This weekend in Generative Media

Hollywood won't use AI; AI scrapers ignore web norms; Record labels vs. AI audio

News and Opinion

• Fear and Loathing (and Hype and Reality) in Los Angeles: Why the Major Studios Won't Use AI Video Generators Extensively Anytime Soon—And Why That Puts Them in a Bind (The Mediator: Doug Shapiro on Substack)

• Exclusive: Multiple AI companies bypassing web standard to scrape publisher sites, licensing firm says (Reuters)

• Major Labels Weighing Lawsuit Against AI Firms Suno and Udio for Alleged Unlicensed Training (Billboard)

• Mixed feelings: A new study shows creatives are fully embracing AI, but not all are happy about it (Fast Company)

• Neo-Nazis Are All-In on AI (Wired)

• I Am Laura Kipnis-Bot, and I Will Make Reading Sexy and Tragic Again (Wired)

• Apple might partner with Meta on AI (TechCrunch)

• Stability AI appoints new CEO, the Information reports (Reuters)

• OpenAI CTO: AI Could Kill Some Creative Jobs That Maybe Shouldn't Exist Anyway (PC Magazine)

• What to know about Adobe Lightroom's new AI feature (Mashable)

• Generative AI Is Turning Memes Into Nightmares (Forbes)

Research

• EvTexture: Event-driven Texture Enhancement for Video Super-Resolution (ICML 2024, project page)

• COLMAP-Free 3D Gaussian Splatting (CVPR 2024, project page)

• HybridNeRF: Efficient Neural Rendering via Adaptive Volumetric Surfaces (CVPR 2024, project page)

• GLHDR: HDR video reconstruction driven by global to local alignment strategy (Computers & Graphics)

• GradeADreamer: Enhanced Text-to-3D Generation Using Gaussian Splatting and Multi-View Diffusion (project page)

• Splatter a Video: Video Gaussian Representation for Versatile Processing (project page)

• D–NPC: Dynamic Neural Point Clouds for Non-Rigid View Synthesis from Monocular Video (project page)

• Invertible Consistency Distillation for Text-Guided Image Editing in Around 7 Steps (project page)

• CamTrol: Training-free camera control for video generation (project page)

• Wild-GS: Real-Time Novel View Synthesis from Unconstrained Photo Collections (arXiv)

• Adversarial Perturbations Cannot Reliably Protect Artists From Generative AI (arXiv)

• A Note on the new attack paper: "Adversarial Perturbations Cannot Reliably Protect Artists From Generative AI" (Glaze at UChicago)

• Glaze are so dishonest about this because the paper doesn't just show that Glaze doesn't work against anyone determined to bypass it, but it actually doesn't work AT ALL, using fine-tuning scripts the authors found on HuggingFace with no effort to bypass Glaze. (X)

Misc

• When you see generative AI models blatantly reproduce prominent tutorials off the web, does it kill the magic for you? (X)

• I'm kind of sad that the internet is going to be crammed with AI slop from now until the end of time. We only got to have a few decades without slop. (X)

• Ok, Magnific just dropped a bomb with the Relight feature. This kind of feature will be a game-changer for testing advertising environments while maintaining product consistency. (X)