Today in Generative Media

Hollywood backlash boosts AI bills; Publishers get cozy with AI; AI gets theatrical

• Washington pushes for new AI rules protecting artists as Hollywood backlash mounts (Yahoo Finance)

• This Week in AI: OpenAI and publishers are partners of convenience (TechCrunch)

• Publishers are already using way too much AI. (Literary Hub)

• AI is Getting Theatrical (Fast Company)

• Make a song from any sound. Coming soon 🎧 (X)

• Diffusion Models (Andrew Chan) Notes on the theory behind models like Stable Diffusion and their applications.

• ReVideo: Remake a Video with Motion and Content Control (project page)

• Pandora: Towards General World Model with Natural Language Actions and Video States (project page)

• Improved Distribution Matching Distillation for Fast Image Synthesis (project page)

• DoGaussian: Distributed-Oriented Gaussian Splatting (project page)

• Images that Sound: Composing Images and Sounds on a Single Canvas (project page)

• Effective large language model adaptation for improved grounding (project page). We introduce AGREE, a learning-based framework that enables LLMs to provide accurate citations in their responses, making them more reliable and increasing user trust.

• Best Practices and Lessons Learned on Synthetic Data for Language Models (arXiv)

• Refusal in LLMs is mediated by a single direction (project preview)

• Comfy3D Update: Integrated Zero123++, Integrated InstantMesh, Add 3D Mesh Orbit Renderer (X)

• A simple but very practical prompt for working on character concepts of all kinds. (X)

• Golden Gate Claude's reductio ad pontem (X)

• most intelligence supply will probably be consumed by use cases we don't foresee yet (X)

• If you want to help the climate, consider environmentally-friendly sourcing for your art by getting it from AI models. (X) [satire?]