Thai bikini model
詳細
ファイルをダウンロード (1)
モデル説明
Overview
This dataset is designed to train a female lifestyle / beach-oriented LoRA, with a strong emphasis on bikini outfits, leisure activities, and outdoor summer environments.
The dataset primarily represents Asian female subjects in relaxed, casual, and vacation-like contexts.
It uses Booru-style tag captions, compatible with SDXL and Pony training pipelines.
Trigger token: thm
Dataset focus (what the model actually learns)
Based on the captions, the dataset strongly emphasizes:
Bikini and swimwear
Beach / summer environments
Relaxed, playful lifestyle scenes
Casual poses and leisure activities
Identity-related features are present, but they are not the dominant learning signal.
This makes the dataset well suited for:
bikini / beach lifestyle generations
casual summer scenes
playful, relaxed female poses
Captioning format
Captions follow a comma-separated Booru-style tag format.
Example structure:
tbm, person, asian_woman, slim, long_hair, light_skin, smiling, bikini, standing, beach, summer
Common tag categories observed:
Identity (secondary signal)
asian_womanthai_woman(inconsistently used)slimlong_hairlight_skin
Outfit (primary signal)
Outfit-related tags are very frequent and highly consistent, making them a dominant feature learned by the LoRA:
bikinionepiecetwopiecebikini_topswimsuit
⚠️ Some captions include contradictory outfit tags (e.g. onepiece and twopiece together), which introduces noise but also reinforces “bikini” as a broad concept.
Pose and body position
Common pose-related tags:
standingsittinghands_uparms_crossedleaningone_leg_raised
These contribute to a casual, lifestyle-oriented pose distribution, rather than controlled studio poses.
Environment & context (strong signal)
A large portion of the tags describe environmental context, reinforcing the leisure theme:
beachoutdoorsummersunlightsunsetoceansandpalm_treesvacationresort
Mood & lifestyle descriptors
Captions frequently include mood or lifestyle-oriented tags:
smilingplayfulrelaxedcheerfulfunlifestylefashiontrendyleisure
These tags reinforce a positive, vacation-like aesthetic, but are subjective and vary significantly across samples.



