This weekend in Generative Media

AI & screenwriting; Spotify tests genAI voice ads; Apple's risky AI image generators

News

• AI & Screenwriting: Cinematic Dream Or Celluloid Nightmare? (Forbes)

• Spotify announces an in-house creative agency, tests generative AI voiceover ads (TechCrunch)

• Apple’s New AI Image Generators Are Its Riskiest Experiment Yet (Inverse)

• AI is the sixth great revolution in filmmaking (and maybe the most important) (VentureBeat)

• The future of AI in filmmaking - from storyboard to screen (Creative Bloq)

• Welcome to the Reply AI Film Festival, the 1st competition connecting AI-talents with the global cinematic industry. (Reply AI)

Software

• Runway AI just dropped Gen-3 Alpha that generates AI video just from text and images. (X)

• Depth Anything V2 just released, enabling real-time depth estimation directly in your browser with Transformers.js and WebGPU acceleration! (X)

• KREA Video Enhancer is now open to everyone. with this new update, we also support higher resolutions and frame rates of up to 120fps. (X)

Research

• Posterior Distillation Sampling (CVPR 2024, project page)

• Rethinking Score Distillation as a Bridge Between Image Distributions (project page)

• Interpreting the Weight Space of Customized Diffusion Models (project page)

• EMMA: Your Text-to-Image Diffusion Model Can Secretly Accept Multi-Modal Prompts (project page)

• AnyFit: Controllable Virtual Try-on for Any Combination of Attire Across Any Scenario (project page)

• Hallo Hierarchical Audio-Driven Visual Synthesis for Portrait Image Animation (project page)

• M-LRM: Multi-view Large Reconstruction Model (project page)

• Interactive 3D Scene Generation from a Single Image (project page)

• Gaussian-Forest: Hierarchical-Hybrid 3D Gaussian Splatting for Compressed Scene Modeling (arXiv)

• An Image is Worth More Than 16x16 Patches: Exploring Transformers on Individual Pixels (arXiv)

• Rich Human Feedback for Text-to-Image Generation (arXiv)

• Understanding Hallucinations in Diffusion Models through Mode Interpolation (arXiv)

Misc

• Step-by-Step Diffusion: An Elementary Tutorial (arXiv)

• Time Machine (X)

• Cool train ride with LumaLabs (X)

• AI is time-consuming, but by the time I grab a coffee and come back, I have exquisite video animations for my images. (X)

• AI community right now (X)