Nano Banana 2 Lite: Text to Image

HOW IT WORKS

Step 1. Write your prompt
Describe the subject, setting, mood, and style. Simple prompts get automatically expanded into a detailed scene by the built-in system prompt; detailed prompts are followed closely as written.
Works great with: scene descriptions · characters · concept art · mood pieces

Step 2. Pick your aspect ratio
Choose the shape that fits your output — 16:9 for widescreen, 1:1 for square, 9:16 for vertical, and so on.

Step 3. Set your generation options
Choose how many images to generate at once, lock or randomize the seed, and set output format.

Step 4. Hit run and download
Nano Banana 2 Lite generates your image directly from the prompt in seconds.
Ready for: Photoshop · Figma · Canva · any editor

First time? Leave every setting as-is. The defaults (16:9 · seed randomized · PNG · thinking level off) are the right starting point for almost everyone.

RECOMMENDED SETTINGS

Quick-start guide. Find the goal that matches yours and copy the settings.

• Standard run (most people) — Start here — 16:9 · seed randomized · PNG · thinking level off. The right starting point for almost everyone.

• Reproduce a result you liked — Lock the seed to the number that produced it, instead of leaving it on randomize.

• Need multiple options fast — Raise num_images to generate several variations from one prompt in a single run.

• Prompt feels ignored or over-interpreted — Turn thinking_level on for more deliberate interpretation of complex prompts; leave it off for speed on simple ones.

• Content getting filtered unexpectedly — Adjust safety_tolerance and enable_safety_checker to match how permissive you need the output to be.

Prompt: Describe the subject, environment, lighting, and style directly — simple prompts are automatically enriched with composition and atmosphere detail by the system prompt, so you don't need to over-write it yourself.

USE CASES

🖼️ Concept & Scene Art Generate full scenes and environments from a single description, no reference image needed.

🎨 Rapid Ideation Produce several prompt variations quickly to explore different visual directions before committing to one.

📐 Any Aspect Ratio Generate directly in the ratio you need for social, print, or web without cropping afterward.

✍️ Text-in-Image Prompt for legible on-image text and have it rendered clearly as part of the scene.

WHAT WORKS BEST / WHAT TO AVOID

✅ Works great

• Clear, descriptive prompts with subject + setting + style

• Letting the system prompt auto-enhance simple ideas

• Locking a seed once you find a result worth iterating on

• Straightforward single-subject compositions

⚠️ May produce softer results

• Overly vague one-word prompts with high expectations for detail

• Very cluttered multi-subject scenes in one prompt

• Conflicting style instructions in the same prompt

• Leaving safety tolerance too strict for the intended content

NEW TO COMFYUI?

Start with the free ComfyUI for Beginners Course on Floyo. Sixteen short videos take you from zero to running your own AI workflows. No setup headaches, no jargon, clear hands-on lessons. Watch the course, then run any workflow here in your browser.

👉 Watch the free ComfyUI for Beginners Course →

FAQ

What is Nano Banana 2 Lite? It's Google's fast image-generation model, served here through fal.ai. This workflow wraps it in a single text-to-image node with a built-in system prompt that improves composition, lighting, and detail automatically.

Does this workflow support image input or ControlNet? No — this graph is text-to-image only. There's no LoadImage node or ControlNet model in the pipeline, so structure (pose, edges, depth) isn't carried over from a reference image.

How is this different from a local ComfyUI checkpoint workflow? This node calls the Nano Banana 2 Lite model via the Floyo API rather than running a local model file, so there's no checkpoint, VAE, or CLIP loader to manage — just prompt and generation settings.

How do I get more consistent results across a batch? Lock the seed and keep the prompt identical, then only vary the parts you want to change (subject, style keyword, etc.).

Can I use the results commercially? Check Google/fal.ai's current terms for Nano Banana 2 Lite outputs, since commercial usage rights depend on the model provider's licensing, not on Floyo itself.

Ready to try it? Write a prompt, set your aspect ratio, and hit run.
→ Launch Workflow, Free

Questions? Watch the free course or check the FAQ above.