Today in Generative Media

Paid AI contributions? AI will trap us like The Truman Show; Reddit tests AI answers

News and Opinion

• Creators Will Soon Contribute To AI “And Get Paid For It”, Google CEO Sundar Pichai Predicts (Deadline)

• AI Will Turn Our Lives into The Truman Show (Scientific American)

• Reddit begins testing AI-powered answers feature to win more users (CNBC)

• If you can make this AI bot fall in love, you could win thousands of dollars (TechCrunch)

• I went to an AI art awards show and it was an eye-opening experience (Creative Bloq)

• The creator of ChatGPT’s voice wants to build the tech from ‘Her,’ minus the dystopia (TechCrunch)

• 15 Times to use AI, and 5 Not to (Ehtan Mollick on Substack)

• How A.I. Can Revive a Love of Learning (New York Times)

• A.I. Will Transform the Global Economy — if Humans Let It (New York Times)

• Watching the Generative AI Hype Bubble Deflate (Harvard Kennedy School)

• OpenAI Sora disappoints (Mehul Gupta on Medium)

Software

• I can now run a GPT-4 class model on my laptop (Simon Willison’s blog)

• With Public Diffusion, Spawning set out to prove that it's possible to train a beautiful AI Art model using only images from the public domain. (X)

• lucataco / real-esrgan-video (Replicate) Real-ESRGAN Video Upscaler

• I'm thrilled to announce that you can now try 4-bit SVDQuant FLUX models in ComfyUI. Our models are approximately 4x smaller and 3x faster than the original 16-bit versions, delivering better efficiency without compromising quality. (X)

Research

• Birth and Death of a Rose (project page)

• Turbo3D: Ultra-fast Text-to-3D Generation (project page)

• Splatter-360: Generalizable 360° Gaussian Splatting for Wide-baseline Panoramic Images (project page)

• PanoDreamer: 3D Panorama Synthesis from a Single Image (project page)

• DNF: Unconditional 4D Generation with Dictionary-based Neural Fields (project page)

• DroneSplat: 3D Gaussian Splatting for Robust 3D Reconstruction from In-the-Wild Drone Imagery (project page)

• MV-Adapter: Multi-view Consistent Image Generation Made Easy (project page)

• AIpparel: A Large Multimodal Generative Model for Digital Garments (project page)

• PBDyG: Position Based Dynamic Gaussians for Motion-Aware Clothed Human AvatarsPermalink (project page)

• [MASK] is All You Need (project page)

• Spatially-Adaptive Hash Encodings For Neural Surface Reconstruction (arXiv)

• GBR: Generative Bundle Refinement for High-fidelity Gaussian Splatting and Meshing (arXiv)

Misc

• issue #24 – Here (befores & afters) Issue #24 of befores & afters magazine covers the visual effects and virtual production of Robert Zemeckis’ Here, including the de-ageing of stars Tom Hanks and Robin Wright, plus the shoot involving a fixed camera and LED wall setup. It also goes into depth on the various machine learning techniques used on the film.