Image created with Flux Pro Ultra. Image prompt: A Minecraft screenshot showing players wearing cubic headsets interacting with floating holographic block structures, split between real and virtual worlds, with “ARVR” written in pixelated Minecraft font across the top
AI Avatars Escape the Uncanny Valley | Andreessen Horowitz https://a16z.com/ai-avatars/
“OmniSVG announced on Hugging Face A Unified Scalable Vector Graphics Generation Model https://x.com/_akhaliq/status/1909935266069938653
DiTaiListener https://cv.maxi.su/DiTaiListener/
Introducing Sponsored AI Lenses https://newsroom.snap.com/sponsored-ai-lenses
“Introducing Mocha ✨ We’re incredibly excited to launch the absolute best way to build full-stack web apps. No code. No templates. Just describe what you want and watch it come to life. https://x.com/nichochar/status/1906748998322655529
“@AgentOpsAI 13/ Create Multilingual 2D Digital Humans for Enterprise Hosted by: Rochelle Pereira, Sr. Director of Engineering, NVIDIA Ragav Venkatesan, Principal Software Engineer, NVIDIA Learn about NVIDIA NIMā„¢ microservices for secure, high-performance AI deployment across various” / X https://x.com/AtomSilverman/status/1907898487154626722
“Meta announced MoCha, an AI that turns speech and text into movie-grade talking/singing character animations It enables multi-character conversations with turn-based dialogue generation, and near-perfect lip-sync https://x.com/adcock_brett/status/1908913575403323545
How Sphere’s new ‘The Wizard of Oz’ experience is coming to life with AI https://blog.google/products/google-cloud/sphere-wizard-of-oz/
“This can fix me.. Transition beyond Gaussian, its now repeating pattern, just like the Generator matching before https://x.com/cloneofsimo/status/1910097234538176650
“Harvard and University of Toronto researchers dropped an AI for 3D medical image & video segmentation Built atop Segment Anything Model 2.1, MedSAM2 generalizes across organs, modalities, and pathologies with an 85%+ reduction in annotation costs https://x.com/rowancheung/status/1909496858122060236
“@AgentOpsAI 5/ Grounding LLMs in Reality: Enhancing SAP’s Document Grounding Hosted by: Atreya Biswas, Lead Architect, SAP Jia Xiang Lim, SAP Document Grounding, SAP’s retrieval-augmented generation (RAG) solution for unstructured data, addresses this challenge by enabling contextual and” / X https://x.com/AtomSilverman/status/1907898216043130998
“At the system level, TPU wins hard – ICI scales to 9,216 chips. However, the 3D torus topology limits programmability. Compare GB200: only 72 chips, but on a switched network. It’s a much more flexible topology, but the switches consume power, and you have to lean on the” / X https://x.com/itsclivetime/status/1910026078405750792
“We’ve made huge improvements to Imagen 3: our highest quality text-to-image model. It’s capable of generating visuals with better detail, richer lighting and fewer distracting artifacts. 🖼️ In #VertexAI, it’s also easier than ever to remove unwanted objects, blemishes, or https://x.com/GoogleDeepMind/status/1910009261075357902
“Ripping off someone else’s work shot-for-shot using Gen AI and passing it off as your own is not cool. The 3D camera intro shot was also a blatant from Miklas aka Sur Render who brought this to my attention on threads.” / X https://x.com/bilawalsidhu/status/1907764126719336826
“Vancouver-based Sanctuary AI demoed their efforts with sim-to-real transfer In this clip, their hydraulic hand can be seen executing an in-hand reorientation policy under a 500g load, trained in simulation https://x.com/adcock_brett/status/1908913441009451137
“New Algorithm from China’s ByteDance: VAPO Value-based Augmented Proximal Policy Optimization framework for reasoning models. VAPO refines value-based learning so long reasoning sequences finally become manageable. Tames lengthy outputs through smart advantage calibration for https://x.com/rohanpaul_ai/status/1909581398991946230
“Optimus job opening for Technical Animator was posted last week: “Optimus simulation team is responsible for building core components to visualize all aspects of Optimus… for understanding, testing bot functionality, and generating synthetic data for training deep NNs.” https://x.com/TheHumanoidHub/status/1909312550564774249
“World models are a puzzle piece for the future of AI They are gen AI systems that learn simulation of real environments to: – predict future states – simulate actions internally – support planning and decision-making All inside the “mental model” without constant real-world https://x.com/TheTuringPost/status/1910467892929585663
“Just dropped a Severance inspired app using @lovable_dev fake tasks, corporate satire, eerie vibes. Built it in a day. Welcome to the Lumon Productivity Portal. “Embrace productivity. Refine your reality.” #buildinpublic #nocode #severance https://x.com/aiwhit781/status/1906718475613000033
“Scanning a space with 3D Gaussian splatting (in this case my dining room) gives you some wild superpowers. Change FOV, pull off impossible camera moves, get x-ray vision through walls — and even classify everything semantically. Reality becomes editable. And queryable. https://x.com/bilawalsidhu/status/1907963498404974832
“Tesla made Optimus walk better with reinforcement learning in simulation and is now working to apply the same technique to manipulation tasks. https://x.com/TheHumanoidHub/status/1909312554838769813
“I am curious about this. GPT-4o multimodal does a better job than Gemini multimodal with images, but it also alters details each time, while Gemini does not. This is especially true of faces and lighting, making image consistency hard. Is that a deliberate choice, OpenAI folks?” / X https://x.com/emollick/status/1909463868230729887
“Tesla engineers talked about the in-house Autopilot simulation pipeline built on top of Unreal Engine on AI Day 2022, and also showed early work in simulating Optimus using the Autopilot simulator. https://x.com/TheHumanoidHub/status/1909312552636760391
“Alibaba just released LAM on Hugging Face Large Avatar Model for One-shot Animatable Gaussian Head https://x.com/_akhaliq/status/1910259092972589432
Scene-Centric Unsupervised Panoptic Segmentation https://visinf.github.io/cups/
[2504.07095] Neural Motion Simulator: Pushing the Limit of World Models in Reinforcement Learning https://arxiv.org/abs/2504.07095
FantasyTalking: Realistic Talking Portrait Generation via Coherent Motion Synthesis https://fantasy-amap.github.io/fantasy-talking/
“Discrete diffusion is winning over AR recently: LLaDA-8B, Dream-7B, UniDisc I could be coping but maybe diffusion isnt dead yet UniDisc did comparison and found diffusion model to perform better than autoregressive, as they were ‘less efficient at training time’ but ‘more https://x.com/cloneofsimo/status/1908148670098538645
Waymo may use interior camera data to train generative AI models, but riders will be able to opt out | TechCrunch https://techcrunch.com/2025/04/08/waymo-may-use-interior-camera-data-to-train-generative-ai-models-sell-ads/
WHAMM! Real-time world modelling of interactive environments. – Microsoft Research https://www.microsoft.com/en-us/research/articles/whamm-real-time-world-modelling-of-interactive-environments/
