This weekend in Generative Media

Musk shares Harris deepfake; Zuck's genAI gamble; How much will AI affect film?

News

• Elon Musk Shares Manipulated Harris Video, in Seeming Violation of X’s Policies (New York Times)

• Why Zuckerberg’s multibillion-dollar gamble doesn’t just matter to Meta (The Guardian)

• ‘Hold on to your seats’: how much will AI affect the art of film-making? (The Guardian)

• AI v The Mind: Can AI tell better jokes than a human? (BBC)

• AI at the Paris 2024 Olympics: From discovering the next Olympians to an athletes chatbot (EuroNews)

• ChatGPT’s long-awaited new Voice Mode will roll out to Plus subscribers 'next week' (TechRadar)

• SIGGRAPH 2024: what to expect from this year's computer graphics conference (Creative Bloq) AI will be top of the agenda, with a rare joint appearance from Mark Zuckerberg and Nvidia CEO Jensen Huang.

• Seattle’s Hiya acquires deepfake startup Loccus and releases voice-cloning detection tool (GeekWire)

Research

• Alchemist: Parametric Control of Material Properties with Diffusion Models (CVPR 2024, project page). Smoothly editing material properties of objects with text-to-image models and synthetic data (Google Research blog)

• NeuralTO: Neural Reconstruction and View Synthesis of Translucent Objects (SIGGRAPH 2024, ACM)

• Learning Unsigned Distance Functions from Multi-view Images with Volume Rendering Priors (ECCV 2024, project page)

• DreamDissector: Learning Disentangled Text-to-3D Generation from 2D Diffusion Priors (ECCV 2024, project page)

• Temporal Residual Jacobians for Rig-free Motion Transfer (ECCV 2024, project page)

• 3D Gaussian Parametric Head Model (ECCV 2024, project page)

• SparseCraft: Few-Shot Neural Reconstruction through Stereopsis Guided Geometric Linearization (ECCV 2024, project page)

• ViPer: Visual Personalization of Generative Models via Individual Preference Learning (ECCV 2024, project page)

• DreamDissector: Learning Disentangled Text-to-3D Generation from 2D Diffusion Priors (project page)

• Surfel-based Gaussian Inverse Rendering for Fast and Relightable Dynamic Human Reconstruction from Monocular Videos (project page)

• CrowdMoGen: Zero-Shot Text-Driven Collective Motion Generation (project page)

• Generalizable 3D Scene Reconstruction via Divide and Conquer from a Single View (project page)

• Diffree: Text-Guided Shape Free Object Inpainting with Diffusion Model (project page) Demo on HuggingFace.

• GaussianSR: High Fidelity 2D Gaussian Splatting for Arbitrary-Scale Image Super-Resolution (arXiv)