This Weekend in Generative Media

Adobe is all in on AI; AI makes robocalling worse, AI tidies up Wikipedia

• "Let the robots do the work" – Adobe is unapologetically all-in on AI (Creative Bloq)

• Robocalling is already an issue for Americans. AI is making it worse (CNBC)

• AI tidies up Wikipedia’s references — and boosts reliability (Nature)

• OpenAI debates when to release its AI-generated image detector (TechCrunch)

• Jacques - “Absolve”, by Paul Trillo (Vimeo). The film fuses traditional VFX with a variety of boundary-pushing AI techniques that give way to a new aesthetic and offer an alternative to how visual effects will be approached in the future. Shot breakdown from “Absolve” (X)

• sakemin/musicgen-chord. (Replicate) Generate music restricted to chord sequences and tempo.

• GenAssist: Making Image Generation Accessible (UIST 2023, project page)

• SadTalker: Learning Realistic 3D Motion Coefficients for Stylized Audio-Driven Single Image Talking Face Animation (CVPR 2023, project page) Colab now available.

• 3D-GPT: Procedural 3D Modeling with Large Language Models (project page)

• DynVideo-E: Harnessing Dynamic NeRF for Large-Scale Motion- and View-Change Human-Centric Video Editing (project page)

• Enhancing High-Resolution 3D Generation through Pixel-wise Gradient Clipping (arXiv)

• Using DALL-E 3 to Generate Consistent Character Images: A Detailed Guide (X)