Today in Generative Media

Pika raises $80M; Justice Dept. AI misuse worries; Nvidia CEO: AI will make games

• AI video start-ups race ahead as Big Tech competition looms (Washington Post)

• Justice Department's 'deepfake' concerns over Biden interview audio highlights AI misuse worries (ABC News)

• Nvidia's Jen-Hsun Huang reflects on how AI already creates pixels and entire frames, before saying that 'games will be generated with AI' (PC Gamer)

• ‘All eyes on Rafah’ is the Internet's most viral AI image. Two artists are claiming credit (NPR)

• Companies like Google and OpenAI are pillaging the internet and pretending it’s progress (BGR)

• NeRF volumetric VR video for Quest and Vision Pro has arrived, and it's open source (Lifecast)

• Omost is a project to convert LLM's coding capability to image generation (or more accurately, image composing) capability. (GitHub)

• Diffusers v0.28.1: HunyuanDiT andTransformer2D model class variants (GitHub)

• Towards 3D Vision with Low-Cost Single-Photon Cameras (CVPR 2024, project page)

• HyperDreamBooth: HyperNetworks for Fast Personalization of Text-to-Image Models (CVPR 2024, project page)

• Learning Temporally Consistent Video Depth from Video Diffusion Priors (project page)

• Enhancing Temporal Consistency in Video Editing by Reconstructing Videos with 3D Gaussian Splatting (project page)

• CamCo: Camera-Controllable 3D-Consistent Image-to-Video Generation (project page)

• RaDe-GS: Rasterizing Depth in Gaussian Splatting (project page)

• UniAnimate: Taming Unified Video Diffusion Models for Consistent Human Image Animation (project page)

• Diffusion On Syntax Trees For Program Synthesis (project page)

• SpatialRGPT: Grounded Spatial Reasoning in Vision Language Model (project page)

• FastLGS: Speeding up Language Embedded Gaussians with Feature Grid Mapping (project page)

• SoundCTM: Uniting Score-based and Consistency Models for Text-to-Sound Generation (project page)

• Kaleido Diffusion: Improving Conditional Diffusion Models with Autoregressive Latent Modeling (arXiv)

• ZeroSmooth: Training-free Diffuser Adaptation for High Frame Rate Video Generation (arXiv)

• Tetrahedron Splatting for 3D Generation (arXiv)

• Guiding a Diffusion Model with a Bad Version of Itself (arXiv)

• Our AI-powered storytelling platform just got even more powerful...We're proud to present Visions, our new expansion pack designed to elevate how you visualize and communicate your next creative idea. (X)

• Skybox AI's incredible realism coming with Model 3.1 is showcased perfectly in the new style Cinematic Realism. Use it for free, coming June 11. (X)

• I'm working on making it easier to publish ComfyUI workflows as Replicate models. (X)