Gemini 101 — Image Generation with Nano Banana 2

Learn how to generate high-fidelity images from text, edit existing visuals, and apply professional style transfers using the latest Nano Banana 2 model.

From Text to Pixels

We have spent the first three modules mastering the art of conversation and text, but Gemini is not just a writer — it is an artist. In 2026, the barrier between thinking of an image and seeing an image has effectively disappeared.

Through the Nano Banana 2 model, Gemini can now understand complex visual metaphors, render text within images accurately, and even follow specific artistic movements like 80s City Pop or Minimalist Pixel Art. Whether you are creating a logo for your startup, storyboarding a video, or just having fun, this module will teach you how to direct the AI's "brush" with precision.

A creative desk where a physical keyboard transforms into a digital paintbrush, vibrant neon colors flowing onto a holographic canvas

What you will learn

The Visual Prompt Formula — beyond "draw a cat"

The Aesthetic Library — mastering styles and eras

The Director''s Cut — controlling lighting and camera angles

Image Editing — changing the world with inpainting

Style Transfer — applying one look to another subject

Text in Images — designing logos and posters

Ethics & Constraints — what the AI will not draw

The Creative Loop — refining your visuals

Section 1 — The Visual Prompt Formula

Just as we used the R-C-T-F formula for text, image generation needs structure to move from "generic" to "gallery-quality." Nano Banana 2 thrives when you provide a mix of Subject, Action, Setting, and Aesthetic.

The Beginner Prompt: "A man working on a laptop." (Result: generic, stock-photo style.)

The Pro Prompt: "A young entrepreneur (Subject) drinking coffee (Action) in a neon-lit cyberpunk cafe in Hong Kong (Environment) during the golden hour with a cinematic 35mm film aesthetic (Lighting/Style)."

A chalkboard showing a chemistry formula for an image: Subject + Setting + Lighting = Masterpiece, with icons representing each element

Section 2 — The Aesthetic Library

One of the most powerful features of Nano Banana 2 is its deep understanding of art history and modern design. You don''t need to be a designer — you just need to know the names of the vibes you want to capture.

Popular 2026 styles:

Vaporwave / City Pop — 80s Japanese aesthetics, pastel pinks and blues, retro-futuristic cityscapes.

Ukiyo-e — Traditional Japanese woodblock prints with bold lines and flat colors.

Isometric Pixel Art — Great for "game-like" environments and technical startup illustrations.

Minimalist Flat Design — Perfect for modern website headers and mobile app icons.

Section 3 — The Director''s Cut

To make your images look professional, you need to talk like a cinematographer. Nano Banana 2 understands technical photography terms that completely change the feel of an image.

Lighting keywords:

Volumetric Lighting — God rays and dramatic fog.

Cyberpunk Neon — High contrast, vibrant purples and cyans.

Soft Bokeh — Blurs the background so the subject pops.

Camera angles:

Bird''s-eye view — Looking straight down (great for maps and layouts).

Low-angle shot — Makes the subject look powerful and heroic.

Macro photography — Extreme close-up (great for textures or nature).

A vintage film camera floating in the air with glowing adjustment dials for ISO, Aperture, and Shutter Speed

Section 4 — Image Editing & Inpainting

One of the most revolutionary updates in Nano Banana 2 is the ability to edit an image after it has been generated. You no longer have to throw away a "nearly perfect" image because of one small mistake.

The Inpainting Workflow:

1. Generate your initial image.
2. Select the Edit tool.
3. Brush over the specific area you want to change (e.g. a person''s shirt).
4. Type a new instruction: Change the shirt to a black hoodie with a startup logo.

This unlocks iterative design — you can start with an empty room and slowly add furniture, change the wall color, or swap the view out the window.

Section 5 — Style Transfer & Multi-Image Composition

Sometimes you have an image of a subject you like, and another image of a style you love. Nano Banana 2 lets you merge them.

How to use reference images:

Composition reference — Upload a sketch of a layout. Gemini will follow that layout for the final render.

Style reference — Upload a Hiroshi Nagai painting and ask Gemini to "Draw the Hong Kong skyline in this exact style."

This is how you get brand consistency. If you are building a series of tutorials, upload your first image as a reference so all subsequent images share the same palette and lighting.

Two images merging in a swirl of pixels — a simple house and a Starry Night painting becoming a Van Gogh style house

Section 6 — Text in Images

Historically, AI was terrible at spelling. Nano Banana 2 has solved this. You can now prompt for specific text to appear on signs, t-shirts, or screens.

Example prompt: "A minimalist neon sign on a brick wall that says ''AfterWork'' in a futuristic font."

This is a game-changer for creating social media assets, YouTube thumbnails, and landing-page hero images without ever opening Photoshop.

Section 7 — Ethics, Safety, and Constraints

To keep the platform safe, Gemini has built-in guardrails. Knowing what it will not do saves you frustration.

If your prompt is rejected, make it more generic. Instead of naming a specific celebrity, describe the type of person — for example, "a charismatic tech founder in his late 30s."

Section 8 — The Creative Loop

The best AI artists don''t get the perfect image on the first try. They iterate.

1. The Sketch — Start with a basic prompt to see if the AI understands the layout.
2. The Refinement — Add lighting and style keywords.
3. The Edit — Use inpainting to fix small details.
4. The Upscale — Request a high-resolution version for the final.