Shrinking a full character into a capsule and photographing it with a macro lens to achieve collectible-figure quality—this "world in your palm" effect is insanely eye-catching on social media. But most beginners crash on their first attempt: the capsule turns opaque, the character's proportions go haywire, or the depth of field blurs everything into mush.
This tutorial starts from zero. Three core concepts + three steps, and your first generation will produce a "hand holding a transparent capsule with a detailed miniature character inside" image.
Final Result Preview — See What You'll Make
A realistic human hand holding a transparent capsule. Inside: an incredibly detailed miniature character model—wrinkles in the clothing, hair sheen, even shoe sole textures all clearly visible. The capsule surface has real light reflections, the skin where fingers press turns slightly pale, and the background dissolves into creamy bokeh.
The whole image looks like a macro photograph of a real collectible figure, not an AI generation. Three things make this work: macro lens depth of field control, capsule material optical realism, and character-to-container scale conflict.
3 Core Concepts You Need to Know
Concept 1: Macro Depth of Field — Why "macro lens" Is Non-Negotiable
A macro lens produces extremely shallow depth of field: subjects in focus are razor-sharp (you can see pores), while everything out of focus melts into creamy blur.
Shallow DOF does two things for capsule figures:
- Forces attention onto the capsule: Fingertips and background blur, only the capsule zone stays sharp
- Creates the "real photograph" illusion: AI images default to full-frame sharpness, which screams "render." Add macro DOF and the brain subconsciously reads "this was shot with a camera"
photographed with a macro lens is the trigger. Writing only close-up without macro lens may give you a close framing but with everything in focus—losing the macro signature.
Concept 2: Capsule Refraction — Three Optical Layers "transparent" Triggers
A transparent capsule isn't just "you can see inside." Real transparent materials produce three optical phenomena:
- Refraction distortion: The curved surface slightly warps the edges of objects inside
- Internal wall reflection: At certain angles, you see faint secondary reflections on the capsule's inner surface
- Caustic highlights: Concentrated light spots where the transparent material focuses incoming rays
transparent capsule triggers AI to render these optical details. If AI makes the capsule frosted or semi-opaque, strengthen: crystal clear transparent capsule with visible light refraction.
Concept 3: Scale Conflict — Where the Visual Impact Comes From
The visual punch comes from a simple physical contradiction: a character that should be full-sized is shrunk to thumb-size and sealed inside a capsule.
The more specific the scale conflict, the stronger the effect:
- Character's palms pressed against the capsule wall (implies "trapped")
- Body curved to fit the capsule's arc
- Clothing details (buttons, zippers) magnified to visible clarity under macro
Use miniature model of [character] instead of small [character]. "Miniature model" implies a crafted figure (like a collectible), while small just means "not big"—AI might interpret it as a distant person.
3 Steps to Generate Your First Capsule Character
Step 1: Choose Your Character and Action
Character choice sets the mood. Action choice sets the story. Four recommended templates:
| Character Type | Prompt Phrasing | Recommended Action | Visual Mood |
|---|---|---|---|
| Cyberpunk Runner | a cyberpunk runner wearing neon-lit jacket | leaning against the glass wall | Lonely, sci-fi |
| Ancient Samurai | an ancient samurai in full armor | drawing sword in cramped space | Power, spatial tension |
| Magical Girl | an anime magical girl in sparkly outfit | pressing palms against capsule wall | Playful, "rescue me" |
| Astronaut | a space astronaut in full suit | floating weightlessly inside | Zero gravity, serenity |
Pick your character and action, then plug them into Step 2's template.
Step 2: Assemble the Complete Prompt
Base template (copy directly):
Close-up, A hand holding a transparent capsule, inside it is
a miniature model of [CHARACTER], wearing [OUTFIT]. The
character is posed in a [POSE], interacting with the inner
surface of the capsule, as if frozen or confined inside.
Hyper-realistic, cinematic, studio lighting setup,
photographed with a macro lens.
Example with "Cyberpunk Runner":
Close-up, A hand holding a transparent capsule, inside it is
a miniature model of a cyberpunk runner, wearing a neon-lit
jacket with glowing circuit patterns. The character is posed
leaning against the glass wall with one hand pressed flat,
interacting with the inner surface of the capsule, as if
frozen or confined inside. Hyper-realistic, cinematic,
studio lighting setup, photographed with a macro lens.
