Today in Generative Media

Google's AI needs nuclear power; The Times vs. Perplexity; Rizz's AI dating coach

News and Opinion

• Google becomes the latest tech giant to strike a nuclear-power deal for AI (MarketWatch)

• The New York Times has had it with generative AI companies using its content (TechCrunch)

• How The 5th Most Downloaded Dating App Is Redefining Digital Relationships (Forbes)

• Adobe aims to teach AI, online content skills to 30 million (Fast Company)

• SAG-AFTRA’s Duncan Crabtree-Ireland on Potential for Video Game Publisher Holiday Season Boycott Amid Strike: ‘It’s a Tool That’s in Our Toolkit’ (Variety)

• Taking the Ideas Seriously: Some AI film developments from people who care about doing it right. (Mike Gioia on Substack)

Software

• Generate Video (beta) on Firefly Web App (Adobe blog)

• Excited to share a new product (and a new category for Adobe): Mood-boarding and concepting in the age of AI with Project Concept. (X)

• Adobe’s New Adaptive Profile Uses AI to Non-Destructively Improve Photos in One Click (PetaPixel)

• Websites Now Self-Improve: Our fine-tuned model generates variations and automatically launches A/B tests. (Keak)

• ClipbookLM. Maximize the value of your long form content. Find clips, generate articles, and more.

• Meet Aria: The New Open Source Multimodal AI That's Rivaling Big Tech (Decrypt)

• 🍓 Ichigo: Llama Learns to Talk (Homebrew AI) Inspired by the Chameleon and Llama Herd papers, llama3-s (Ichigo) is an early-fusion, audio and text, multimodal model.

Research

• HART: Efficient Visual Generation with Hybrid Autoregressive Transformer (project page)

• Sana: Efficient High-Resolution Image Synthesis with Linear Diffusion Transformer (project page)

• Semantic Image Inversion and Editing using Stochastic Rectified Differential Equations (project page)

• Refinement of Monocular Depth Maps via Multi-View Differentiable Rendering (project page)

• ElasticTok: Adaptive Tokenization for Image and Video (project page)

• Gaussian Garments: Reconstructing Simulation-Ready Clothing with Photo-Realistic Appearance from Multi-View Video (project page)

• 4-LEGS: 4D Language Embedded Gaussian Splatting (project page)

• DisEnvisioner: Disentangled and Enriched Visual Prompt for Customized Image Generation (project page)

• Look Gauss, No Pose: Novel View Synthesis using Gaussian Splatting without Accurate Pose Initialization (IROS 2024, arXiv)

• Towards Foundation Models for 3D Vision: How Close Are We? (arXiv)

Misc

• Adobe has released a tool that combines Generative Gaussian Splats with a Diffusion layer and it's not all over the internet? WHAT IS GOING ON :D (X)

• AI invented the bottom part of the image (in which it believed that women's shirts should be unbuttoned further, with some tension around the buttons, and revealing a little hint of something underneath). (X)

• How To: Magic Ink Painting Made with Luma DreamMachine (X)