Nanobanana Generate Image

Generate and transform images using Google Gemini 3 Pro Image Preview model through the Zenmux API gateway.

Capabilities

•Text-to-Image Generation: Create images from text descriptions
•Image-to-Image Transformation: Modify existing images (e.g., translate text, inpaint, edit)

Prerequisites

•ZENMUX_API_KEY: Environment variable must be set with your Zenmux API key
•Python packages: google-genai, Pillow

Install dependencies:

bash

pip install google-genai Pillow

Instructions

Text-to-Image Generation

Use scripts/generate_image_from_text.py as reference:

python

import os
from google import genai
from google.genai import types

api_key = os.getenv("ZENMUX_API_KEY")
if not api_key:
    raise ValueError("ZENMUX_API_KEY environment variable is not set")

client = genai.Client(
    api_key=api_key,
    vertexai=True,
    http_options=types.HttpOptions(api_version='v1', base_url='https://zenmux.ai/api/vertex-ai')
)

prompt = "Your image description here"

response = client.models.generate_content(
    model="google/gemini-3-pro-image-preview",
    contents=[prompt],
    config=types.GenerateContentConfig(
        response_modalities=["TEXT", "IMAGE"]
    )
)

for part in response.parts:
    if part.text is not None:
        print(part.text)
    elif part.inline_data is not None:
        image = part.as_image()
        image.save("generated_image.png")
        print("Image saved as generated_image.png")

Image-to-Image Transformation

Use scripts/generate_image_from_image.py as reference:

python

import os
from google import genai
from google.genai import types
from PIL import Image

api_key = os.getenv("ZENMUX_API_KEY")
if not api_key:
    raise ValueError("ZENMUX_API_KEY environment variable is not set")

client = genai.Client(
    api_key=api_key,
    vertexai=True,
    http_options=types.HttpOptions(api_version='v1', base_url='https://zenmux.ai/api/vertex-ai')
)

# Load local image
local_image = Image.open("input_image.png")

prompt = "Your transformation instructions here"

response = client.models.generate_content(
    model="google/gemini-3-pro-image-preview",
    contents=[prompt, local_image],
    config=types.GenerateContentConfig(
        response_modalities=["IMAGE"]
    )
)

for part in response.parts:
    if part.text is not None:
        print(part.text)
    elif part.inline_data is not None:
        image = part.as_image()
        image.save("generated_image.png")
        print("Image saved as generated_image.png")

Key Configuration

Parameter	Value
Model	`google/gemini-3-pro-image-preview`
API Base URL	`https://zenmux.ai/api/vertex-ai`
API Version	`v1`

Response Modalities

•["TEXT", "IMAGE"]: Get both text explanation and generated image
•["IMAGE"]: Get only the generated image
•["TEXT"]: Get only text response

Examples

Generate a creative image

python

prompt = "Create a picture of a nano banana dish in a fancy restaurant with a Gemini theme"

Translate text in an image

python

language = "English"
prompt = f"Identify and translate the text in this image to {language}. Use inpainting to seamlessly replace the original text while preserving the visual context."

Edit an image

python

prompt = "Change the background color to blue while keeping the main subject intact"

Reference Scripts

•scripts/generate_image_from_text.py: Text-to-image generation example
•scripts/generate_image_from_image.py: Image transformation example

Troubleshooting

•
ZENMUX_API_KEY not set: Export the environment variable before running
bash
```
export ZENMUX_API_KEY="your-api-key"
```
•
Missing packages: Install required dependencies
bash
```
pip install google-genai Pillow
```
•
Image not saving: Ensure you have write permissions in the output directory