This weekend in Generative Media

Aussies find AI 'pillaging'; Canadian news vs. OpenAI; Elon Musk's AI game studio

News

• Amazon, Google and Meta are ‘pillaging culture, data and creativity’ to train AI, Australian inquiry finds (The Guardian)

• Major Canadian News Outlets Sue OpenAI in New Copyright Case (New York Times)

• Elon Musk Says He’ll Make An xAI Game Studio To ‘Make Games Great Again’ (Forbes)

• TV Writers Found 139,000 of Their Scripts Trained AI. Hell Broke Loose (The Ankler)

• Amazon develops video AI model, The Information reports (Reuters)

• ElevenLabs’ new feature is a NotebookLM competitor for creating GenAI podcasts (TechCrunch)

• Bluesky’s open API means anyone can scrape your data for AI training (TechCrunch)

• Google uses search remedies trial to subpoena OpenAI, Perplexity and Microsoft over their generative AI efforts (Digiday)

• Adobe and GenAI (FX Guide)

• Uber for AI labeling (The Verge)

Software

• How to Train Your Mochi: Introducing LoRA fine-tuning. Customize Mochi on a single GPU with just a few videos. Create any effect or create consistent characters. Make Mochi 1 truly yours. (X)

• MeshGen: A Blender addon for generating meshes with AI. (GitHub)

• GaussianAnything: Interactive Point Cloud Latent Diffusion for 3D Generation (GitHub)

Research

• CAT4D: Create Anything in 4D with Multi-View Video Diffusion Models (project page)

• 🧙‍♂️ AnimateAnything: Consistent and Controllable Animation for video generation (project page)

• Diffusion Self-Distillation for Zero-Shot Customized Image Generation (project page)

• HI-SLAM2: Geometry-Aware Gaussian SLAM for Fast Monocular Scene Reconstruction (project page)

• Make-It-Animatable: An Efficient Framework for Authoring Animation-Ready 3D Characters (project page)

• EchoMimic: Towards Striking, Simplified, and Semi-Body Human Animation (project page)

• SmileSplat: Generalizable Gaussian Splats for Unconstrained Sparse Images (project page)

• PreF3R: Pose-Free Feed-Forward 3D Gaussian Splatting from Variable-length Image Sequence (project page)

• NovelGS: Consistent Novel-view Denoising via Large Gaussian Reconstruction Model (arXiv)

• Boost 3D Reconstruction using Diffusion-based Monocular Camera Calibration (arXiv)

• Representing Long Volumetric Video with Temporal Gaussian Hierarchy (ACM Transactions on Graphics)

• DreamUDF: Generating Unsigned Distance Fields from A Single Image (ACM Transactions on Graphics)

Misc

• Gen-3 Alpha Expand Video from RunwayML is one of the most impressive AI models I've seen recently. (X)

• And we’re back for the Annual Gen AI Thanksgiving Day Parade (X)

• We took our new 3D sculpt model, generated a blender scene and then asked a video-to-video model to render. (X)

• This is a slippery slope to nonconsensual explicit deepfakes, and it’s kids in schools who’ll bear the brunt — not Taylor Swift. The flood of explicit content on demand is a nightmare in the making. I don’t envy this generation dealing with the fallout. (X)