it's pretty good – try using personalization or messing around with the prompts, every time there's a new model you have to tweak things a bit settings / prompt wise
GENERATIVE AI
-
World Models Development Excitement Shared by AI Researchers
By
–
Love this 🧵@ccamp___ and @Tocelot ! Thanks for sharing the excitement of building World Models! @theworldlabs https://t.co/ANyrvRSExs
— Fei-Fei Li (@drfeifei) 4 avril 2025Love this @ccamp___ and @Tocelot ! Thanks for sharing the excitement of building World Models! @theworldlabs
-
Understanding AI behavior through image differences and mathematics
By
–
it's not trying to make your images look like the ones you chose it's looking at the differences between the images and trying to understand how that effects your behavior – trust the math!
-
AI Model Discovers Novel Physics Solutions at Brookhaven Laboratory
By
–
o3-mini-high helped a Brookhaven National Laboratory researcher find novel exact solutions to a physical model: https://
arxiv.org/pdf/2503.23758 -
Gemini Team Achieves Recognition for AI Advancement
By
–
Thanks! Kudos should go to the whole Gemini team!
-
Making AI-generated images look more natural and realistic
By
–
we tried really hard to make the images look more natural
-
Fixed Patch Compression Destroys Small Image Details
By
–
pretty well! but nothing is perfect as the images go through a fixed patch size compressor that will eventually destroy small details
-
New slow HQ mode announced for low latency performance
By
–
we like low latency, give it a shot! or use the 'slow' hq mode (which is our main announcement)