Today in Generative Media

Wired tests ethical genAI music; Google's genAI video presos; IATSE AI protections

News and Opinion

• The New ‘Ethical’ AI Music Generator Can’t Write a Halfway Decent Song (Wired)

• Google tests out Gemini AI-created video presentations (The Verge)

• How IATSE’s AI Protections Are More ‘Evolved’ Than the Other Guilds’ (IndieWire)

• These IATSE Artists Are Voting ‘No’ on Their Next Contract, and AI Is to Blame (The Wrap)

• Job seekers are using controversial AI 'teleprompter' apps to 'crush' their interviews (Business Insider)

• Seeing is not Believing with AI (Mike Gioia on Substack)

• Mango creates the first campaign generated by artificial intelligence for its Teen line (Mango Fashion Group)

Research

• Mile-High AI: NVIDIA Research to Present Advancements in Simulation and Gen AI at SIGGRAPH (NVIDIA blog)

• Dynamic Neural Radiance Field From Defocused Monocular Video (ECCV 2024, GitHub)

• RodinHD: High-Fidelity 3D Avatar Generation with Diffusion Models (ECCV 2024, project page)

• TCAN: Animating Human Images with Temporally Consistent Pose Guidance Using Diffusion Models (project page)

• RRM: Relightable assets using Radiance guided Material extraction (CGI 2024, arXiv)

• PanDORA: Casual HDR Radiance Acquisition for Indoor Scenes (project page)

• CopyBench: Measuring Literal and Non-Literal Reproduction of Copyright-Protected Text in Language Model Generation (project page)

• Tailor3D: Customized 3D Assets Editing and Generation with Dual-Side Images (project page)

• Read, Watch and Scream! Sound Generation from Text and Video (project page)

• Mira: A Mini-step Towards Sora-like Long Video Generation (project page)

• MiraData: A Large-Scale Video Dataset with Long Durations and Structured Captions (arXiv)

• Solutions to Deepfakes: Can Camera Hardware, Cryptography, and Deep Learning Verify Real Images? (arXiv)

Misc

• Here is a manual [for LLMs] in two tweets: (X)

• Unsampling for AnimateDiff/Hotshot - An Inner-Reflections Guide (X)

• Build an AI Movie Production Agent with Claude Sonnet 3.5 in just 30 lines of Python Code (step-by-step instructions): (X)

• New way to navigate latent space. It preservers the underlying image structure and feels a bit like a powerful style-transfer that can be applied to anything. (X)

• Catwalk (X)