Bringing Character Illustrations to Life with NanoBanana Pro: Directing Expressions & Composition (Before vs. After)

Images That Look Okay but Don’t Grab You
When generating character images with AI, sometimes the colors are fine and the design is decent, but the image strangely lacks charm. The problem can usually be summarized in one sentence.
“There is a face, but no attitude.”
In this post, I will select a decent but strangely unappealing character cut made with NanoBanana Pro and show you how it changes when only the expression, gaze, and camera are altered while keeping the design, colors, and background mostly the same, comparing Before and After.
Before – The Bland Default Image
Character Settings
- Character: Top-tier female student at a prestigious magic academy
- Atmosphere: Slightly haughty, but not too strong
- Location: Magic academy terrace with sunset
Anime illustration of an honor student girl at a magic academy. Upper body shot standing on a terrace in the sunset, wearing a uniform, detailed background, soft lighting. ar2:3
This image feels roughly like this
- Camera height: Eye level
- Aspect ratio: Upper body shot, character placed dead center
- Expression: Bland expression with closed mouth, staring straight ahead
- Background: Wall with sunset light coming in, cityscape slightly visible in the distance
Tip: There is almost no information about emotion, gaze direction, or camera position in the prompt. For the model, these are conditions where it has no choice but to draw a safe graduation photo composition.
Why this image doesn’t stick
It looks fine on the surface. But it is an image that instantly disappears when mixed into a feed.
1. The expression does not introduce the character
The setting is an honor student, but the face feels closer to a blank graduation album shot. She looks like someone just standing there because she was told to stand in front of the camera, conveying almost no emotional information.
- Is she haughty?
- Is she confident?
- Is she the type who sneers out of exhaustion?
It reads as none of these. If the expression is neutral, the character is remembered as neutral.
2. The gaze does not speak to the audience
The eyes seem to be facing forward, but they don’t throw any message to the viewer.
- No tension of being watched
- No sign that she is about to say something
Because these elements are completely absent, the image just feels like a character information card.
3. The camera is too nice
Front view, eye level, center of the screen. This is a good basic setting for information delivery. However, it is also a combination close to the worst for making a strong impression of a character. In a feed full of AI illustrations, such settings often get buried immediately.
Directing Notes – Summary of Points to Tweak
After seeing the Before image, the correction points noted can be summarized into four categories.
- Emotion
- This character needs to look like someone who knows very well that she is an honor student.
- Instead of a cold striking feeling, she needs a light confidence that says “I’m a cut above.”
- Gaze
- A gaze looking down lightly from slightly above the audience fits well.
- Not completely frontal, but a direction that is conscious of the audience while feeling relaxed is good.
- Camera Position
- It is better to lower the position slightly below eye level to create a composition looking up at the figure.
- Doing this makes her look like someone dominating the scene rather than an ordinary student.
- Placement in Frame
- Move the character slightly to the right from the center of the screen.
- Leave the left margin for the sunset and school building so that the background explains the character’s level and worldview together.
After – The Result of Changing Only Expression and Composition
This is a version where only the expression, gaze, camera, and screen placement were modified while maintaining the same character settings, similar outfit, and similar background.
Anime illustration of a confident honor student girl at a magic academy. Shot in 3/4 view standing on a terrace in the sunset, placed slightly to the right of the frame. Camera angle is set slightly low, looking up at her. Gentle confident smile, relaxed half-closed eyes, hair and uniform fluttering in the wind. Magic academy building and city skyline unfold in the left background, applying cinematic lighting effects.
ar2:3
Key Differences Before / After
- Added emotion and attitude sentences
- confident
- gentle confident smile
- relaxed half-closed eyes
- Added prompts for camera and framing
- Camera angle is set slightly low, looking up at her
- placed slightly to the right of the frame
In the end, I just specified where to shoot and where to place her in the screen with clear sentences. The model doesn’t always reflect it perfectly, but even this much brings it much closer to a character cut with presence suitable for a main thumbnail or cover, rather than a graduation album impression.
Item | Before | After |
|---|---|---|
Emotion | Bland neutral expression | Smile revealing light confidence |
Gaze | Frontal, meaningless stare | Relaxed gaze conscious of the audience |
Camera Height | Eye level | Looking up from a slightly lower position |
Position in Frame | Center of screen | Shifted to right side forming story with background |
Impression | Character cut for information | Character cut with presence enough for a cover |
Tip: The After version didn’t drastically increase details, but organized it so that the viewer can read ‘what kind of character this is’ at a glance.
Hints form a Directing Perspective
Even if colors, outfit, and background are similar, changing just the expression and composition makes ‘who this character is’ feel completely different. So when writing prompts, it helps to write down these three lines first.
- Who is this character speaking to right now?
- What attitude are they treating that person with?
- Where should the camera be to show that attitude?
If you organize this content in natural language first and then translate it into prompt sentences, you can experience the quality of the character cut rising a step without changing the model or style.
Most bland images can be seen as cases where these three lines were omitted from the prompt.
[Practical Prompt] Magic Academy Honor Student, Confident Terrace Shot Ver.
The first line describes the scene, and then it is divided into blocks. A scene of a top-ranking female student standing leaning against the railing of a magic academy terrace where sunset is settling, holding a folded exam result paper in her hand and looking down at the audience.

[Character]
1girl, late teens, top of the class student, calm but confident attitude, gentle smile that hints at self assurance, half-closed relaxed eyes, looking slightly down toward the viewer, holding a folded report paper in one hand, the other hand casually resting on the terrace railing, long straight hair flowing with the wind, subtle earrings as a small accent
[Outfit]
prestigious magic academy uniform, neatly arranged tie, well fitted blazer with golden piping, pleated skirt, soft fabric cape hanging from her shoulders, small metal emblem of the academy on her chest, slightly wind lifted cape edges, refined but not flashy accessories
[Atmosphere]
3/4 shot framing, camera positioned a bit lower than her eye level, slight upward angle to emphasize presence, character placed on the right side of the frame, wide empty space on the left showing the sunset sky and part of the academy tower, warm orange rim light outlining her silhouette, faint magical sigils glowing above the distant city roofs, cinematic depth of field, soft clouds catching the last light of the day
--ar 2:3The important parts here are:
- Emotion: gentle smile, self assurance, relaxed eyes
- Gaze: looking slightly down toward the viewer
- Camera: a bit lower than eye level, slight upward angle
- Placement: placed on the right side of the frame





