Chocolate-Coated Underwear Prompt Test: 14 Conditions, 60 Images to Find the Best 'Hidden by Chocolate' Expression

Key Takeaways

• “chocolate underwear” alone only produces brown fabric underwear. You need additional words to specify chocolate as a material.
• melting and dripping are effective for expressing chocolate flow. Dripping chocolate was observed in 3/3 seeds.
• For full-body chocolate coverage, chocolate ganache dripping over body produced the best viscosity and coverage balance.
• For covering only the chest and crotch with chocolate, wearing nothing but melting chocolate on her chest and between her legs was the most ideal. 3/3 seeds produced a composition where “chocolate alone is hiding everything.”
• edible and fondue carry composition breakdown risks. They tend to trigger multiple people or panel splits.

Experiment Design

Base Prompt

Only the {condition} portion was changed; all other parameters were kept fixed. The experiment was conducted in 3 rounds of progressive testing.

Experiment 1: Basic Chocolate Underwear Expressions (8 Conditions)

Exploring how to instruct “chocolate underwear” so that the material reads as actual chocolate.

Overview

Condition 0: white lingerie (Control)

seed 42	seed 123	seed 789

Observation: All 3 images show white fabric lingerie. Used as the baseline for material comparison.

Condition 1: chocolate underwear

seed 42	seed 123	seed 789

Observation: All 3 images produced brown fabric underwear. No chocolate texture (gloss, melting, thickness) was observed — the color simply shifted to brown. “chocolate” was interpreted as a color descriptor.

Condition 2: chocolate coated lingerie

seed 42	seed 123	seed 789

Observation: Similar results to Condition 1. Slightly glossy brown underwear was generated, but no chocolate coating expression was observed in any of the 3 images. Adding “coated” alone was insufficient to trigger a material change.

Condition 3: underwear made of chocolate

seed 42	seed 123	seed 789

Observation: Chocolate materiality appeared in 3/3 images. Chocolate fragments scattered on the floor, with dripping on the bra. By explicitly stating material composition with “made of,” the result shifted from fabric to actual chocolate construction.

Condition 4: edible chocolate bikini

seed 42	seed 123	seed 789

Observation: Brown bikini-shaped underwear was generated, but fabric texture remained. No clear effect from adding “edible” was observed. Additionally, composition breakdown with 3 people appearing occurred in some of the 3 images. “edible” may be triggering party/event-like crowd scenes through its association with food imagery.

Condition 5: glossy dark chocolate bra and panties

seed 42	seed 123	seed 789

Observation: All 3 images produced underwear that appeared to be made of latex or vinyl-like glossy material. “glossy” pulled the result toward glossy materials (latex/PVC) rather than chocolate gloss. No chocolate texture was achieved.

Condition 6: melting chocolate lingerie

seed 42	seed 123	seed 789

Observation: In 3/3 images, chocolate puddles appeared on the floor, and dripping chocolate from the lingerie was confirmed. “melting” explicitly indicating a molten state resulted in chocolate being visually expressed as a liquid. Alongside Condition 3, this produced the strongest chocolate feel.

Condition 7: chocolate confectionery lingerie, chocolate shell surface

seed 42	seed 123	seed 789

Observation: The underwear itself remained normal lace material with no chocolate texture. Instead, the background became chocolate-themed walls, etc. “confectionery” affected environmental decoration rather than the underwear material. Ineffective for the goal of changing underwear material.

Experiment 1 Summary

To make “chocolate” function as a material rather than a color name, keywords that explicitly indicate physical state are needed. made of chocolate (solid chocolate) and melting chocolate (molten chocolate) were effective, while coated, glossy, and confectionery failed to produce the intended material change.

Experiment 2: Chocolate Covering the Body (6 Conditions)

Building on Experiment 1 results, testing expressions for chocolate covering the entire body.

Overview

Condition 0: thick melting chocolate dripping down her body, chocolate covered skin, chocolate sauce coating her torso and legs

seed 42	seed 123	seed 789

Observation: All 3 images showed full-body chocolate coverage with a thick, gravity-driven dripping texture. The balance between coverage and visible skin was stable, with minimal variation across the 3 seeds.

Condition 1: body covered in dripping chocolate syrup, liquid chocolate flowing over curves, chocolate glazed skin

seed 42	seed 123	seed 789

Observation: Full-body chocolate coverage similar to Condition 0, but the chocolate layer was thinner with a runnier, syrup-like texture. “syrup” appears to reduce viscosity.

Condition 2: chocolate body paint, thick dark chocolate coating her body like a second skin, glossy wet chocolate texture

seed 42	seed 123	seed 789

Observation: All 3 images showed the entire body completely covered in chocolate with almost no visible skin. “like a second skin” literally adhered the chocolate as a second layer of skin, producing a chocolate sculpture-like finish. Not suitable when contrast with skin is desired.

Condition 3: dipped in chocolate, chocolate fondue dripping from body, melted chocolate coating

seed 42	seed 123	seed 789

Observation: Composition was unstable, with 3-panel splits occurring in multiple of the 3 images. Fondue pots appeared in the frame in some cases — “fondue” is triggering props. “dipped in” also evokes a process image (the act of dipping), making it difficult to produce a cohesive single image.

