ComfyPilotComfyPilot
模型库 / LoRA / Takagi (Teasing Master Takagi-San/Karakai Jouzu no Takagi-san)
Takagi (Teasing Master Takagi-San/Karakai Jouzu no Takagi-san)

Takagi (Teasing Master Takagi-San/Karakai Jouzu no Takagi-san)

1043767

LoRA 3,751 次下载 426 喜欢 更新于 2026-11-15

底模:Krea 2

takagi-santakagiteasing master takagi-sankarakai jouzu no takagi-sancharacter

下载模型 在源站打开

触发词

takag1, 1girl, solo, long hair, sidelocks, brown hair, brown eyes, foreheadschool uniform outfit, school uniform, serafuku, school briefcase, serafuku, briefcase, blue pleated skirt, white shirt, sailor collar, short sleeves, red neckerchief, socks, shoescow outfit, animal ears, animal costume, cow ears, cow print, cow horns, cow costume, sleeves past wrists, animal print, neck bellcowgirl outfit, cowboy hat, bandana, cosplay | hand gun, weapon, aiming at viewer, aiming, holding gun, holding weapon, revolver | bar (place), bottlefrilly pink swimsuit outfit, swimsuit, pink bikini, navel, frills, side-tie bikini bottom, frilled bikini, hair bunprincess outfit, dress, tiara, cosplay, gloves, elbow glovesSanta outfit, Santa costume, Santa hat, sack, hat, boots, belt, long sleeves, red footwear, fur trim, belt

文件清单 线路:原始(huggingface.co) · 切换

下载后放进对应的 ComfyUI 目录。models/ 指 ComfyUI 的模型根目录。

权重 Takagi__Teasing_Master_Takagi-San_Karakai_Jouzu_no_Takagi-san__krea2_3471021_epoch_10.safetensors 218 MB 下载

说明

Takagi (Teasing Master Takagi-San/Karakai Jouzu no Takagi-san)

作者:MetalchromeX

If you like what I've done, consider supporting me at https://patreon.com/erodandy! Update 11-14-26: I did a Krea 2 version plus made some tweaks to the Anima version and uploaded a V2. Update 7-21-26: Since I was commissioned to do a Takagi Anima LoRA, I thought I'd first go back and update the dataset. V3 has a further refined dataset with improved tagging and training settings, which should lead to an even better and more flexible LoRA than V2. Update 3-18-25: I added a V2. V1 was already pretty good, but V2 is an improvement by reduction; I removed some questionable quality and redundant images and lowered the step count to make it less overtrained. Even so, the dataset is still about 750 images compared to the almost 1k of V1, so this is still a very detailed model!

示例图