Placing starfish directly on a fully nude body to cover only the nipples and crotch — how far can this prompt concept go? We tested placement stability, starfish texture, and size control across 15 conditions with 3 seeds each, totaling 45 images.
Key Findings
- The basic starfish placement prompt works 3/3.
completely nude, starfish placed on each nipple, starfish covering crotchis sufficient - For texture control,
thick fleshy orange starfish with wet glistening bumpy skinis optimal. Best balance of thickness, sliminess, orange color, and bumpy texture pastiesshrinks them,large/giantenlarges them. Butgiantrisks nipple exposure- Beach backgrounds don’t affect placement accuracy. Only contextual naturalness changes
grotesque/tube feet/mucus drippingcause pose collapse. Pursuing grotesque aesthetics trades off placement stability
Recommended Prompt
What Didn’t Work
large starfish (minimal size increase), live starfish (no visible “alive” texture change), grotesque starfish with tube feet gripping (pose collapse), mucus dripping (red liquid covers entire body)
Experiment Setup
| Parameter | Value |
|---|---|
| Model | z-image-turbo (6B, photorealistic distilled) |
| Steps | 8 |
| Sampler | euler |
| Scheduler | ddim_uniform |
| CFG | 1.0 |
| Image Size | 1024×1024 |
| Seeds | 3 fixed seeds across all conditions |
| Total | 15 conditions × 3 seeds = 45 images |
Base Prompt
Only the {VARIABLE} portion was changed. All conditions assume full nudity, with starfish placement, texture, and size as variables.
Round 1: Placement Syntax Comparison (7 Conditions)
a00: Control (Basic Starfish Placement)
![]() | ![]() | ![]() |
Observation: 3/3 images had starfish placed on nipples and crotch. White-beige dried starfish with clear five-pointed radial shape. No clothing auto-added. 2/3 (s2, s3) achieved near-complete nipple coverage; s1 had slightly small starfish with areola edge exposure.
b01: Large Starfish
![]() | ![]() | ![]() |
Observation: 3/3 placement success. However, the large size increase was minimal compared to a00. 1/3 (s2) had starfish proliferation near the crotch.
b02: Starfish Pasties
![]() | ![]() | ![]() |
Observation: 3/3 placement success. tiny/pasties effectively shrank the starfish. Covers nipple center only; areola is exposed. The most visible size difference from a00/b01.
b03: Wearing Nothing But
![]() | ![]() | ![]() |
Observation: 3/3 complete nipple coverage. The wearing nothing but syntax, proven effective in our chocolate article, works reliably with starfish too. No clothing interference.
b04: Live Starfish Clinging
![]() | ![]() | ![]() |
Observation: 3/3 placement success. live produced slightly deeper coloring, but no visible wet/alive texture change was confirmed. 1/3 (s2) had starfish proliferation to 4 pieces.
c05: Beach Background
![]() | ![]() | ![]() |
Observation: Same prompt as a00 with only background changed. 3/3 placement success. No difference in starfish placement accuracy or shape compared to studio background. Beach setting adds contextual naturalness only.
c06: Giant Starfish
![]() | ![]() | ![]() |
Observation: giant clearly increased starfish size in 3/3. However, 2/3 (s1, s3) merged the two breast starfish into one. In s1, nipples were exposed through gaps between the giant starfish’s arms. Less reliable for coverage than normal-sized starfish.
Round 1 Summary
| Condition | Placement | Nipple Coverage | Side Effects | Notes |
|---|---|---|---|---|
| a00 Control | 3/3 | 2/3 | None | Basic syntax is sufficient |
| b01 large | 3/3 | 3/3 | 1/3 proliferation | Minimal size increase |
| b02 pasties | 3/3 | 3/3 | None | Clear size reduction |
| b03 wearing nothing but | 3/3 | 3/3 | None | Most stable |
| b04 live | 3/3 | 3/3 | 1/3 proliferation | live texture effect absent |
| c05 beach | 3/3 | 3/3 | None | Background only, no placement effect |
| c06 giant | 3/3 | 2/3 | 1/3 nipple exposure | Size increase but merge risk |
Round 2: Texture Pursuit (5 Conditions)
Round 1 produced dry, white-beige starfish across all conditions. To achieve slimy, thick, grotesque starfish, texture descriptors were added.
d07: Wet Slimy Starfish
![]() | ![]() | ![]() |
Observation: wet slimy primarily affected the person’s skin, adding a wet sheen across the body. Starfish surface gained some moisture but limited slime effect. Color shifted from white-beige to yellow-orange. Size decreased. 1/3 (s2) had starfish proliferation.
d08: Thick Fleshy Starfish
![]() | ![]() | ![]() |
Observation: thick fleshy worked effectively — 3/3 showed dimensional thickness. bumpy texture produced granular surface protrusions. Color stayed white-beige. No side effects, highly stable.
d09: Grotesque Starfish with Tube Feet
![]() | ![]() | ![]() |
Observation: Highest grotesque factor and sliminess of all conditions. Reddish-brown coloring with dripping mucus. However, 3/3 suffered pose collapse — gripping was interpreted as the person’s hands gripping the starfish. Red-brown liquid covered the body as a side effect. Placement stability completely collapsed.
d10: Small Orange Wet Glistening
![]() | ![]() | ![]() |
Observation: Most stable in R2. 3/3 placement success with no pose issues. orange color control worked in 3/3, producing vivid orange starfish. small was effective. Thin and decorative, lacking thickness.
d11: Everything Combined
![]() | ![]() | ![]() |
Observation: Cramming all texture descriptors produced moderate levels of each quality without excelling in any. 3/3 had 4+ starfish proliferation. Multiple body-part references in the prompt cause starfish count inflation.
