Adam here! This week, we explore Runway ML's Gen-3 Alpha for video generation and Meta's Meta 3D Gen for text-to-3D asset creation. We also feature the first brand film created with Sora, showcasing the story of Toys"R"Us and Geoffrey the Giraffe.

In Tweets of the Week, Min Choi demonstrates AI applications with Claude 3.5 Sonnet, and Dylan Field addresses concerns about Figma's Make Design feature.

Check out More AI in Action for interactive learning with Claude and meme animation with Kling AI.

🚀 Introducing Gen-3 Alpha: Elevating Video Generation Technology

Runway ML has released Gen-3 Alpha, enhancing high-fidelity, controllable video generation. This model offers significant improvements in fidelity, consistency, and motion over Gen-2. Key features include Text to Video, Image to Video, advanced control modes like Motion Brush, and photorealistic human generation.

Gen-3 Alpha is built on a new infrastructure for large-scale multimodal training, ensuring detailed, realistic video outputs. Safeguards include an improved visual moderation system and C2PA provenance standards.

While Sora isn't here yet, explore the innovative capabilities of Gen-3 Alpha now.

Need help with getting started? Check Getting Started with Gen-3 Alpha.

🖼️ New Research: Meta 3D Gen Delivers High-Quality Text-to-3D Generation

Meta has introduced Meta 3D Gen, a new system for text-to-3D generation that produces high-quality geometry and textures. This system generates 3D assets with high-resolution textures and material maps end-to-end, achieving results that are superior to previous solutions, and at speeds 3-10 times faster.

• High-Quality 3D Assets: Generates assets with detailed textures and material maps.

• Fast Performance: Produces high-quality 3D shapes and textures in under a minute.

• Advanced Rendering: Supports physically-based rendering (PBR) for realistic relighting.

• Generative Retexturing: Allows retexturing of generated or artist-created 3D shapes using additional text inputs.

Meta 3D Gen combines two key components: Meta 3D AssetGen for generating 3D models from text and Meta 3D TextureGen for high-quality texture generation. This integration enables Meta 3D Gen to represent 3D objects in view space, volumetric space, and UV (texture) space, enhancing the fidelity and quality of the generated assets.

The system generates several views of the object, reconstructing colors, metallicity, and roughness in 3D, with a final texture refinement step significantly improving sharpness and details.

Meta's research shows that Meta 3D Gen outperforms industry baselines in terms of prompt fidelity and visual quality for complex textual prompts, making it a leading solution in text-to-3D generation.

🎥 Watch the Story of a Dream Come True: The Creation of Toys"R"Us and Geoffrey the Giraffe

The first brand film created with Sora, a new technology from OpenAI, has been released. This film, produced in partnership with Native Foreign, tells the story of Charles Lazarus, the founder of Toys"R"Us, and the beloved mascot Geoffrey the Giraffe.

• New Technology: Sora can generate realistic scenes and multiple characters from text instructions, creating up to one-minute-long videos.

• Historical Reenactment: The film brings to life young Charles Lazarus and his dreams for the Toys"R"Us brand in the early 1930s.

• Beloved Mascot: Experience the story of Geoffrey the Giraffe, an enduring symbol of Toys"R"Us.

This film shows the potential of Sora for storytelling and video creation. While Sora's broader release is awaited, this project offers a glimpse into the future of content creation.

🐥 Tweets of the Week

Claude 3.5 Sonnet Over A Week: Min Choi (@minchoi) shares several AI applications, including a music generation app using genetic algorithms, quick illustration of concepts, passing the AI Mirror Test, creating interactive prompt engineering cheat sheets, and generating 3D simulations from pictures.

Figma's Make Design Feature Update: Dylan Field (@zoink) addresses concerns about Figma's new Make Design feature, clarifying that it is not trained on Figma content or community files. He explains how the feature works using off-the-shelf LLMs and commissioned design systems, acknowledges the issues found, and announces the temporary disabling of the feature for further QA.

🎬 More AI in Action this Week

💡 Interactive Learning with Claude: Discover how to create interactive learning experiences with Claude in seconds. By uploading Japanese lessons and prompting for an interactive dashboard and quiz, you get a professional and neat interface. You can continually improve it by adding new lessons over time.

adamstewartmarketing

View more on Instagram

🚀 Bringing Memes to Life with Kling AI: Kling AI is now animating memes, with early examples including the distracted boyfriend meme, Justin Timberlake's mugshot, and Mona Lisa in between posing. Watch these static images come alive with AI.

Discover these AI Tools and Apps

1. Claude - Claude is an Artificial Intelligence, trained by Anthropic using Constitutional AI to be safe, accurate, and secure — the best assistant for you to do your best work.

2. Kling AI - Kling is a text-to-video AI tool. It empowers users with remarkable video generation capabilities, allowing them to effortlessly and efficiently create artistic videos.

3. Runway Gen-3 - A new frontier for high-fidelity, controllable video generation.

4. Wegic.AI - An AI web designer & developer by your side.

5. PlusDocs - Generate AI presentations and edit slides with AI.