Unwrap your creativity: A gen media special

DevSember Week 3: Unwrap Your Creativity - Summary

Key Concepts:

• Gen Media: The generation of production-ready, high-fidelity, brand-consistent, and multimodal assets (visuals, audio, video).
• Nano Banana Pro: A Google AI model capable of high-quality image generation, search grounding, accurate text rendering, and strong prompt adherence.
• Search Grounding: The ability of an AI model to access and incorporate live data from Google Search into generated content.
• VO (Video Operations): Utilizing AI to generate video content from images and other assets.
• Gemini 3: The foundational large language model powering prompt creation, brand guideline generation, and overall creative direction.
• Vertex AI: Google Cloud’s platform for building and deploying AI applications, offering features like provisioned throughput, SynthID watermarking, and indemnification.
• AI Studio: A Google tool for rapid prototyping and experimentation with generative AI models.
• Creative Personas: Categorizing users (Builders, Professionals, AI Natives) to tailor tool development to their specific needs.

1. Introduction & The Rise of Gen Media

The session, part of DevSember week three, focused on shifting from text-based generative AI to “Gen Media” – creating production-ready visuals, audio, and video. The discussion highlighted the rapid advancements in generative AI, moving beyond simple memes to assets suitable for professional use. The theme of the week is “unwrap your creativity.”

2. Nano Banana Pro: Community Showcase & Capabilities

Kulan showcased examples of community creations using Nano Banana Pro on platforms like X, Instagram, and LinkedIn. Key examples included:

• Search Grounding: Pavle created images of cities based on live weather data from Google Search, demonstrating real-time data integration.
• Text Rendering: Anders generated infographics with remarkably accurate text, a significant improvement over previous models. This unlocks applications in fields like consulting and education.
• Camera Type Simulation: Users are testing Nano Banana Pro’s ability to realistically simulate different camera types (e.g., disposable cameras) based on prompts.
• Character Consistency: JB created images with consistent character appearance, eliminating the need for complex training or skin enhancement tools.
• Deblurring: Haleem demonstrated the model’s ability to deblur images.
• Multilingual Rendering: The model successfully generated images with text in various languages, including Korean, Japanese, and Ebo (a Nigerian language).
• Video Generation (with VO): Hussein created 3D floor plans and walkthrough videos, showcasing the power of combining Nano Banana Pro with VO.

3. The Gen Media Stack: Powering the Workflow

The core of Gen Media on Vertex AI consists of:

• Gemini Robotics: For potential touch-based interactions.
• Gemini 3 Pro Image (Nano Banana Pro): The image generation engine.
• LIA: For generating music.
• Gemini TTS (Text-to-Speech): For creating voiceovers.
• Gemini 3: The overarching “brain” providing creative direction and prompt assistance.

4. Building a Brand Ecosystem: A Live Demo

Kulan demonstrated a complete workflow using Gemini 3, Nano Banana Pro, and VO to create a brand ecosystem around a fictional “Get Lit” jumper:

• Brand Guideline Generation: Gemini 3 was used to generate both a text-based and visual brand guideline based on the jumper’s design.
• Localized Ads: Ads were created for the UK, Japan, and France, adapting the scene while maintaining brand consistency.
• Video Creation: Images were generated and then animated using VO, incorporating music generated by LIA and a voiceover created with Gemini TTS.
• Website Generation: AI Studio was used to generate a basic e-commerce website with the ability to virtually “try on” the jumper using a user’s uploaded photo.

5. Key Features for Developers & Creative Workflows

Kulan highlighted three crucial features for developers building creative tools:

• Thinking Process Exposure: Nano Banana Pro can reveal its reasoning behind image generation, helping users understand and refine prompts.
• Search Grounding: Access to live Google Search data for dynamic content creation.
• Large Context Window & Multi-Input Support: The ability to upload PDFs, images, and code (up to 14 images) as references for consistent generation.

6. Creative Personas & Building for the Right User

Kulan categorized users into three personas:

• Creative Builders: Technically proficient users who want programmability and control.
• Creative Professionals: Experienced designers who need granular control and consistency.
• AI Native Creatives: Users new to creative tools who need guided experiences and intuitive interfaces.

Developers should tailor their tools to the specific needs of their target persona. Analyzing prompt logs can provide valuable insights into user behavior and desired features.

7. Why Build on Vertex AI?

Vertex AI is recommended for enterprise-level applications due to:

• Provisioned Throughput: Guaranteed compute resources for consistent performance.
• SynthID Watermarking & Indemnification: Features for tracking AI-generated content and ensuring compliance.

AI Studio is suitable for rapid prototyping and personal projects.

8. Prompting Strategies & Gemini’s Role

The discussion emphasized the increasing importance of reference images over complex prompts. Gemini 3 is invaluable for:

• Prompt Generation: Assisting users in crafting effective prompts.
• Prompt Enhancement: Refining existing prompts for better results.
• Critiquing Outputs: Evaluating generated images and suggesting improvements.

9. Notable Quotes:

• “The relationship between code and creativity is really changing.” – Kulan
• “You’re not just building just another content management system, you are really building the infrastructure for imagination.” – Kulan
• “Everybody’s a builder.” – Stephanie
• “The key is reducing the gap between your curiosity and executing on your curiosity.” – Kulan

10. Conclusion & Takeaways

The session demonstrated the transformative potential of Gen Media, powered by models like Nano Banana Pro and Gemini 3. The key takeaways are:

• Gen Media is here: AI-generated assets are now production-ready.
• Focus on consistency: Utilize reference images and brand guidelines.
• Understand your users: Tailor tools to specific creative personas.
• Leverage Gemini 3: Use it for prompt creation, refinement, and overall creative direction.
• Consider Vertex AI: For enterprise-grade applications requiring scalability and security.
• Just start building: Experiment and iterate to unlock the full potential of generative AI.