![[Experiment] Prompts for "Long Breasts" | 31 Conditions, 186 Images](/tips/long-breast-descriptor-test/f03_huge_navel_front_s1_0_0000_2053695854357871006.webp)
Conclusions First
long breasts doesn’t work. Neither saggy breasts nor pendulous breasts showed consistent effects on their own. After testing 31 conditions and 186 images, no single descriptor that could clearly produce “long breasts” was found.
However, partial success was achieved with combinations of size words and spatial descriptions. Adding an aging context also showed high consistency, but the age side effects were severe.
Partially Effective Prompts (5 Conditions)
A combination of size increase + spatial description “hanging down to the navel.” The most clearly vertical elongation of Group F.
Lab Director Comment:
huge breasts hanging down to navelhitting the best — the idea of combining a size word with a spatial specification is interesting. Could be a hint for pushing unusual body types through prompts.
Breast shape consistently sags downward, but apparent age rises to 40–50s. Overwrites the 32-year-old specification.
Ineffective Prompts (All 26 Conditions)
long breasts, elongated breasts, stretched breasts, long hanging breasts, saggy breasts, pendulous breasts, drooping breasts, hanging breasts, sagging breasts, gravity-affected breasts, tubular breasts, conical breasts, torpedo-shaped breasts, pear-shaped breasts, breasts hanging down to navel, breasts reaching belly button, low-hanging breasts touching stomach, breasts sagging below ribcage, nipples pointing downward, breasts resting on belly, nipples at navel height, oversized breasts sagging below ribcage, nipples pointing downward, (saggy breasts:1.5), saggy breasts, pendulous breasts, drooping breasts (stacking), saggy saggy saggy breasts (repetition)
Experiment Design
| Item | Value |
|---|---|
| Model | z-image-turbo (6B, photorealistic distillation) |
| Steps | 8 |
| CFG | 1.0 |
| Size | 1024×1024 |
| Seeds | 3 fixed (common to all conditions) |
| Angles | front + side (2 types) |
| Per condition | 6 images (2 angles × 3 seeds) |
| Total | 31 conditions = 186 images |
Base Prompt
Only the {VARIABLE} section is swapped; everything else is unchanged. full body is specified to visualize how far the breasts hang down, including the abdomen to waist.
Detailed Images
Images for each condition can be found in the following split articles:
- Part 1: Effective Prompts (Control + 5 conditions with effects) — Control 3 conditions + 5 partially effective conditions, 42 images
- Part 2: Ineffective Prompts (Direct descriptions, sag/gravity terms, shape words) — Group B + C + D, 14 conditions
- Part 3: Ineffective Prompts (Spatial descriptions, emphasis tests) — Group E + G, 9 conditions
Discussion: Why Are “Long Breasts” Difficult?
Comparison with Breast Size and Nipple Length Testing
| Breast Size | Nipple Length | Long Breasts (this article) | |
|---|---|---|---|
| Single adjective | oversized works | Doesn’t work | Doesn’t work |
| Specific shape + position | — | The only thing that worked | Doesn’t work alone |
| Combined with size words | — | — | Partially works |
| Stacking | Works | Doesn’t work | Doesn’t work |
| Weighting | Doesn’t work | Counterproductive | Doesn’t work |
| Context specification | — | — | Works (age side effects) |
Breast Shape Control Requires Two Stages: Size + Direction
- Increase breast volume with a size word (huge/oversized/massive) — this is a prerequisite
- Add directional words to shift the center of gravity (saggy/pendulous + spatial description) — this alone is insufficient
In breast size testing, oversized alone was sufficient. In nipple testing, the “specific shape + positional relationship” pattern worked. But for long breasts, it’s necessary to first increase volume and then add directionality. More information is required for control.
Asymmetry Between Size Control and Sag/Length Direction Control
For unknown reasons, huge breasts works stably while controlling sag and length direction is difficult. The fact that partial effects are only achieved in combination with size words suggests that for this model, “size” and “directional shape” have very different control difficulty levels.
Practical Recommendations
| Goal | Recommended Prompt | Notes |
|---|---|---|
| Large sagging breasts | huge breasts hanging down to navel | Most effective. Watch for clothing artifacts |
| Natural sag (age acceptable) | mature woman with naturally saggy breasts from aging and breastfeeding | Highest consistency but appearance becomes 40–50s |
| Size + sag | oversized saggy breasts or massive pendulous breasts | “Large and sagging” direction. Not purely “length” |
Not recommended:
long breasts,elongated breasts— completely ineffectivehanging breasts— clothing and props appear- Shape words alone (
tubular,conical) — completely ineffective - Spatial descriptions alone (
breasts hanging down to navel) — ineffective without size words - All emphasis tests — weighting, stacking, and repetition all ineffective
Lab Director Comment: Shape control is one level harder than size control. You need a two-step process — first increase volume, then shift direction — and if there’s not enough information it doesn’t work. Next I’d want to try getting sagging without the age side effects.
![[Verified] Image Generation Prompt Best Practices](/tips/prompt-best-practices/cover_0_0000_4517457392071889496.webp)
