Creating a Series of Illustrations with a Stable Character in Gemini

Until recently, creating a consistent series of illustrations in image generators was a challenging task. The same character could look different in each frame: hair color, clothing, body proportions, or even facial features would often change. To maintain consistency, users had to generate dozens of variations and describe every tiny detail with extreme precision.

Today, leading services are gradually solving this problem. With its latest update, Gemini makes it possible to create cohesive stories with a stable character. Below is a step-by-step guide on how to prepare prompts for a sequence of illustrations and lock the character’s appearance across all images.

1. Create a Story

The first step is to define a simple storyline that will guide the illustrations. For example:

A girl finds an old map inside a book that leads into the forest. She follows it and arrives at a cave, where among the rocks she discovers a rusty candy tin. Inside are childhood items — a folding pocket knife, a toy car, a small ball, a toy soldier, and a badge. She brings the box home, and when her father sees his forgotten childhood belongings, he cannot hold back his tears.

2. Character Description

To keep the character consistent across scenes, the description must be simple, precise, and unchanged. In this story we only need to describe the girl.

A 12-year-old girl with shoulder-length wavy black hair, natural dark brown eyes, and light skin. She wears round thin-rimmed glasses. Her outfit is always the same: a light cream short-sleeved shirt, knee-length denim shorts, and simple brown sandals.

The father only appears in the final scene and does not need a detailed description — the model will generate his appearance automatically.

3. Fix the Style

Locking the illustration style is crucial. This ensures all scenes look cohesive and belong to the same series.

Clean digital illustration style, consistent line art, flat soft shading, pastel color palette, smooth simple shapes, stable proportions, minimal visual noise, full-bleed composition filling the entire canvas with no borders or margins.

4. Storyboard

Break the story into key scenes. Each frame should represent a meaningful moment and flow naturally into the next.

• The girl finds the map.
• The girl heads into the forest.
• At the entrance of the cave.
• Inside the cave — the box.
• The open box with childhood items.
• The girl shows the box to her father.

5. Prompts for Illustrations

To generate the full series, open Gemini in image mode and enter the prompts one by one in the order of the storyboard. Start with the first prompt that includes the full character description and style, then continue with the following prompts to gradually build the story scene by scene.

Scene 1 — Finding the Map

A 12-year-old girl with shoulder-length wavy black hair, natural dark brown eyes, and light skin. She wears round thin-rimmed glasses, a light cream short-sleeved shirt, knee-length denim shorts, and simple brown sandals. Clean digital illustration style, consistent line art, flat soft shading, pastel color palette, smooth simple shapes, stable proportions, minimal visual noise, full-bleed composition filling the entire canvas with no borders or margins. The girl is standing in a dusty attic, holding an old folded map she has just found inside a large book, looking at it with curiosity.

Scene 2 — Walking in the Forest

The same 12-year-old girl with a small backpack on her back. She is walking along a narrow forest path on a sunny summer day, holding the unfolded map in her hands, looking ahead with anticipation.

Scene 3 — At the Cave Entrance

The same 12-year-old girl with a small backpack on her back. She is standing in front of a large cave entrance surrounded by rocky cliffs and tall trees, the cave covered with green ivy, holding the unfolded map in her hands and looking at the dark opening with curiosity.

Scene 4 — Inside the Cave

The same 12-year-old girl. She is inside a dark cave with a small backpack on her back, holding a vintage flashlight in one hand and shining it onto a small closed metal candy box lying among the rocks. The box looks old, slightly rusty, and worn. Her face shows curiosity as she gazes at it.

Scene 5 — Opening the Box

The same 12-year-old girl. She is sitting on the ground just outside the cave in the forest, with her small brown backpack placed nearby. In front of her is a small gray rusty metal candy box, now open, filled with childhood treasures. Inside the box are a tiny toy car and a seashell. On the ground beside it lie a toy soldier and a few small marbles. She is holding an old notebook in her hands, reading it with wide eyes and a very surprised expression.

Scene 6 — At Home with Her Father

The same 12-year-old girl. She is standing beside her father, a man around 40 years old, in a cozy evening living room. On a table in front of him rests a simple rusty metal box with a few old toys inside. He is holding a toy soldier, his eyes wide with a mix of joy, nostalgia, and excitement, while with his other hand he touches his head as if overwhelmed by emotion. The girl smiles with anticipation and warmth, watching his reaction.

Practical Tips

• Always include the full character description and style in the very first prompt to lock appearance and rendering.
• In subsequent prompts, simply refer to “the same 12-year-old girl” and adjust only the location, pose, and actions.
• Use short, clear sentences without metaphors or artistic flourishes. AI models handle simple language more reliably.
• For long sessions, start a new chat for each series to avoid drift and accumulated errors.
• Keep props and background details consistent by naming them directly (e.g., “the same rusty candy box”).