OpenAI’s New Image Generator: An AI Revolution!
Check out Weights & Biases and sign up for a free demo here:
4o Image Generation:
Apple terminal:
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:
Benji Rabhan, B Shang, Christian Ahlin, Gordon Child, John Le, Juan Benet, Kyle Davis, Loyal Alchemist, Lukas Biewald, Michael Tedder, Owen Skarpness, Richard Sundvall, Steef, Taras Bobrovytsky, Thomas Krcmar, Tybie Fitzhugh, Ueli GallizziIf you wish to appear here or pick up other perks, click here:
My research:
X/Twitter:
Thumbnail design: Felícia Zsolnai-Fehér –
Wow, I got my ex to chat me “I love you” and it made me blush
bro u deadass posted the second comment on this video and it was about ur niche personal affairs
@@Nayo987 Man says what he feels, unlike my mom, she slaps.
@@Nayo987 Man says what he feels, unlike my mom, she slaps.
the GOAT 2 minutes papers is here again
WHAT A TIME TO BE ALIVE !!!!!
Next up: the same quality in Sora
Just gave it a try, it managed to generate a glass of wine filled to the brim
It still struggles with analog clocks though
@@mihirvd01 yeah that’s gonna be hopeless for a long time I guess
its amazing how well it captures the facial expressions in the “style transfer”
it does a very good job, but it isn’t perfect. exact facial expressions, with the feelings they convey can be really hard to translate, if you are already translating all elements of an image into a different style.
i would like to use the silver medalist shooter Yusuf Dikec at 2:10 as an example. the memetic and cultural significance of the original image lies in how nonchalant and casual he is, as compared to his opposition.
the AI redraws make him look a lot more stern, tense, focused, and serious. thus the core concept the image is meant to convey is lost.
if *any* artist is given the original image, sans this context, and is then tasked to redraw the image, it’s likely the details and emotions in this microexpressions will be lost. i’m sure that if the AI was prompted *with* context and *why* it’s a meme, it would succeed in conveying the necessary microexpressions.
i saw another post about an AI redraw of the “distracted boyfriend” meme where the expressions were off, thus the meaning of the image was completely butchered. it doesn’t take a lot for art to change / lose meaning.
The blend of precision and imagination in these outputs is remarkable. Especially impressive to see such character consistency sustained across multiple scenes.
The next generation of ChatGPT’s image generator is, in my opinion, what will make this technology truly revolutionary.
Now we are waiting for the chinese open-source model to do it at 10x speed
@@Alpha13yt we need that
open source
This officialy hails in the next generation of AI image generation. The prompt-fidelity is unmatched by far. What a time to be alive!
It’s unreal how good it looks
My team and I are already using it (it’s overloaded as I’m writing this) for our series about sentient robots. What would take our tiny team years to produce can now be accomplished in a month or less. While other artists protest and fight against AI tools, we decided to embrace them and create today the folklore of tomorrow. Not sure that our studio will survive, but now we definitely have a chance. Wish us luck!
Good luck!
@@IceMetalPunk Thanks a lot!
Evolution favors the adaptable. You chose wisely.
Keep your AI generated series away from my eyes please, you should disclose that you used AI so that I can stay away from it forever.
@@turnip5178 i’ll do everything in my power to ensure you can’t tell, and then only reveal to you something was AI once you’ve enjoyed it
sweet dreams
In one update OpenAI made so many of your old videos completely obsolete. What a time to be alive indeed.
I like the comparison to older generations at the end there
You’re my original channel for generated images and even in the AI revolution you’re still the best IMO.
FIY The image appear blurry during generation because there is a CSS blur filter applied. Under it, the frontend receives a series of intermediary images one replacing another, with the top part more or less ready and the bottom part heavily pixelated. You can easily see how the FE swaps the images under even the blur. It also appears that after the whole image is generated, it undergoes some kind or an additional refinement that adds details
We’re living in exciting times!
“I felt a great disturbance in the Force, as if millions of graphic designers, animators, illustrators, and photographers suddenly cried out in terror and were suddenly silenced.”
i don’t u have met famous persons like these, it’s truelly amazing to see that.
what a time to be alive