Image created with gemini-2.5-flash-image with claude-sonnet-4-5. Image prompt: A pristine 1961 Ferrari 250 GT California Spyder in Rosso Corsa red positioned in a minimalist white art gallery, surrounded by floating illuminated frames displaying artistic reflections and variations of the car’s form, premium museum lighting with soft shadows, elegant and contemplative composition, high-end automotive photography meets contemporary art installation.
We’re rolling out an upgrade to @NotebookLM’s Video Overviews: New visuals powered by Gemini’s image generation model, Nano Banana. Plus, we’re introducing a new “”Brief”” format for quick summaries.”” / X https://x.com/Google/status/1977769195187126396
💃MPMAvatar: Learning 3D Gaussian Avatars with Accurate and Robust Physics-Based Dynamics”” TL;DR: photorealistic 3D human avatars from multi-view videos with realistic, robust dynamics; tailored Material Point Method for garments + 3DGS for rendering https://x.com/Almorgand/status/1978068062340337881
Nano Banana is now live in Search, so you can use Lens & AI Mode to instantly transform photos with our latest image editing model. Here’s a guide to get you started: 1. Open Lens in the Google app for Android and iOS. 2. Tap the new Create mode — look for the yellow banana 🍌 https://x.com/Google/status/1978857184566837735
Riverflow 1 from Sourceful debuts at #1 in the Artificial Analysis Image Editing Leaderboard for All Listings, using a vision language model trained by Sourceful to allow chain of thought reasoning in combination with a third-party open weights diffusion model Riverflow 1 https://x.com/ArtificialAnlys/status/1978891167795417092
Introducing MAI-Image-1, debuting in the top 10 on LMArena | Microsoft AI https://microsoft.ai/news/introducing-mai-image-1-debuting-in-the-top-10-on-lmarena/
ROGR: Relightable 3D Objects using Generative Relighting”” TL;DR: From posed images (unknown lighting), reconstructs a relightable neural radiance field; can be rendered under any novel environment map without further optimization, on-the-fly relighting and novel view synthesis. https://x.com/Almorgand/status/1976708356149768556
Wan 2.5 debuts at #5 in Text to Video and #8 in Image to Video, marking Alibaba Wan’s strongest video model yet, but breaks from tradition by not being open weights Wan 2.5 is the latest release from Alibaba Wan, representing a substantial upgrade from their Wan 2.2 A14B model. https://x.com/ArtificialAnlys/status/1977910656566489143
Its staggering to see people use dinov2 over and over to fit Imagenet, and choose to not account the pretraining compute in the overall considerations. https://x.com/cloneofsimo/status/1978326330132283891
🎬 Video Arena Update! New model provider @Kling_ai enters the leaderboard with ⚡ Kling 2.5 Turbo 1080p: a big leap from version 2.1, supporting generations up to 1080p resolution! With this model, the cost is as low as $0.15 per 5-second, 1080p video. 🔹 Image-to-Video: tied https://x.com/arena/status/1977836507546423321
We think you’re going to like these updates a bunch 🍌 Introducing six new visual styles for Video Overviews (powered by Nano Banana) and a shorter “”Brief”” option for quick insights. These will start rolling out to Pro users this week and to all users in the upcoming weeks. https://x.com/NotebookLM/status/1977790565983945199
