Today in Generative Media

Attenborough 'disturbed' by AI replica; Authors asked to sell AI rights; AI as clip art

News

• David Attenborough Reacts to AI Replica of His Voice: ‘I Am Profoundly Disturbed’ and ‘Greatly Object’ to It (Variety)

• HarperCollins is asking authors to sell their books to the A.I. woodchipper (A.V. Club)

• More proof that AI images are becoming modern-day clip art (DigitalTrends)

• This is how Tom Hanks was so convincingly de-aged in Here (Creative Bloq)

Software

• Mochi 1 Preview from Genmo (HuggingFace) [Now available in the diffusers library]

• ElevenLabs now offers ability to build conversational AI agents (TechCrunch)

• NotebookLlama: An Open Source version of NotebookLM (GitHub)

• This repository is an implementation of "Vidu4D: Single Generated Video to High-Fidelity 4D Reconstruction with Dynamic Gaussian Surfels", in NeurIPS 2024. (GitHub)

• OmniVision-968M: a new local VLM for edge devices, fast & small but performant (X)

Research

• ProvNeRF: Modeling per Point Provenance in NeRFs as a Stochastic Process (project page)

• Novel View Synthesis with Pixel-Space Diffusion Models (arXiv)

• Efficient Density Control for 3D Gaussian Splatting (arXiv)

Misc

• AI isn't changing what's possible - it's changing the implementation stack. (X)

• Movie producer Jerry Bruckheimer on AI in Hollywood, new movie 'F1' (YouTube)

• It is now possible to generate 3D models using Gaussian Splatting with RunwayML's new Vid2Vid 20-second videos. (X)