video/13231*