Today in Generative Media

Meta's Llama 3.2, AI voices; Celebs 180 on AI; Zuck: creators overestimate their value

News

• Meta Releases Llama 3.2—and Gives Its AI a Voice (Wired)

• Kristen Bell told Instagram to ‘get rid of AI’ before she became its official voice (The Verge)

• Mark Zuckerberg: creators and publishers ‘overestimate the value’ of their work for training AI (The Verge)

• Meta Spurns EU’s Voluntary AI Safety Pledge Ahead of New Law (Yahoo Finance)

• OpenAI’s chief research officer has left following CTO Mira Murati’s exit (TechCrunch)

• OpenAI planning to become for-profit company, say reports (theguardian.com)

• OpenAI rolls out Advanced Voice Mode with more voices and a new look (TechCrunch)

• OpenAI’s Advanced Voice mode is unavailable in the EU, and now we might know why (TechRadar)

• AI crawlers are hammering sites and nearly taking them offline (Fast Company)

• ‘Robot lawyer’ company faces $193,000 fine as part of FTC’s AI crackdown (The Verge)

Software

• fofr / expression-editor: Quickly edit the expression of a face (Replicate)

• Omni Zero Couples A diffusion pipeline for zero-shot stylized portrait creation (HuggingFace)

• Create high-quality, ultra-realistic characters in just a minute. Try it today at https://hedra.com (X)

• Browse ~4k FLUX LoRAs on Hugging Face! (X)

• This is a ComfyUI node for Clarity AI, the creative image upscaler. It upscales and enhances images to high resolution (GitHub)

• Molmo Multimodal Open Language Model built by Ai2 (Ai2)

• Meta has officially entered the world of Radiance Fields with the announcement of Meta Horizon Hyperscape. Demo available right now in the US with a Quest 3. (X)

Research

• I2VEdit: First-Frame-Guided Video Editing via Image-to-Video Diffusion Models (SIGGRAPH Asia 2024, project page)

• Spline-based Transformers (ECCV 2024, project page)

• DC-Gaussian: Improving 3D Gaussian Splatting for Reflective Dash Cam Videos (NeurIPS 2024, project page)

• Latent Intrinsics Emerge from Training to Relight (NeurIPS 2024, arXiv)

• Zero-Shot Detection of AI-Generated Images (project page)

• TextBoost: Towards One-Shot Personalization of Text-to-Image Models via Fine-tuning Text Encoder (project page)

• MIMO: Controllable Character Video Synthesis with Spatial Decomposed Modeling (project page)

• MaskBit: Embedding-free Image Generation via Bit Tokens (project page)

• GSplatLoc Grounding Keypoint Descriptors into 3D Gaussian Splatting for Improved Visual Localization (project page)

• Plenoptic PNG: Real-Time Neural Radiance Fields in 150 KB (arXiv)

• Human Hair Reconstruction with Strand-Aligned 3D Gaussians (arXiv)

• BetterDepth: Plug-and-Play Diffusion Refiner for Zero-Shot Monocular Depth Estimation (arXiv)

Misc

• Murati is leaving, Schulman left, Karpathy left, Ilya left, Leike left, Brockman is on leave, perhaps a dozen others left, GPT-5 hasn’t dropped, Sora hasn’t shipped, the company had an operating loss of $5b last year, there is no obvious moat, Meta is giving away similar software for free,many lawsuits pending, and people are valuing this company at $150 billion dollars? (X)