Generative AI Google Brain

gemini-use-cases-marketing-creative-content-generation

3 min read

Creative Content Generation with Gemini in Vertex AI and Imagen

Source notebook

Repo path: gemini/use-cases/marketing/creative_content_generation.ipynb · Open on GitHub · intermediate

Generates and personalizes GShoe marketing copy with Gemini, then outpaints product images with Imagen.

Summary

This notebook teaches a marketing workflow that starts with Gemini text generation from a product name, then improves prompts with a product description and product image. It demonstrates targeting posts for social platforms, adapting copy for professions, generations, countries, and languages, then using Imagen outpainting to expand a product image from 1:1 to 16:9.

Key code patterns

Initialize Vertex AI

import vertexai
 
PROJECT_ID = "YOUR_PROJECT_ID"
REGION = "us-central1"
vertexai.init(project=PROJECT_ID, location=REGION)

Sets the project and region required before using Vertex AI models.

Generate copy with Gemini

from vertexai.generative_models import GenerativeModel
 
model = GenerativeModel("gemini-2.0-flash")
prompt = f"""
Generate a few social media posts about {product_name}.
This is the product description: {product_description}
"""
response = model.generate_content(prompt).text

Shows iterative prompt enrichment from product name to detailed marketing context.

Use product image input

from vertexai.generative_models import Part
 
product_image = Part.from_uri(
    "gs://github-repo/use-cases/marketing/gshoe-images/gshoe-01.jpg",
    mime_type="image/jpeg",
)
content = [prompt, product_image]
posts = model.generate_content(content).text

Uses Gemini multimodal input to ground social posts in a product image.

Personalize existing posts

prompt = f"""
Reference to these social media posts: \n{social_media_posts}
Make the posts suitable for students
"""
result = model.generate_content(prompt).text

Reuses generated copy as context for audience-specific adaptation.

Outpaint with Imagen

from vertexai.preview.vision_models import Image, ImageGenerationModel
 
imagen_model = ImageGenerationModel.from_pretrained("imagegeneration@006")
base_img = Image.load_from_file(location=image_path)
mask_img = Image.load_from_file(location=mask_path)
images = imagen_model.edit_image(
    base_image=base_img,
    mask=mask_img,
    edit_mode="outpainting",
    prompt="a shoe surround by flowers",
)

Expands a prepared product image and mask using Imagen outpainting.

Models & APIs used

  • Models: gemini-2.0-flash, imagegeneration@006
  • APIs / services: Vertex AI, Gemini API in Vertex AI, Imagen on Vertex AI, Cloud Storage
  • SDKs / libraries: google-cloud-aiplatform, vertexai, requests, PIL

When to use this

Use this pattern to prototype marketing content generation and platform-specific creative assets from product text and imagery.

Gotchas & caveats

  • Requires an existing Google Cloud project with billing enabled.
  • Vertex AI API must be enabled.
  • Local execution requires Cloud SDK installation.
  • Colab requires explicit Google Cloud authentication; Vertex AI Workbench does not.
  • Notebook was tested with Python 3.11 and google-cloud-aiplatform 1.54.0.
  • Imagen features in the sample require allowlisting through the Imagen on Vertex AI access request form.
  • The outpainting helper expects a 1:1 input image and raises an error otherwise.
  • The sample uses us-central1 unless REGION is changed.

Best practices

  • Add product description to improve generated output quality.
  • Use product images as multimodal context for marketing messages.
  • Target generated posts to specific platforms such as Facebook, Instagram, LinkedIn, and Twitter.
  • Reuse generated posts as reference context when personalizing for audience segments.
  • Prepare an expanded base image and mask before calling Imagen outpainting.
  • Clean up by deleting projects or resources and disabling the Vertex AI API to avoid charges.

Graph View

Backlinks