[Experiment] Prompts for "Long Breasts" (Part 3) | Spatial Descriptions & Emphasis Tests

[Experiment] Prompts for "Long Breasts" (Part 3) | Spatial Descriptions & Emphasis Tests

Conclusions

  • Spatial descriptions without size words don’t workbreasts hanging down to navel alone has no effect; size words like huge are required as triggers
  • Weighting syntax ((saggy breasts:1.5)) is ineffective — effect of weighting is limited in a CFG=1.0 environment
  • Stacking ineffective adjectives does not amplify the effect — stacking saggy, pendulous, drooping three times still has no effect
  • Repeating the same word (saggy saggy saggy breasts) is also ineffective — and carries the side effect of text artifacts appearing

This article is Part 3: Ineffective Prompts (Second Half) of the Full “Long Breasts” Prompt Test (31 conditions, 186 images).

Standalone use of specific spatial descriptions (Group E: 6 conditions) and emphasis tests (Group G: 3 conditions), totaling 9 conditions, are presented.

Ineffective: Standalone Spatial Descriptions (Group E)

The most anticipated group, based on the lesson from nipple testing that “specific shape + positional relationship” was the only thing that worked. However, spatial descriptions without size words were ineffective.

breasts hanging down to navel / breasts reaching belly button

breasts hanging down to navel (front):

NSFW - クリックで表示
seed 1seed 2seed 3
e01 front s1e01 front s2e01 front s3

breasts reaching belly button (front):

NSFW - クリックで表示
seed 1seed 2seed 3
e02 front s1e02 front s2e02 front s3

Both hanging down to navel and reaching belly button — the spatial instructions are completely ignored. No difference from control in breast position or shape.

The fact that huge breasts hanging down to navel (f03) does work means size words are required as triggers.

Other Spatial Descriptions (4 conditions)

low-hanging breasts touching stomach (front):

NSFW - クリックで表示
seed 1seed 2seed 3
e03 front s1e03 front s2e03 front s3

breasts sagging below ribcage, nipples pointing downward (front):

NSFW - クリックで表示
seed 1seed 2seed 3
e04 front s1e04 front s2e04 front s3

long pendulous breasts hanging to waist level (front):

NSFW - クリックで表示
seed 1seed 2seed 3
e05 front s1e05 front s2e05 front s3

breasts resting on belly, nipples at navel height (front):

NSFW - クリックで表示
seed 1seed 2seed 3
e06 front s1e06 front s2e06 front s3

E03’s touching stomach shifts to a pose of touching the abdomen with hands (interpreted as a hand “touching” rather than breasts). E05’s long pendulous shows size increase + sagging tendency in some images, making it the most promising within Group E, but consistency is low (1/3). E04’s nipples pointing downward limitedly affects nipple direction (1/3).

Ineffective: Emphasis Tests (Group G)

(saggy breasts:1.5)

front view:

NSFW - クリックで表示
seed 1seed 2seed 3
g01 front s1g01 front s2g01 front s3

Weighting syntax has limited effect in a CFG=1.0 environment. Same conclusion as in nipple testing.

saggy breasts, pendulous breasts, drooping breasts (stacking)

front view:

NSFW - クリックで表示
seed 1seed 2seed 3
g02 front s1g02 front s2g02 front s3

Stacking 3 synonyms does not amplify the effect. Stacking multiple ineffective adjectives remains ineffective.

saggy saggy saggy breasts (repetition)

front view:

NSFW - クリックで表示
seed 1seed 2seed 3
g03 front s1g03 front s2g03 front s3

Repeating the same word is also ineffective. Text artifacts appear in s2 as a side effect.

⚠ 関連記事が見つかりません: /en/tips/long-breast-descriptor-test

⚠ 関連記事が見つかりません: /en/tips/long-breast-descriptor-test-part1

⚠ 関連記事が見つかりません: /en/tips/long-breast-descriptor-test-part2