# Generative Designing is The Biggest Update To Your Workflow | Gemini 3 Source: https://www.youtube.com/watch?v=XP6TJRbOdGo Recap page: https://rapidrecap.app/video/XP6TJRbOdGo Generated: 2026-01-12T14:32:28.552+00:00 --- ## Quick Overview Google's latest AI tools, including Gemini 3, Veo, Imagen 3, and experimental platforms like Whisk and Flow, dramatically enhance creative workflows by automating complex tasks such as image generation, video creation, and scroll-driven web animations, often requiring minimal prompting compared to previous methods. **Key Points:** - Gemini 3 is highlighted as Google's most intelligent model, powering new capabilities like generative interfaces and complex task completion via Gemini Agent. - Google introduced Flow, an AI filmmaking tool custom-designed for its advanced models (Veo, Imagen, Gemini) to help storytellers create cinematic clips. - Whisk, an experimental tool powered by Nano Banana, allows users to remix reference images by selecting subject, scene, and style, offering more control than simple text-to-image generation. - Veo 3.1 Fast video generation costs $0.15 per second in the Paid Tier, while the Standard version costs $0.40 per second, and neither is available on the Free Tier. - The video demonstrates using Veo's Frames to Video feature to create a complex camera deconstruction animation by providing distinct start and end frames. - The author used Claude 3 with XML tags to structure a complex prompt for integrating a complex 'Background Paths' component into a Next.js project, achieving high fidelity results. - The use of WebP format for hero section animations is advocated over MP4 due to its smaller file size, faster load times, and superior scroll interaction mapping capabilities. ![Screenshot at 00:01: 48:Demonstration of the Whisk tool, where a complex prompt combining image references \(subject, scene, style\) is used to generate a highly detailed, stylized sequence of images in one go.](https://ss.rapidrecap.app/screens/XP6TJRbOdGo/00-00-01.jpg) **Context:** This video explores the recent advancements in Google's generative AI suite, focusing heavily on the integration and practical application of new tools like Gemini 3, Imagen 3, Veo (for video), and the experimental platform Whisk. The presenter demonstrates how these tools streamline creative and development workflows, from generating complex images and animations to providing detailed code scaffolding for web development projects, often contrasting the new, highly controllable methods with older, less efficient techniques. ## Detailed Analysis The video showcases several powerful updates and new AI tools from Google, starting with the announcement of Gemini 3, described as their most intelligent model, featuring generative interfaces and the Gemini Agent for complex task execution. Next, the presentation pivots to image and video generation experiments. Whisk, an experimental tool, is introduced, which excels at image remixing by allowing users to combine reference images for subject, scene, and style, providing granular control over the output. The video then dives into Google Flow, a new AI filmmaking tool built specifically for Google’s advanced models (Veo, Imagen, Gemini), designed to create cinematic clips with consistent characters and physics-based realism. The presenter demonstrates Veo’s 'Frames to Video' feature by creating a complex, seamless animation of a Sony camera assembling and disassembling itself based on start and end frames. The pricing structure for Veo is detailed, showing that video generation is exclusively on paid tiers ($0.15/sec for Fast, $0.40/sec for Standard) as the free tier does not support it. Transitioning to web development, the presenter uses Claude 3 to generate code for a scroll-driven animation on a camera landing page, specifically utilizing XML tags within the prompt to ensure the AI correctly separates context, requirements, and animation behavior. The final section argues for using the WebP format over MP4 for hero section animations due to its performance benefits, including smaller file size, reduced bandwidth, and easier scroll interaction mapping via the standard tag, requiring no media player wrapper. ### Google AI Models & Tools - Gemini 3 launches as the most intelligent model - Veo is the new AI filmmaking tool custom-designed for Veo, Imagen, and Gemini - Whisk enables remixing images using subject, scene, and style references for greater control ### Veo Pricing & Usage - Veo 3.1 Fast costs $0.15/sec (Paid Tier) and is unavailable on the Free Tier - Frames to Video successfully generates a complex camera lens deconstruction animation ### Code Generation with Claude 3 - Claude 3 successfully integrates a 'Background Paths' UI component into a Next.js project using structured prompts with XML tags - Agent provides step-by-step integration guidelines including dependency installation ### WebP for Animation - WebP is preferred over MP4 for hero sections because it maps scroll interactions easier - WebP files are more compact than MP4, leading to faster load times and lower bandwidth usage ### UI Component Libraries - Inspiration for component design is sourced from libraries like 21st.dev (offering components like 'Background Paths') and HeroUI (a React UI library built on Tailwind CSS and React Aria) ![Screenshot at 00:01: 41:The output of the Veo Frames to Video tool successfully showing the camera lens elements separating in a technically accurate, animated sequence.](https://ss.rapidrecap.app/screens/XP6TJRbOdGo/00-00-01.jpg) ![Screenshot at 00:05: 55:The Whisk interface showcasing various generated images and the ability to select subject, scene, and style references for remixing.](https://ss.rapidrecap.app/screens/XP6TJRbOdGo/00-00-05.jpg) ![Screenshot at 00:17: 06:A pricing table from the Gemini API showing that Veo 3.1 Fast costs $0.15 per second in the Paid Tier and is not available on the Free Tier.](https://ss.rapidrecap.app/screens/XP6TJRbOdGo/00-00-17.jpg) ![Screenshot at 00:39: 00:Claude Code interface showing a detailed prompt structured with XML tags for integrating a component into a Next.js project.](https://ss.rapidrecap.app/screens/XP6TJRbOdGo/00-00-39.jpg) ![Screenshot at 00:50: 50:Demonstration of the Whisk tool generating sequences of images \(like the soda can advertisement\) based on simple descriptive text inputs.](https://ss.rapidrecap.app/screens/XP6TJRbOdGo/00-00-50.jpg)