What each part does:
| Prompt Section | Function | What Happens Without It |
|---|---|---|
Close-up |
Locks tight framing | AI may generate full-body shot, capsule becomes tiny |
A hand holding |
Anchors scale reference | No hand means capsule size has no context |
transparent capsule |
Defines see-through material | May become opaque container |
miniature model of |
Forces collectible-figure quality | Character looks like a shrunken real person |
interacting with the inner surface |
Creates character-container contact | Character looks Photoshopped in |
macro lens |
Triggers shallow depth of field | Everything in focus, loses real-photo feel |
Step 3: Generate and Check 3 Core Indicators
After generating, check these one by one:
- Is the capsule transparent? You should clearly see the full character inside. If frosted, add
crystal clear glass capsule with visible refraction - Does the character look like a "model"? Should have collectible-figure quality—slight plastic or resin sheen. If too realistic, add
figure-like, collectible toy quality - Does depth of field exist? Fingertips and background should be blurred, only capsule sharp. If everything's crisp, add
extremely shallow depth of field, f/2.8 macro bokeh
3 Secrets to Getting It Right the First Time
Secret 1: Description order determines weight — Hand → Capsule → Character → Lens. AI weights earlier content more heavily. Put "hand" first to guarantee it appears, "character" in the middle to place it inside the capsule, "lens parameters" last as global modifiers.
Secret 2: Character actions must physically interact with the container. interacting with the inner surface is the line between "character happens to be in a capsule" and "character is trapped inside." Without interaction it's a still life; with interaction (pressing, curling, struggling) it becomes a story.
Secret 3: Don't manually specify too many colors. Capsule figure color should come from studio lighting and the character's own outfit. Let studio lighting setup handle color temperature automatically—it'll look far more natural than manually writing "red background" or "blue lighting."
Level-Up Challenge: 3 Advanced Variations
Variation 1: Add Preservative Liquid — From "Collectible" to "Sci-Fi Specimen"
After transparent capsule, add: filled with pale blue preservative liquid
Effect: The character appears submerged in lab fluid, hair floating upward, clothing edges drifting. The narrative shifts from "collectible showcase" to "alien research specimen."
Variation 2: Multi-Capsule Display — From "Single Piece" to "Collection Shelf"
Replace A hand holding a transparent capsule with: Three transparent capsules arranged on a laboratory shelf, each containing a different miniature character
Effect: Three different miniature characters in separate capsules, like a museum display cabinet. Great for series posters or character showcases. Note: with multiple capsules, each character's detail decreases—keep it to 3 max.
Variation 3: Cracking Escape — From "Static" to "Action Moment"
Append: the capsule has a visible crack with the character's tiny fist punching through it
Effect: Cracks radiate from where the character's fist is breaking through. From "static collectible" to "escape in progress"—visual impact doubles instantly.
Test these variations in nanobanana pro, adding one at a time and comparing each step's difference.
5 Common Beginner Mistakes
Mistake 1: Capsule Turns Opaque
Cause: transparent description isn't strong enough; AI defaults to semi-opaque material.
Fix: Upgrade to crystal clear transparent glass capsule with visible light refraction and caustic highlights. Three-layer description stacking (crystal clear + transparent glass + visible refraction) makes opacity nearly impossible.
Mistake 2: Character Scale Goes Wrong — Too Big or Too Small
Cause: AI interprets miniature model size unpredictably.
Fix: Add explicit proportion: the character occupies about 70% of the capsule's interior space. This gives AI a clear spatial guideline.
Mistake 3: Hand Generation Errors — Wrong Finger Count
AI hand generation remains a universal challenge. Two tips: use A hand not hands (single hand is easier to get right). If hands keep failing, switch to capsule resting on a velvet surface to bypass the hand entirely, using the velvet surface as your scale reference instead.
Mistake 4: No Depth of Field — Everything Is Sharp
Cause: macro lens weight gets diluted by other descriptions.
Fix: Append at the end: shallow depth of field, soft bokeh background, only the capsule is in sharp focus. Turns depth of field from an implication (macro lens) into a direct instruction.
Mistake 5: Character Looks Pasted In — No Interaction
Cause: Missing specific physical interaction descriptions.
Fix: the character's hands are pressed against the capsule wall, fingers slightly splayed, creating visible pressure points on the glass. The more specific the physical contact, the more AI renders "actually inside" rather than "composited on top."
Interested in transparent material optical control? Our glass fruit prompt word-by-word breakdown explains refraction, bubbles, and caustic effects in detail.
FAQ
Can I use a different container instead of a capsule?
Absolutely. Replace transparent capsule with glass jar, snow globe, test tube, or light bulb. Each container's shape and optical properties produce completely different visual effects. Light bulbs are especially interesting—the character gets bathed in warm tungsten filament glow.
Can AI generate existing IP characters?
AI recognizes most well-known characters, but facial details and logos may not be perfectly accurate. For better results, use detailed appearance descriptions instead of character names (e.g., "wearing red and blue tight suit with spider web pattern"). Mind copyright for commercial use.
What if hands keep generating incorrectly?
Two solutions: First, add A realistic human hand with visible fingerprints and skin texture before A hand to strengthen hand realism. Second, if errors persist, remove the hand entirely—use capsule standing upright on a dark reflective surface as an alternative composition base.
What aspect ratio works best?
Portrait (3:4 or 4:5) works best since hand-holding-capsule compositions are naturally vertical. Square (1:1) also works well for social media. Landscape is not recommended—too much empty background weakens the macro close-up focus.
Want to explore more 3D rendering styles? Our glassy neon 3D art guide shows how transparent materials behave under different lighting environments.