Today in Generative Media

Customers say no to AI; Mixed judgment in Anthropic copyright case; 1984 with LLMs

News and Opinion

• ‘It destroys the purpose of humanity’: Customers are saying no to AI (Washington Post)

• Judge Alsup grants partial summary judgment to Anthropic, ruling training copies were fair use. But Judge rules no fair use in pirated copies of books used to build a central library. They are infringing. (ChatGPT Is Eating the World)

• Elon Musk says xAI will retrain Grok: 'Far too much garbage' (Business Insider)

• 1984, but with LLM’s (Gary Marcus on Substack)

• Mastodon updates its terms to prohibit AI model training (TechCrunch)

• Thanks to social media, consumers have more power than ever. Just wait until generative AI becomes commonplace (Fast Company)

• What happens when AI comes for our fonts? (The Verge)

Software

• Pre-assembled gizmos for ComfyUI within Nuke (GitHub)

• 🤔👀🎬🖥️ Kimi-VL-A3B-Thinking-2506: A Quick Navigation (HuggingFace)

• Claude Code for VSCode (Anthropic)

• Graph-Code: A Multi-Language Graph-Based RAG System (GitHub) An accurate Retrieval-Augmented Generation (RAG) system that analyzes multi-language codebases using Tree-sitter, builds comprehensive knowledge graphs, and enables natural language querying of codebase structure and relationships.

Research

• GenMOJO: Robust Multi-Object 4D Generation for In-the-wild Videos (project page)

• A Feed-Forward 4D Foundation Model for Text-Driven Universal Mesh Animation (project page)

• UniRelight: Learning Joint Decomposition and Synthesis for Video Relighting (nvidia.com)

• WildCAT3D: Appearance-Aware Multi-View Diffusion in the Wild (project page)

• ContentV: Efficient Training of Video Generation Models with Limited Compute (project page)

• Align Your Flow: Scaling Continuous-Time Flow Map Distillation (project page)

• Metropolis-Hastings Sampling for 3D Gaussian Reconstruction (project page)

• UltraZoom: Generating Gigapixel Images from Regular Photos (project page)

• PosterCraft: ✨⭐ Rethinking High-Quality Aesthetic Poster Generation in a Unified Framework (project page)

• Lumina: Real-Time Mobile Neural Rendering by Exploiting Computational Redundancy (arXiv)

Misc

• El Sailon, Santa Fe’s original AI Salon, is going global with its first-ever online webinar: "AI in Film." [June 24, 5pm Pacific time]

• Yes, AI can be efficient. Yes, it can save time. But telling the world it’s “$10 to make a whole movie” is suicidal if you ever want to get hired. (X)

• WAN and Nuke: extraction (YouTube)

• "Vlogwartz" is going insane on TikTok right now. 2 week old account, 15 million views. (X)