Ben Poole

knock knock #veo3

@SpAItial_AI yes please. tdm?

We wired 3D Gaussian Splatting into the 1999 Quake 3 engine - it's fully playable! Echo-2 generates the game levels, and it's rendering spaces directly inside the open source version of id Tech 3. Anyone wants to try it?

@SpAItial_AI congrats! what's world score evaluation?

Echo-2 can generate a diverse set of environments. Results are spatially-persistent by design, a critical difference from many video models. We use 3DGS for extremely fast, real-time rendering; visual quality achieves state-of-the-art using the world score evaluation.

🚀Echo-2 is here - our new world model! These aren’t videos. These are 𝟑𝐃 𝐬𝐜𝐞𝐧𝐞𝐬. Generated from a single image. - Stunning visual quality. - Real-time rendering. - Interactive camera control. - Physically grounded. 🧵More details👇

Meet Vision Banana 🍌 from @GoogleDeepMind! We provide strong evidence that image generators are generalist vision learners. Traditional computer vision tasks (segmentation, depth estimation, normal prediction) can now be performed at/near SOTA with a single generalist model derived from an image generation model. 🖼️ Explore the results: vision-banana.github.io 📄 See details at: arxiv.org/abs/2604.20329

We release Diamond Maps💎 unlocking accurate and efficient guidance for diffusion models. Our experiments show that our methods scale incredibly well. Excited to see what people will build with this! Accurate guidance has been a notoriously hard problem, but in this work, we’re bringing TWO (!) solutions to the table. The recipe for success: 1️⃣ Speed: Use distilled models (flow maps, mean flows, consistency models). 2️⃣ Exploration: Inject stochasticity to properly explore your search space. Because this fundamentally improves anything using flow matching and diffusion, we see a lot of potential for applications across audio, robotics, molecules, and beyond. Paper: arxiv.org/abs/2602.05993 Code: github.com/PeterHolderrie… Huge thanks to an amazing team: Douglas Chen, @LucaEyring, @ishin_shah, Giri Anantharaman, @electronickale, @zeynepakata, Tommi Jaakkola, @nmboffi, and @max_simchowitz. It was awesome bringing this to life together!

@DylanTFWang looks dope!!

Genie3 generates videos. We generate 𝟯𝗗 𝘄𝗼𝗿𝗹𝗱𝘀 you can actually use. Launching tomorrow — Tencent #HYWorld 2.0, an engine-ready World Model🚀 This isn't a video. It's a real 3D scene, all generated & editable. One image in. A whole 3D world out. 🔥Open-source tomorrow

Today, we're open sourcing the code behind "The Infinite Crate," our VST/AU plugin that lets you play with Lyria RealTime directly inside your favorite DAW! 🎧🎶 Since we released The Infinite Crate last year, it’s been used by some of our favorite artists - including a wonderful showcase with @daitomanabe in Tokyo - and being featured as a top new music tool at NAMM 2026. We’re now fully open sourcing the plugin for developers to fork, modify, and make their own under the permissive Apache 2.0 license. 💾 Get it here: github.com/magenta/the-in…

@robrombach @venturetwins sick

New paper out! We present a training method for multimodal generative models, called Self-Flow, which combines classic flow matching and representation learning. Why? Unlike most representation alignment methods, our new approach does not require external, pretrained models and thus scales gracefully to joint multimodal training on images, videos and audio. How? It combines per-timestep flow matching with dual-timestep representation learning, improving the models' internal representations. This approach outperforms prior methods and shows promising scaling behavior in multimodal pretraining. It also enables downstream applications such as action prediction for embodied AI. webpage+paper: bfl.ai/research/self-… code: github.com/black-forest-l… Credit to @hila_chefer, @pess_r, Dominik, @dustin_podell, Vikash, @Vinh_Suhi and Antonio. If you enjoy doing open research like this, come and join BFL! We are actively hiring🌲

1/6 Introducing Unified Latents: what if your diffusion model's latents were measured in bits? Instead of relying on dimensionality reduction, we learn a latent AE with explicit bitrate control. Paper: arxiv.org/abs/2602.17270 @emiel_hoogeboom, @TimSalimans

neuroAI comparisons of ANNs to brains do have a range of problems. Even more than I had realized. And I was worried before: biorxiv.org/content/10.110…

World models for the real world! Awesome work with incredible colleagues at Waymo 🤖🚙

We’re excited to introduce the Waymo World Model—a frontier generative mode for large-scale, hyper-realistic autonomous driving simulation built on @GoogleDeepMind’s Genie 3. By simulating the “impossible”, we proactively prepare the Waymo Driver for some of the most rare and complex scenarios—from tornadoes to planes landing on freeways—long before it encounters them in the real world. waymo.com/blog/2026/02/t…

The Genie is out of the bottle 🧞‍♂️ Try it now: labs.google/ProjectGenie

🚨NEW LABS EXPERIMENT🚨 Introducing Project Genie, an experimental prototype that lets you create and explore infinitely diverse worlds! Prompt with images or text to create a living, expanding world that builds itself in real-time around you. Access is rolling out today to Google AI Ultra subscribers (US only, 18+) Learn more: labs.google/projectgenie

Introducing Project Genie: An experimental research prototype powered by Genie 3, our world model, that lets you prompt an interactive world into existence — and then step inside 🌎

I got early access to Project Genie from @GoogleDeepMind ✨ It's unlike any realtime world model I've tried - you generate a scene from text or a photo, and then design the character who gets to explore it. I tested dozens of prompts. Here are the standout features 👇

🧞‍♂️🧞‍♂️🧞‍♂️

This is absolutely shameful. Agents of a federal agency unnecessarily escalating, and then executing a defenseless citizen whose offense appears to be using his cell phone camera. Every person regardless of political affiliation should be denouncing this.

I'm proud to have had @JeffDean as my lead at Google for the last 10 years. He is one of the few people of his stature in our industry always willing to stand up and say the obvious. We can't let this administration scare us away from acknowledging the reality we are living in. It's time to stand up and use our voices to end this.

While generative AI continues to impress, it takes a dedicated group of incredible artists, storytellers, and researchers to create a compelling film. Check out the awesome work from Connie and team on Dear Upstairs Neighbors:

Vision-language models are getting better every day. Can we use them to improve image compression? Yes! For my internship, working w/ @GoogleDeepMind, @GoogleResearch, we designed VLIC, a diffusion autoencoder post-trained with VLM preferences. Our preprint is out today! A🧵: