This weekend in Generative Media

Sand's MAGI-1 censors; Cheehoo gets $10M for AI animation; AI vs. the app economy

News and Opinion

• A Chinese AI video startup appears to be blocking politically sensitive images (TechCrunch)

• AI animation startup Cheehoo lands $10M funding from Greycroft (TechCrunch)

• AI could make apps irrelevant, says Meta's tech chief (Business Insider)

• Adobe releases ‘created without generative AI’ tag to label human-generated art (Fast Company)

• It's becoming less taboo to talk about AI being 'conscious' if you work in tech (Business Insider)

• Will the Humanities Survive Artificial Intelligence? (The New Yorker)

• AI might be coming for your job, but it can't take your creativity (Creative Bloq)

• NFL, Adobe team up for AI-powered league and fan-generated content (MarketingDive)

• I compared Adobe’s new Firefly Image Model 4 to ChatGPT’s image generator, and it’s like they went to the same art school (TechRadar)

Software

• MVSAnywhere: A multi-view stereo depth estimation model which works anywhere, in any scene, with any range of depths (GitHub)

• UniAnimate-DiT is based on a state-of-the-art DiT-based Wan2.1-14B-I2V model for consistent human image animation. (GitHub)

• Shape-Guided Clothing Warping for Virtual Try-On (GitHub)

• Tina: 1.5B model that beats other 1.5B reasoning models trained with full-parameter RL, at just $9 total cost. (X)

Research

• Step1X-Edit: A Practical Framework for General Image Editing (project page)

• TAPIP3D: Tracking Any Point in Persistent 3D Geometry (project page)

• PartField: Learning 3D Feature Fields for Part Segmentation and Beyond (project page)

• VideoSPatS: Video SPatiotemporal Splines for Disentangled Occlusion, Appearance and Motion Modeling and Editing (project page)

• FaceCraft4D: Animated 3D Facial Avatar Generation from a Single Image (arXiv)

• ePBR: Extended PBR Materials in Image Synthesis (arXiv)

• Token-Shuffle: Towards High-Resolution Image Generation with Autoregressive Models (arXiv)

• ViSMaP: Unsupervised Hour-long Video Summarisation by Meta-Prompting (arXiv)

Misc

• AI and CGI can work together ! Not against ! I made all this using ComfyUI with Wan 2.1 14B model on a H100. (LinkedIn)