OpenAI's DALL-E 3 - The King Is Back!
4:51

OpenAI's DALL-E 3 - The King Is Back!

Two Minute Papers 22.09.2023 174 185 просмотров 6 817 лайков

Machine-readable: Markdown · JSON API · Site index

Поделиться Telegram VK Бот
Транскрипт Скачать .md
Анализ с AI
Описание видео
❤️ Check out Lambda here and sign up for their GPU Cloud: https://lambdalabs.com/papers The DALL-E 3 announcement is available here: https://openai.com/dall-e-3 Their video is available here (thanks for the Scholarly representation!): https://www.youtube.com/watch?v=sqQrN0iZBs0 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, Bret Brizzee, Bryan Learn, B Shang, Christian Ahlin, Gaston Ingaramo, Geronimo Moralez, Gordon Child, Jace O'Brien, Jack Lukic, John Le, Kenneth Davis, Klaus Busse, Kyle Davis, Lukas Biewald, Martin, Matthew Valle, Michael Albrecht, Michael Tedder, 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/ #openai #dalle3

Оглавление (1 сегментов)

Segment 1 (00:00 - 04:00)

Big day today, DALL-E 3, the third version of the  legendary text to image AI is coming! No, we can’t   try this yet, there is no product or paper yet,  only the initial announcement. Now, it seems that   it does three things better than all the other  techniques out there. So, what does it do well? Well, first, it listens! Now, I hear you  asking, Doctor, what do you mean by that? Well,   with other techniques, important parts  of our prompts often get lost. However,   not here. When we enter a detailed prompt, it  really tries to take all of it into consideration.    If I write the prompt, I want to get that  moustache to be noteworthy, and this one,   I would say, is. And some of these prompts can get  bloody long, and I just kept looking and looking,   but I think I might see everything that was  requested here. Even for things that are   really difficult to paint, sometimes,  even imagine. Try to imagine a whirlwind of   porcelain fragments, harmonious fluid forms in  a dreamlike atmosphere and so on. Excellent. Two, I am wondering, can this really  compete? I mean, Midjourney and Stable   Diffusion have become so incredibly  good, I can imagine that even DALL-E   3 might have a hard time competing. Now,  we have a little indication here. Oh yes,   the legendary prompt from DALL-E 2. “An expressive  oil painting of a basketball player dunking,   depicted as an explosion of a nebula. ” And now,  look, this is what it looks like with version 3.    More detail. More definition. More life. I find  it a great deal better. Great job, little AI! And three, now hold on to your papers Fellow  Scholars, for better integration with ChatGPT.    You see, we don’t even need to write the prompts  directly, we can even ask it to make up a new   character. Let’s call him Larry the hedgehog. And  we get Larry. Larry is amazing, so let’s get some   more images, and now you immediately see that  getting several images created about the same   character is not a trivial matter at all. We’ve  looked at a bunch of papers that tried to pull   that off, and now it is here. But, every hedgehog  needs a house, so it can imagine that too,   that is a beautiful house, and, look at that!   Text! Fantastic. I really want that because   we have been promised proper text support in  text to image, and at least for me, it never   quite worked out without significant cherry  picking. I hope it is going to be better here. And then, finally, we can ask for  stickers of the amazing Larry,   and even a bedtime story with him. I have a  7 year old daughter and just thinking about   the amount of fun we are going to have with this  gives me goosebumps. Love it. Thank you so much! Now, one more thing. Note  that there is no paper. Here,   I can only comment on what I see in  the announcement, and the cases that   we see here are usually not the average,  but the best case scenarios. However,   we will be able to try this and find out  soon, no doubt. What a time to be alive! I was happy to see that it will  not create an image in the style   of a living artist. And when I watch  their video, what do I see? Proper,   and I mean proper scholarly representation,  that is what I see. Well done Fellow Scholars!

Другие видео автора — Two Minute Papers

Ctrl+V

Экстракт Знаний в Telegram

Экстракты и дистилляты из лучших YouTube-каналов — сразу после публикации.

Подписаться

Дайджест Экстрактов

Лучшие методички за неделю — каждый понедельник