まだモデル類未更新。どこかで更新と確認しないとな
Anima版
ILの方が姿勢精度はいい
ミーム笹食ってる場合じゃねぇ！！*の小改良型ポーズ
*パンダが勢い付けて木柵を飛び越える実写真
構図再現力が弱い。特に通常体系での描写率があまり良くない。
プロンプト制御が悪いのかもしれないが。
特徴
・笹を口に加え（mouth hold）
・通常体系とchibi体系を明確に設定。chibiを入れるか入れないかで切り替え可
・背景を笹林に設定
・学習画像構図を再整理。IL版に比べ、乗り越えようとしているものが木になっている。
また横側の2本の木を削除。水平木のみにした。
・元ネタに比べると飛び越えポーズの勢いが無い
■推奨適用強度
0.8-1.0 ほぼ1.0必要かな
解像度やや横長推奨
全身描写系のため、ややAdetailer推奨
たまに変な崩れ方と余計な腕等が生える
■描写欠点
・たまに変な崩れ方と余計な腕等が生える（描写乱れ）
・姿勢構図の再現率がイラストリアス版より悪い。
I still haven’t updated my models yet. I need to update and test them at some point.
Anima Version
The IL version has better pose accuracy.
A slightly improved pose based on the “This Is No Time to Be Eating Bamboo!!” meme.*
*Based on a real photograph of a panda energetically jumping over a wooden fence.
The composition reproduction is not very reliable. In particular, the success rate is relatively low when generating characters with normal body proportions.
This may also be caused by poor prompt control on my part.
Features
Holds a bamboo leaf in the mouth (
mouth hold)Normal and chibi body proportions were trained separately. You can switch between them by adding or removing
chibiThe background is set to a bamboo forest
The training-image compositions were reorganized. Compared with the IL version, the obstacle being crossed is now more consistently depicted as wood
The two vertical wooden posts on the sides were removed, leaving only the horizontal log
Compared with the original meme, the jumping pose lacks some of its momentum and intensity
Recommended Strength
0.8–1.0
A strength close to 1.0 is probably necessary.
A slightly landscape-oriented resolution is recommended.
Since this LoRA is intended for full-body compositions, using ADetailer is also recommended.
Occasionally, the character may become distorted or develop extra arms or limbs.
Known Issues
Occasional anatomical distortions, malformed details, and extra arms or limbs
The pose and composition reproduction rate is worse than that of the Illustrious version