Today in Generative Media

Instagram teases AI video editing; Decart raises $32M; AI copyright suits, visualized

News

• Instagram teases AI editing tools that will completely reimagine your videos (The Verge)

• Decart nabs $32M at $500M+ valuation to build AI tech and ‘open world’ apps (TechCrunch)

• Every AI Copyright Lawsuit in the US, Visualized (Wired)

• Generative AI Still Needs to Prove Its Usefulness (Wired)

• Adobe Lightroom's new photo culling tool could soon choose your best photos for you during import (Yahoo Tech)

Software

• First Update of LTX-Video is Here! (X)

• Google releases its own ‘reasoning’ AI model (TechCrunch)

  • Gemini 2.0 Flash Thinking Mode (Google)

• MarkItDown is a utility for converting various files to Markdown (e.g., for indexing, text analysis, etc). (GitHub)

• Finally, a Replacement for BERT (HuggingFace)

Research

• INFP: Audio-Driven Interactive Head Generation in Dyadic Conversations (project page)

• SqueezeMe: Efficient Gaussian Avatars for VR (project page)

• Real-time Free-view Human Rendering from Sparse-view RGB Videos using Double Unprojected Textures (project page)

• Prompting Depth Anything for 4K Resolution Accurate Metric Depth Estimation (project page)

• AniDoc: Animation Creation Made Easier (project page)

• CATSplat: Context-Aware Transformer with Spatial Guidance for Generalizable 3D Gaussian Splatting from A Single-View Image (arXiv)

• TraSCE: Trajectory Steering for Concept Erasure (arXiv)

Misc

• Welcome to the Machine Learning Road Map: A succinct guide to learning ML fundamentals for free! (GitHub)

• Discover and hire the creatives shaping the future of media. All of the world's top Al experts. All in one place. (Runway Talent Network) A new media platform to discover and hire the creatives and companies actively using AI to shape the future of media.

• I generally don't like being controversial, but I am sad that this release is completely misleading. It is NOT generative AI. It's a cool **physics simulation library**. The generative part isn't there, and the referenced paper is an LLM that calls the right assets and code. (X)