# NANOBANANA PRO: IMAGE PROMPT ANATOMY REFERENCE
# Reference: a cheesecake baby character from a food-character series
# Model: nano-banana-2 via Kie AI | Aspect: 9:16 | Resolution: 2K

## WHAT THIS REFERENCE DOES

This document dissects a high-performing Nanobanana Pro image prompt layer by layer.
Use it to calibrate the depth, specificity, and structure of every prompt you write.
The goal is not to reproduce this image, it is to understand how each layer
of description contributes to the final render.

================================================================
REFERENCE IMAGE DESCRIPTION
================================================================

Subject: A hyper-realistic baby doll figure whose body is made entirely of
strawberry cheesecake, smooth cream cheese filling for skin, a biscuit crumble
base, strawberry compote dripping from fresh strawberries stacked on the head.
The figure is mid-bite, being fed by a human hand from the left holding a spoon.
Setting: Warm wooden surface, bokeh background with amber out-of-focus lights.

================================================================
LAYER-BY-LAYER PROMPT ANATOMY
================================================================

## LAYER 1: CORE SUBJECT

Weak version: "a cute baby made of cheesecake"

Expert version:
"A chubby baby doll figure whose entire body is sculpted from layered
strawberry cheesecake, the skin surface is smooth cream cheese filling
with a matte-satin sheen, interrupted by visible strawberry swirls at
the torso. The body sits on a thick graham cracker biscuit base that
provides a rectangular plinth. Fresh strawberries crown the head,
partially sliced to reveal the red interior, with glossy strawberry
compote dripping in thick rivulets down the forehead and cheeks.
The figure has oversized circular black doll eyes with a subtle specular
catch light. Rosy blush circles on both cheeks. Mouth open slightly,
lower lip extended."

What changed:
- Named the material (cream cheese, not just cream)
- Specified surface quality (matte-satin, not just smooth)
- Grounded the base (biscuit plinth gives spatial logic)
- Described the strawberries with specificity (sliced to reveal interior)
- Named the sauce behaviour (thick rivulets, not just dripping)
- Eyes: named the type (doll eyes), gave them a catch light
- Blush: located and described (circles, not just rosy)

## LAYER 2: COMPOSITION

Weak version: "front view, close up"

Expert version:
"Three-quarter low-angle shot, camera positioned slightly below subject
eye level. The figure occupies the centre two-thirds of the frame
vertically. The graham cracker base is fully visible at the bottom edge,
grounding the subject. The human hand entering from the left occupies the
left quarter of the frame at mid-height. Depth of field: extremely
shallow, the figure's face is sharp, the torso begins to soften toward
the lower third, the background is fully abstracted. No geometric
cropping, full figure visible from base to strawberry crown."

What changed:
- Named the angle precisely (three-quarter low-angle)
- Described what the camera sees relative to the subject's eye level
- Specified frame occupation (two-thirds vertically)
- Described the base as visible and grounding
- Placed the hand in the frame with spatial precision
- Described depth of field behaviour across the frame zones

## LAYER 3: LIGHTING

Weak version: "warm lighting, bokeh background"

Expert version:
"Key light: soft diffused source from upper-right, approximately 45° above
horizontal, creates smooth highlight across the right side of the figure's
face and cheek, with no hard shadow edge. Fill light: ambient warm bounce
from the background bokeh cluster (amber/candlelight temperature, ~2700K),
wrapping the left side of the figure with a very gentle warm-tone shadow fill.
No rim light. The cream cheese surface catches the key light as a broad
soft highlight, not a specular point, the material behaves like a
slightly translucent matte gel. The strawberry sauce reflects the key light
with a higher gloss index, wet, syrup-like sheen. The biscuit base
receives the least direct light, slightly underlit, warm shadow."

What changed:
- Named the light type (soft diffused, not just warm)
- Gave the light an angle (45° above horizontal)
- Described how the fill light works (bounce from background, not a separate source)
- Gave a colour temperature number (2700K)
- Described material-specific light behaviour:
  the cream catches light differently from the sauce
- The base is lit intentionally differently from the face

## LAYER 4: MATERIAL & TEXTURE

Weak version: "looks like real food, very detailed"

Expert version:
"Cream cheese body surface: smooth with micro-pore texture visible at
1:1 detail, not plastic-smooth, not rough. Surface tension implies
soft deformability. The strawberry swirls within the torso are visible
as embedded veins of colour, not painted on. The biscuit crumble base:
irregular grain structure, some larger crumb fragments visible at the edges,
powdery dust where it contacts the wooden surface. The strawberry compote:
behaves like a thick preserve, moves in wide rivulets, not thin drips.
The leading edge of each rivulet has a raised meniscus. Fresh strawberry
cut surfaces: visible achene (seed) indentations, bright interior red
against the darker rind exterior."