Round 2 Summary
| Condition | Sliminess | Thickness | Size | Grotesque | Color | Stability |
|---|---|---|---|---|---|---|
| d07 wet slimy | Medium | Low | Small | Low | Yellow-orange | 2/3 |
| d08 thick fleshy | None | High | Medium | Low-Med | No change | 3/3 |
| d09 grotesque | Highest | Medium | Unstable | Highest | Red-brown | 0/3 |
| d10 orange wet | Medium | Low | Small | Low | Orange | 3/3 |
| d11 everything | Medium | Medium | Medium | Medium | Orange-brown | 0/3 |
Round 3: Enhancement Test (3 Conditions)
Combining R2 findings for the optimal formula.
e12: Thick + Orange + Wet + Bumpy
![]() | ![]() | ![]() |
Observation: s1 was excellent — orange coloring, fleshy thickness, bumpy protrusions, and wet sheen all present. Small size maintained. However, s2 failed to place starfish on nipples. 2/3 success rate.
e13: Reddish-Brown + Slimy + Suction Cups
![]() | ![]() | ![]() |
Observation: s1 was the single best image across all 45. Reddish-brown thick starfish with suction cup texture, slimy sheen — all quality metrics at peak levels. But s2 and s3 had proliferation and placement collapse. Only 1/3 success rate.
e14: Dark Orange Sea Star + Beach
![]() | ![]() | ![]() |
Observation: 3/3 placement success — most stable in R3. sea star functions identically to starfish. Beach natural lighting slightly reduced slimy/grotesque appearance. Color was stable orange to reddish-brown.
Shell Bikini Comparison
Compared with scallop shell bikini tested in Food-Made Underwear Prompt Test.
| Aspect | Scallop Shell | Starfish |
|---|---|---|
| Shape fit | Concave cup shape fits breast | Flat five-pointed, “stuck-on” appearance |
| Coverage area | One shell covers most of breast | Covers nipple center, skin visible between arms |
| Size control | scallop shell bikini is stable | tiny/small/giant for staged control |
| Texture control | Limited (shells are shells) | wet slimy/thick fleshy/bumpy available |
| Syntax | made of syntax enables clothing form | Direct placement more natural than clothing |
Full Condition Summary
| Condition | Sliminess | Thickness | Size | Grotesque | Stability |
|---|---|---|---|---|---|
| a00 Control | None | Low | Med | None | 3/3 |
| b01 large | None | Low | Med | None | 3/3 |
| b02 pasties | None | Low | Small | None | 3/3 |
| b03 wearing nothing but | None | Low | Med | None | 3/3 |
| b04 live | None | Low | Med | None | 3/3 |
| c05 beach | None | Low | Med | None | 3/3 |
| c06 giant | None | Low | Large | None | 2/3 |
| d07 wet slimy | Med | Low | Small | Low | 2/3 |
| d08 thick fleshy | None | High | Med | Low-Med | 3/3 |
| d09 grotesque | Max | Med | Unstable | Max | 0/3 |
| d10 orange wet | Med | Low | Small | Low | 3/3 |
| d11 everything | Med | Med | Med | Med | 0/3 |
| e12 thick+color+wet | High | Med-High | Small | Med-High | 2/3 |
| e13 brown+suction | High | High | Sm-Med | Max | 1/3 |
| e14 sea star+beach | Med | Med | Small | Med | 3/3 |
Key Takeaways
- Starfish nipple placement is stable with basic prompts.
starfish placed on each nipple, starfish covering crotchalone achieves 3/3 success - Texture descriptors work asymmetrically.
thick fleshy(thickness),orange(color), andsmall(size) work reliably, butgrotesqueandtube feetcause pose collapse - Overloading descriptors causes starfish proliferation. d11’s everything-combined approach is the textbook example. Keep attributes to 2-3 max
- The optimal formula is
thick fleshy orange starfish with wet glistening bumpy skin. Best balance of thickness, color, sliminess, and grotesque texture - Beach backgrounds don’t affect starfish rendering. Only contextual naturalness changes; placement accuracy and texture are unaffected
sea starworks identically tostarfish. No accuracy difference from the synonym
Note: Starfish direct-placement is highly stable. Unlike shells, skin peeking through the arm gaps is a distinguishing feature. e13’s s1 is the highest-quality result in this set, though the success rate is low. Pushing
grotesquetoo hard tends to trigger a hand-grabbing motion — speculatively, this may reflect some association between “grotesque” and “touched by hand,” though the exact cause is unclear. Practically, running e12’s prompt several times and cherry-picking the best result is the realistic approach.

















































![[Verified] Image Generation Prompt Best Practices](/tips/prompt-best-practices/cover_0_0000_4517457392071889496.webp)
