Foreigners who never met before when they run into each other in Japan 🇯🇵

What a time to be alive! #StarWarsOutlaws #PlayStationPortal #Tokyo

Great food, great fun, great city. See you soon Tokyo! 👋🛫

"We have the core gameplay, now we just need the writers to come up with a story." The core gameplay:

Thoroughly enjoying exploring Star Wars Outlaws and the fantasy around visiting the local cantina, work for hire and gambling.

GameNGen (Valevski et al., 2024) can interactively simulate DOOM at over 20 FPS on a single TPU. 1) an RL-agent learns to play, training sessions are recorded; 2) a diffusion model produces the next frame, conditioned on past frames and actions. Project: gamengen.github.io

PartwiseMPC (Khoshsiyar et al., 2024) for contact-rich physics-based character control without motion capture. Paper: cs.ubc.ca/~van/papers/20…

Just Nix, living its best life. #StarWarsOutlaws

SkillMimic (Yinhuai et al., 2024) learns diverse skills from human-ball motion datasets, enabling smooth skill switching even if these switches are not present in the reference dataset. Project: ingrid789.github.io/SkillMimic/ Paper: arxiv.org/abs/2408.15270…

Massive congrats (pun intended) to everyone at Ubisoft on the launch of #StarWarsOutlaws 👏🎉 Feels like the Rogue One of Star Wars video games to me. Very much enjoying exploring it, living in it. Fantastic work 🫡

MarioVGG by Virtuals Protocol is a text-to-video diffusion model that can continuously generate scenes and levels of Super Mario Bros., simulate the physics and movements of a controllable player all through video. Project: virtual-protocol.github.io/mario-videogam… Paper: virtual-protocol.github.io/mario-videogam…

$699 will seem a bit much for Knack 3, but don't worry, it comes bundled with a PS5 Pro.

DataGemma, open models built from the same research and technology used to create the Gemini models, enhance LLM factuality by grounding them with real-world data from Data Commons. Blog: goo.gle/4gooxdG

NotebookLM's Audio Overview can generate an engaging discussion about your latest project, summarized and analyzed from PDFs, Google Docs, Slides etc. by two AI hosts like you were listening to a podcast. It's really quite something, try it out! ▶️ notebooklm.google

GameGen-O from Tencent (Haoxuan Che et al., 2024) is a diffusion transformer model for generating controllable videos of open-world video games. Project: gamegen-o.github.io

PALO (Vivek Myers et al., 2024) is a novel approach for few-shot adaptation to unseen tasks that exploits the semantic understanding of task decomposition provided by VLMs. It achieves the same performance as finetuning with >15x less data. Project: palo-website.github.io

WonderWorld (Haoyi Duan et al., 2024) generates 3D scenes interactively. It takes an input image and follows player movement to create a 3D scene in less than 10 seconds. Project: kovenyu.com/WonderWorld/ Paper: arxiv.org/pdf/2406.09394

Imagen 3 & Veo, our best image and video generation models, are coming to Dream Screen to empower YouTube creators. Each creation will be watermarked using SynthID, and YouTube will apply a label that communicates to viewers that it was generated with AI. deepmind.google/discover/blog/…

Find someone who looks at you like Nishino-san looks at the PlayStation 30th Anniversary Collection.

Gemini 1.5 Flash & Pro models got better and cheaper! 💲-50% token cost on Pro (both input & output for prompts <128K) 🙌 2x higher rate limits on Flash, ~3x on Pro ⏩ 2x faster output, 3x lower latency 💪 Improved quality in math, long context and vision developers.googleblog.com/en/updated-pro…