Today in Generative Media

NYT inspects OpenAI's code; OpenAI plans to lose $44B; 2 weeks speaking to AI

News

• Why The New York Times' lawyers are inspecting OpenAI's code in a secretive room (Business Insider)

• SF's OpenAI reportedly plans to lose more than $44 billion before it turns a profit (SFGate)

• I've spent the last 2 weeks speaking to AI – here are my 3 biggest takeaways (TechRadar)

• Using ChatGPT to make fake social media posts backfires on bad actors (Ars Technica)

• A Romance Between Two AIs Sparks a New Language—And a Polycule (Decrypt)

Software

• New high quality AI video generator Pyramid Flow launches — and it’s fully open source! (VentureBeat)

  • Pyramidal Flow Matching for Efficient Video Generative Modeling (project page)

  • Code on GitHub

  • Demo on HuggingFace

  • Pyramidal Flow Matching for Efficient Video Generative Modeling [text to video] and [image to video] now on Tost and Runpod (X)

• Google Gemini’s new AI image generator just rolled out to everyone for free – with one annoying limitation (TechRadar)

• FLUX is now much faster on Replicate. Generate images in as little as 300ms from request to response. (X)

• Synchronous API (Replicate)

• Video Background Remover & Changer: You can replace image background with any color, image or video (HuggingFace)

• new video models previewing in Krea (X)

• Readyverse launches Promptopia generative AI creation tool for digital worlds (VentureBeat)

• CogVideoX Factory: Fine-tune Cog family of video models for custom video generation under 24GB of GPU memory (GitHub)

Research

• Depth Any Video with Scalable Synthetic Data (project page)

• RelitLRM: Generative Relightable Radiance for Large Reconstruction Models (project page)

• DifFRelight Diffusion-Based Facial Performance Relighting (project page)

• Fast Feedforward 3D Gaussian Splatting Compression (project page)

• Efficient Perspective-Correct 3D Gaussian Splatting Using Hybrid Transparency (project page)

• DressRecon: Freeform 4D Human Reconstruction from Monocular Video (project page)

• FabricDiffusion: High-Fidelity Texture Transfer for 3D Garments Generation from In-The-Wild Images (project page)

• TextToon: Real-Time Text Toonify Head Avatar from Single Video (project page)

• AniSDF: Fused-Granularity Neural Surfaces with Anisotropic Encoding for High-Fidelity 3D Reconstruction (project page)

• Introducing TouchInsight for detecting touch input from all ten fingers on any surface, purely based on vision-based hand tracking. (X)

Misc

• Welcome to WikiProject AI Cleanup—a collaboration to combat the increasing problem of unsourced, poorly written AI-generated content on Wikipedia. (Wikipedia)

• Behind the product: NotebookLM | Raiza Martin (Senior Product Manager, AI @ Google Labs) (YouTube)

• Create CONSISTENT CHARACTERS for your projects with FLUX! (ComfyUI Tutorial) (YouTube)

• Comparing AI video results:  • Gen-3 • Kling AI 1.5 • Hailuo MiniMax • Luma Dream Machine (X)

• Can you tell which one is me and which one is AI Caleb? (X)

• I found a 3D VR platform that has taken scenes from the mip-NeRF 360 dataset and packaged them up as though they're art pieces. (X)

• Hollywood is the history of technology (X)

• lol (X)