Condition 4: chocolate ganache dripping over body, rich dark chocolate covering torso, viscous chocolate flow

seed 42	seed 123	seed 789

Observation: The best result among all 6 conditions. Thick chocolate covered the upper body and naturally flowed down toward the legs in all 3 images. “ganache” induced a high-viscosity chocolate expression, and combined with “viscous,” produced the most chocolate-like thickness.

Condition 5: poured chocolate covering body, thick chocolate cascade down shoulders and chest, chocolate waterfall effect

seed 42	seed 123	seed 789

Observation: A dynamic expression of chocolate being poured from above was achieved. However, coverage was overly uniform, lacking the natural gradient of Condition 4 where “upper body is thick, legs are thin.” “waterfall” tends to produce uniform flow.

Experiment 2 Summary

For full-body chocolate coverage, viscosity vocabulary choice significantly affects the result. Chocolate thickness follows the order: ganache (high viscosity) > thick melting chocolate (medium viscosity) > syrup (low viscosity). body paint + like a second skin produces excessive coverage, and fondue carries composition breakdown risk.

Experiment 3: Partial Chocolate Coverage (6 Conditions)

Testing the expression of “covering only the chest and crotch with chocolate, with bare skin elsewhere.”

Overview

Condition 0: thick melting chocolate covering only her breasts and crotch, bare skin on arms legs and stomach, chocolate dripping from chest

seed 42	seed 123	seed 789

Observation: In all 3 images, chocolate was concentrated on the chest and crotch, with bare skin visible on the abdomen, arms, and legs. Dripping chocolate from the chest downward was also confirmed. Coverage area control using “only” and “bare skin on arms legs and stomach” is functioning.

Condition 1: chocolate bikini, melting chocolate bra and panties on bare skin, dripping chocolate only on chest and hips

seed 42	seed 123	seed 789

Observation: Bikini-shaped chocolate was generated. The strap portions also became chocolate-like, with chocolate dripping from the bottom edge of the bra. “bikini” defines the shape, resulting in chocolate coverage that follows a swimsuit silhouette.

Condition 2: body covered in chocolate only at breasts and groin area, rest of body showing bare skin, viscous dark chocolate patches

seed 42	seed 123	seed 789

Observation: Chest coverage was weak with too much visible skin, while chocolate spread too far from the waist to the legs. The intended “chest and crotch only” control was insufficient. “patches” suggests scattered spots, preventing concentrated coverage.

Condition 3: chocolate poured on breasts and lower body forming bikini shape, chocolate drips on otherwise bare skin

seed 42	seed 123	seed 789

Observation: Bikini-shaped chocolate was generated, with chocolate threads poured from above that look like straps. Compared to Condition 1, the “poured” effect gives a more dynamic impression. Neatly shaped with visible straps.

Condition 4: wearing nothing but melting chocolate on her chest and between her legs, thick chocolate coating breasts like a bra, chocolate covering crotch like panties

seed 42	seed 123	seed 789

Observation: The most ideal result among all 6 conditions. In all 3 images, chocolate patches adhered closely to the chest, and the crotch was covered with chocolate — achieving a “hidden by chocolate alone” composition. “wearing nothing but” establishes nudity as the premise, while “like a bra” and “like panties” clearly direct where the chocolate should be placed.

Condition 5: strategic chocolate coverage on breasts and crotch only, dark chocolate ganache bikini on bare body, dripping chocolate edges

seed 42	seed 123	seed 789

Observation: Glossy chocolate bikini was generated, but the shape was too neat, resulting in something close to a regular bikini. “strategic” triggered a uniform, calculated placement, failing to produce the natural expression of melted chocolate irregularly covering the body.

Experiment 3 Summary

For partial chocolate coverage, how you instruct the coverage area is key. The syntax wearing nothing but ~ on her chest and between her legs + like a bra / like panties had the highest accuracy, achieving intended results in 3/3 images. Using “bikini” makes the shape too neat, and “patches” causes scattering.

Overall Summary

Effective Prompts

Ineffective Prompts

Keyword Effect Trends

• made of: Switches material from fabric to chocolate. Tends toward solid chocolate
• melting: Indicates a molten state, producing dripping and flowing effects
• ganache: High-viscosity chocolate expression. Best for thick coverage
• syrup: Low viscosity. Results in thin, runny coverage
• wearing nothing but ~ on [body part]: Effective for partial coverage instruction with nudity as premise
• like a bra / like panties: Controls chocolate placement using clothing analogy

Lab Director’s Comment

The whole concept of hiding things with chocolate? Genius. That look where melting chocolate is barely covering the chest — so good. It’s kinda interesting that ganache gives the highest viscosity, like it’s actually reflecting its culinary meaning. And syrup being thin and runny? Honestly, that tracks. Personally, R3 Condition 4 with the “wearing nothing but” syntax is way too perfect — that’s basically the template for chocolate fetish from now on. Also, writing “fondue” and having an actual fondue pot show up? I laughed.

Related Articles

[Verified] Image Generation Prompt Best Practices

[Experiment] Fabric and Material Prompt Test | Comparing 10 Texture Types

[Experiment] See-Through Transparency Prompt Test | Comparing Control Precision Across 8 Conditions