🎨 AI-Powered Studio-Quality Product Photos with Vertex AI Imagen 3

🔗 Try it in Colab: Google Colab Notebook 📂 Source Code: image_generation.py

Generate professional, studio-quality product images using Google Vertex AI Imagen 3 — entirely from text prompts. Whether you're building an e-commerce store, a marketing campaign, or a product catalog, this tool lets you produce high-resolution AI-powered visuals at scale.

You can generate images of any product (sneakers, jewelry, clothing, electronics, and more), showcasing them from multiple angles while highlighting key details like materials, textures, branding, and lighting.

Run the notebook instantly in Google Colab — no setup required:

Or clone and run locally (see Installation below).

Before you begin, make sure you have:

• A Google Cloud account with billing enabled
• Vertex AI API enabled
• IAM role: Vertex AI User
• Python 3.8+

git clone https://github.com/RobinaMirbahar/vertex-ai-imagen3.git
cd vertex-ai-imagen3-colab

# Create and activate virtual environment
python -m venv venv
source venv/bin/activate        # Linux / macOS
venv\Scripts\activate           # Windows

# Install dependencies
pip install -r requirements.txt

from google.colab import auth
auth.authenticate_user()

import vertexai
vertexai.init(project="YOUR-PROJECT-ID", location="us-central1")

export GOOGLE_APPLICATION_CREDENTIALS="path/to/service-account-key.json"

from vertexai.preview import vision_models
from IPython.display import display

# Initialize the model
generator = vision_models.ImageGenerationModel.from_pretrained("imagegeneration-002")

# Generate product images
response = generator.generate_images(
    prompt="Professional studio photo of luxury sneakers, 360 view, 4K resolution",
    number_of_images=4,
    aspect_ratio="1:1",
    guidance_scale=15,
    seed=1234
)

# Display results
for idx, image in enumerate(response.images):
    print(f"Image {idx + 1}")
    display(image._pil_image)

Use negative prompts, custom aspect ratios, and fine-tuned parameters for more control:

response = generator.generate_images(
    prompt="Modern office chair with leather upholstery, studio lighting",
    number_of_images=3,
    negative_prompt="low quality, blurry, distorted, watermark, shadow artifacts",
    aspect_ratio="16:9",        # Supported: "1:1", "4:3", "16:9"
    guidance_scale=20,          # Higher = more prompt-adherent (range: 1–30)
    temperature=0.7,            # Controls creativity (range: 0.0–1.0)
    person_generation="avoid"   # Options: "allow", "avoid"
)

Category	Prompt	Preview
📷 General Product	"Professional studio photo of sneakers on a minimalistic background, soft diffused lighting, multiple angles"
🛒 E-Commerce Listing	"Product images of a smartwatch — front, back, side views, plain white background, even lighting, realistic shadows"
💎 Luxury / Premium	"Studio-quality luxury wristwatch on black satin, dramatic lighting, fine textures and engravings highlighted"
🏡 Lifestyle Shot	"Aesthetic lifestyle shot of a designer handbag in a modern home, natural lighting, warm tones, cozy atmosphere"
🔄 360° View	"Seamless 360-degree product series of a high-end DSLR camera, consistent lighting, high clarity for interactive shopping"
⚪ Minimalist	"Clean product image of a wireless earbud case on a neutral gradient background, soft shadows, studio lighting"
🌿 Outdoor / Adventure	"Rugged outdoor product shot of a hiking backpack in a natural setting, earthy tones, dynamic lighting, scenic backdrop"

Writing better prompts leads to significantly better results. Here are some best practices:

• Be specific about lighting: "soft diffused lighting", "dramatic side lighting", "golden hour sunlight"
• Specify background: "plain white background", "black satin surface", "blurred outdoor setting"
• Include view angles: "front view", "360-degree view", "top-down flat lay"
• Add quality modifiers: "4K resolution", "ultra-realistic", "studio quality", "sharp focus"
• Use negative prompts to exclude unwanted artifacts: "no watermark, no blurring, no distortion"

📖 See the full Prompt Engineering Guide for more.

# Update the Vertex AI SDK
pip install --upgrade google-cloud-aiplatform

# Enable required GCP services
gcloud services enable aiplatform.googleapis.com

# Verify authentication
gcloud auth application-default print-access-token

vertex-ai-imagen3-colab/
├── image_generation.py       # Core image generation script
├── requirements.txt          # Python dependencies
├── Images/                   # Sample output images
│   ├── img0.png              # Sneakers example
│   ├── img1.png              # Smartwatch example
│   └── ...
└── README.md                 # This file

Contributions are welcome! Here's how to get started:

1. Fork the repository 🔄
2. Clone your fork:

   git clone https://github.com/your-username/vertex-ai-imagen3-colab.git
   cd vertex-ai-imagen3-colab

3. Create a feature branch:

   git checkout -b feature/your-feature-name

4. Commit your changes:

   git commit -m "Add: description of your change"

5. Push and open a Pull Request 🚀

Please follow existing code style and include comments where appropriate.

This project is licensed under the Apache 2.0 License. See the LICENSE file for details.

All generated images must comply with Google's Generative AI Prohibited Use Policy. Please ensure responsible and ethical use of AI-generated content.

Built with ❤️ using Google Vertex AI Imagen 3.

If you find this project helpful:

• ⭐ Star the repository
• 🔄 Fork it to build your own version
• 📣 Share it with others who might benefit

Happy generating! 🚀

Repository	Architecture & Focus	Live Status
gemma4-multilora-adk-eval	Dynamic Multi-LoRA Hot-Swapping & CI/CD Agent Evaluation (adk eval) on Cloud Run GPUs with Gemma 4 & vLLM
slothsre-napops	SlothSRE ("NapOps") — Overnight Gemma 4 & Google ADK Incident Digest Agent with OpenTelemetry & Sentry AI Monitoring
streamforge-event-adk	Real-Time Event-Driven Multi-Agent Pipelines on Google Cloud with ADK (ParallelAgent & LoopAgent), Eventarc & Pub/Sub
a2a-mcp-mesh-cloudrun	Agent-to-Agent (A2A) & Multi-MCP Mesh on Cloud Run with Gemma 4, Google ADK, Spanner Graph & AlloyDB
cloudguard-gemma-agent	6-Part Production Gemma Enterprise Agent Series (ADK, Cloud Run L4 GPU, Firestore, Model Armor, Vertex AI QLoRA)
Comic-Studio-Ai	Multi-Agent Comic Generator powered by Google Gemini API & FastAPI deployed on Google Cloud Run
vertex-ai-imagen3	Generate AI-powered product & editorial images using Google Vertex AI Imagen 3 in Google Colab
Python-Learning-Assistant	AI-powered Python tutor covering 50+ interactive exercises with real-time feedback built with Streamlit + Gemini API
cozy-ai-agent-cafe-game	Prompt Paws: 60 FPS Cozy AI Agent Bakery Simulation Game teaching RAG, Guardrails, Context Caching & Multi-Agent Loops

If this project is helpful to your work:

• Star this Repository: Star RobinaMirbahar/vertex-ai-imagen3 on GitHub to bookmark it and help others discover it.
• Fork & Customize: Fork RobinaMirbahar/vertex-ai-imagen3 to build and experiment with your own workflows.
• Sponsor on GitHub: Sponsor @RobinaMirbahar on GitHub Sponsors (github.com/sponsors/RobinaMirbahar) to support future open-source Google (@google), Google Gemini (@google-gemini), Google DeepMind (@google-deepmind), and Google Cloud (@GoogleCloudPlatform) cookbooks!