What changed:
- Each material gets its own paragraph
- Described how surfaces behave under deformation (soft deformability)
- Distinguished embedded vs painted-on colour
- Named physical phenomena (meniscus, achene)
- Described what the material looks like at different scales

## LAYER 5: COLOR & ATMOSPHERE

Weak version: "warm, cozy, pastel"

Expert version:
"Dominant palette: ivory-cream (#F5ECD7), strawberry red (#C8263C),
warm biscuit tan (#C4953A). Accent: the amber bokeh spheres carry
a secondary palette of warm gold (#E8A030) and burnt orange (#C06020).
Overall grade: slightly desaturated mid-tones, warm shadows, no cool tones
anywhere in the frame. The cream body reads as slightly warmer than pure
white, never clinical. Background atmosphere: diffused restaurant or
café warmth, the bokeh suggests string lights or candles, never fluorescent."

What changed:
- Named actual hex-range colours for each key element
- Described the grade approach (desaturated mids, warm shadows)
- Established a constraint (no cool tones) which prevents model drift
- Described the background's implied environment through its light character

## LAYER 6: TECHNICAL MODIFIERS

Weak version: "photorealistic, high quality, 4K"

Expert version:
"Photographic style: editorial food photography crossed with collectible
toy photography. Lens character: 85-105mm equivalent, slight portrait
compression. Rendering: physically-based materials, subsurface scattering
active on the cream cheese surface to give the slight translucency of
dairy-based food. No film grain. No vignette unless natural optical falloff.
Avoid: CGI plastic sheen on skin, hard specular highlights on the cream
surface, oversaturated strawberry red, symmetrical composition."

What changed:
- Named the photographic genre (editorial food + collectible toy)
- Specified focal length equivalent (gives rendering a compression feel)
- Named a specific rendering technique (subsurface scattering)
- Listed explicit avoidances, negative guidance built into the positive prompt

================================================================
THE COMPLETE ASSEMBLED PROMPT (what you actually send)
================================================================

"A chubby baby doll figure whose entire body is sculpted from layered
strawberry cheesecake. Skin surface: smooth cream cheese filling with
matte-satin sheen, strawberry swirls embedded in the torso. Body sits
on a thick graham cracker biscuit plinth, crumble fragments visible at
edges. Fresh strawberries crown the head, one sliced open to show red
interior, with thick glossy strawberry compote dripping in wide rivulets
down the forehead. Oversized circular black doll eyes with single catch
light. Rosy blush circles on both cheeks. Mouth open slightly, lower lip
extended. A human hand enters from the left holding a spoon loaded with
cream filling.

Three-quarter low-angle shot, camera slightly below eye level. Figure
occupies centre two-thirds of frame vertically. Shallow depth of field:
face sharp, torso softens, background fully abstracted. Full figure
visible from biscuit base to strawberry crown.

Soft diffused key light from upper-right, 45° above horizontal, broad
smooth highlight on face, no hard edge. Warm ambient fill from background
(~2700K, candlelight), wrapping left side. Cream surface: broad soft
highlight, not specular point. Strawberry sauce: high gloss, wet syrup
sheen. Biscuit base: underlit, warm shadow.

Palette: ivory-cream (#F5ECD7), strawberry red (#C8263C), biscuit tan
(#C4953A). Warm background bokeh (amber gold). Grade: desaturated
mid-tones, warm shadows, no cool tones. Background: café warmth,
string light or candle bokeh.

Editorial food photography crossed with collectible toy photography.
85-105mm compression. Subsurface scattering on cream surface. No film
grain. No hard specular on cream. No oversaturated reds. No plastic sheen.
No symmetrical composition. 9:16 vertical, 2K, photorealistic."

================================================================
END OF REFERENCE DOCUMENT
================================================================
