⚡️ Revolutionizing Character Rendering

Good Morning. The era of flat screens and limited camera angles could soon be a thing of the past, thanks to an artificial intelligence method that synthesizes lifelike images of characters from fresh perspectives on the fly. Let’s dive in.

Today’s Highlights:

• Stability AI's 'smol' language model enters the chat

• Google's Gemini AI faces scrutiny over misleading demo

• EU lawmakers take a break from marathon AI regulation talks

DEEP DIVE

GPS-Gaussian: Revolutionizing Character Rendering with AI

The Harbin Institute of Technology and Tsinghua University have unveiled GPS-Gaussian, transforming the way characters are rendered in real-time through a new artificial intelligence method for synthesizing novel views of characters.

GPS-Gaussian is a groundbreaking approach that uses large 3D human scan models to create a generalizable Gaussian model to rapidly depict human appearances. It overcomes the limitations of previous novel view synthesis (NVS) techniques, which struggle under sparse-view camera settings and often require dense input views or accurate proxy geometry.

How real can it get?

• It's about more than just avoiding choppy holograms on stage. We're talking holographic communication and ultra-immersive sports broadcasting where the action feels inches away.

• The solution combines Gaussian splatting with neural networks, emerging as a far more efficient contender against slow-rendering techniques like Neural Radiance Fields (NeRF).

The methodological twist is a set of 2D Gaussian parameter maps that are elevated into 3D space, neatly sidestepping the hefty computational burden of 3D operators.

Despite the complexities of significant self-occlusions inherent in human figures, GPS-Gaussian demonstrates its prowess by producing 2K novel view images at a remarkable frame rate of over 25 FPS using just a high-end graphics card. The method assures instant rendering of unseen characters without the need for optimization or fine-tuning, thanks to its broad generalizability and swift rendering.

Could this be the future of real-time character rendering in holographic communication?

Keep reading here.

PUNCHLINES

To The Moon: GOOGL rallies 5% after Google announces Gemini AI model.

In Cells We Trust: Seattle biotech hub pursues ‘DNA typewriter’ tech with $75M from tech billionaires.

Gold Rush: The Caribbean island of Anguilla is raking in millions of dollars every month from the .ai domain.

Check Mate: ChatGPT beaten by 1960s computer program ELIZA in Turing test study.

Frontline Fortune: SF startup MaintainX raises $50M to bring AI to industrial operations.

TLDR

Stability AI's new 'smol' language model, Zephyr 3B: Stability AI introduces a 3B parameter LLM, StableLM Zephyr 3B, with a focus on chat applications. Optimized for Q&A and following instructions, it runs on diverse hardware and trumps larger models in specific benchmarks. This comes alongside other creations like StableCode and Stable Audio, as part of their aim to make generative models widely accessible.

Google's Gemini AI presentation raises doubts: Google's showcase of its Gemini AI was discovered to be misleading, as the "Hands-on with Gemini" demo seems to have used edited footage instead of live responses. The exposé casts a shadow on Google's transparency and the actual capabilities of Gemini.

EU AI Regulation Talks Hit Pause: Exhausted after 22 hours, the EU paused critical discussions on AI rules set to pick up again on Friday. Key issues include how to regulate foundational AI and facial recognition. Agreement is essential for the AI Act, which, if approved by the European Parliament, would be effective no earlier than 2025.

Microsoft to expand LLM offerings beyond OpenAI: Microsoft hints at introducing new LLMs to Azure AI, aiming to give customers more options beyond its OpenAI partnership. This comes as enterprises increasingly adopt AI and amid criticisms for relying heavily on a single model provider.

TRENDING TOOLS

🎧 Fathom: Explore podcasts with AI-assisted search and interactive transcripts

💬 Simple Analytics AI: Engage in conversations with your website analytics

🚀 Streak: AI-powered assistance for your CRM workflows

🎼 Music AI: Access state-of-the-art music APIs and audio solutions on one platform

🤖 Kommunicate GenAI: Integrate generative AI bots seamlessly across communication channels

That’s all for today—if you have any questions or something interesting to share, please reply to this email. We’d love to hear from you!

P.S. If you want to sign up for the Supercharged newsletter or share it with a friend, you can find us here.