Today in Generative Media

AI can be fun; Stability's upscaler; Gen-1 ups output to 15s

• What gets lost in the AI debate: It can be really fun. (Vox)

• Stability AI’s New Upscaler Can Quadruple the Resolution of an Image (PetaPixel). Here’s Stability’s blog post.

• RunwayML’s Gen-1 now supports 15 second generations. (Twitter)

• This startup wants to train art-generating AI strictly on licensed images. (TechCrunch)

• Spotify confirms it’s working with labels to quell copyright threat from AI music. (Music Business Worldwide)

• Mark Zuckerberg says Meta wants to ‘introduce AI agents to billions of people’ (The Verge)

• MosaicML trained Stable Diffusion from scratch for $50K.

• DeepFloyd IF (GitHub) is a novel state-of-the-art open-source text-to-image model with a high degree of photorealism and language understanding.

• LLaVA: Large Language and Vision Assistant (project page, with links to arXiv, code, demo, & dataset)

• Pointersect: Neural Rendering with Cloud-Ray Intersection (Apple ML)

• Pepperoni Hug Spot - AI TV commercial 👀 (Reddit)

• Can’t wait to see all the AI progress… (Twitter)