Today in Generative Media

TikTok's AI influencers; Runway's Gen-3 video model; Adobe Firefly gets competition

News

• TikTok is getting into the murky business of AI-generated influencer ads (Business Insider)

• Runway unveils new hyper realistic AI video model Gen-3 Alpha, capable of 10-second-long clips (VentureBeat)

• Runway Gen-3 Can Make AI Videos of ‘Photorealistic Humans’ (PetaPixel)

• Adobe’s Firefly AI is getting competition at the worst time (DigitalTrends)

• Mira Murati and David Droga on why creatives should—and shouldn’t—worry about AI (Fast Company)

Research

• Generating audio for video (Google DeepMind)

• HiFi4G: High-Fidelity Human Performance Rendering via Compact Gaussian Splatting (CVPR 2024, project page)

• DiffuScene: Denoising Diffusion Models for Generative Indoor Scene Synthesis (CVPR 2024, project page)

• Instruct 4D-to-4D: Editing 4D Scenes as Pseudo-3D Scenes Using 2D Diffusion (CVPR 2024, project page)

• CoPoNeRF: Unifying Correspondence, Pose and NeRF for Pose-Free Novel View Synthesis from Stereo Pairs (CVPR 2024, project page)

• DSINE: Rethinking Inductive Biases for Surface Normal Estimation (CVPR 2024, project page)

• Neural Geometry Fields for Meshes (SIGGRAPH 2024, project page)

• GGHead Fast and Generalizable 3D Gaussian Heads (project page)

• StableMaterials: Enhancing Diversity in Material Generation via Semi-Supervised Learning (project page)

• Eye-for-an-eye: Appearance Transfer with Semantic Correspondence in Diffusion Models (project page)

• Ctrl-X: Controlling Structure and Appearance for Text-To-Image Generation Without Guidance (project page)

• FontStudio: Shape-Adaptive Diffusion Model for Coherent and Consistent Font Effect Generation (project page)

• GS-IR: 3D Gaussian Splatting for Inverse Rendering (project page)

• Layered Image Vectorization via Semantic Simplification (project page)

• MultiDiff: Consistent Novel View Synthesis from a Single Image (CVPR 2024, PDF)

• Unified Gaussian Primitives for Scene Representation and Rendering (arXiv)

Misc

• Well, that was quick! LumaLabs [Dream Machine] now supports extending your generations with a simple button push right inside the interface. (X)

• Here’s hoping the AI tanker ships actually release Sora and Veo level tech into the wild vs teasing collabs with A-list celebs (X)

• Once AI video is indistinguishable from film -- and it will be -- audiences won't care. They just won't. (X)

• And to think that this video is 100% generated by AI, it's total madness the news we have about AI videos these days. Runway Gen-3 (X)

• From sketch to animation with AI.. an early glimpse of future workflows. Still with too little actual control, but at least we've moved beyond the text-field input :D (X)

• Sometimes there are cool DreamMachine fails. (X)