Nano Banana 2 by Google: Examples & Prompts
This page is a Nano Banana 2 capability explorer. The main experience is the selector above. What follows is the technical appendix.
Endpoints
- Official Google overview: https://gemini.google/overview/image-generation/
fal-ai/nano-banana-profor text-to-imagefal-ai/nano-banana-pro/editfor single-image and multi-image edits
On fal, Nano Banana 2 currently maps to the nano-banana-pro endpoints above.
When Nano Banana 2 is worth it
- You need text or layout fidelity.
- You need cleaner subject consistency.
- You want higher-resolution marketing stills.
- You need edits that feel controlled instead of merely plausible.
Minimal Python (text-to-image)
Make sure FAL_KEY is set in your environment.
python - <<'PY'
import urllib.request
import fal_client
result = fal_client.subscribe(
"fal-ai/nano-banana-pro",
arguments={
"prompt": "A minimalist product photo of a tiny banana-shaped robot resting on a matte black pedestal, studio softbox lighting, clean white background, subtle shadow, macro photography.",
"aspect_ratio": "1:1",
"num_images": 1,
"output_format": "png",
"resolution": "1K",
"seed": 123,
"limit_generations": True,
},
)
url = result["images"][0]["url"]
urllib.request.urlretrieve(url, "nano-banana-pro-product.png")
print("saved nano-banana-pro-product.png")
PY
Minimal Python (background replacement)
python - <<'PY'
import urllib.request
import fal_client
# Run this from the post directory, or update paths if needed.
base_url = fal_client.upload_file("./cover.png")
result = fal_client.subscribe(
"fal-ai/nano-banana-pro/edit",
arguments={
"prompt": "Keep the banana robot unchanged. Replace the background with a deep blue to magenta gradient, add a clean cinematic rim light, preserve the subject details, and keep the macro depth of field.",
"image_urls": [base_url],
"num_images": 1,
"output_format": "png",
"resolution": "1K",
"limit_generations": True,
},
)
url = result["images"][0]["url"]
urllib.request.urlretrieve(url, "nano-banana-edit-pro.png")
print("saved nano-banana-edit-pro.png")
PY
Minimal Python (multi-image edit)
python - <<'PY'
import urllib.request
import fal_client
# Run this from the post directory, or update paths if needed.
subject_url = fal_client.upload_file("./cover.png")
environment_url = fal_client.upload_file("./example-terrarium-empty.png")
result = fal_client.subscribe(
"fal-ai/nano-banana-pro/edit",
arguments={
"prompt": "Use image 1 as the subject and image 2 as the environment. Place the tiny banana robot from image 1 naturally inside the terrarium from image 2. Preserve the robot's design and materials, keep the glass geometry and moss, add soft morning light and realistic reflections, and maintain a premium macro-photography look.",
"image_urls": [subject_url, environment_url],
"num_images": 1,
"output_format": "png",
"resolution": "1K",
"limit_generations": True,
},
)
url = result["images"][0]["url"]
urllib.request.urlretrieve(url, "nano-banana-multi-edit-pro.png")
print("saved nano-banana-multi-edit-pro.png")
PY
References
- Official Google overview: https://gemini.google/overview/image-generation/
- Nano Banana Pro text-to-image API: https://fal.ai/models/fal-ai/nano-banana-pro/api
- Nano Banana Pro edit API: https://fal.ai/models/fal-ai/nano-banana-pro/edit/api
- Nano Banana 2 dev guide: https://fal.ai/learn/devs/nano-banana-2-api-developer-guide
- Nano Banana Pro launch post: https://blog.fal.ai/nano-banana-pro-is-now-available-on-fal/