用戶LoRA

XL
角色

Pavi - Avatar: Seven Havens - Illustrious by azertag

#阿凡達

#科拉傳奇

#降世神通:最後的氣宗

#阿凡達

#少女

Pavi - Avatar: Seven Havens - Illustrious by azertag - AI Model cover image
上傳時間
2026年4月30日 下午6:28
使用數
30
評論
暫無評論
觸發詞

pavi, 1girl, dark-skinned female, green eyes, short hair, purple hair, sidelocks, ahoge, hair between eyes, hairband, yellow hairband, wristband, single boot, amputee, prosthesis, running blade, bandaged leg, leg ribbon, leg bow, asymmetrical legwear, short sleeves, cape, green cape, brooch, short dress,

許可權
  • 允許圖片生成與分享
  • 允許使用者下載此模型
  • 允許生成圖片用於商業用途

描述

https://civitai.red/models/2075069/pavi-avatar-seven-havens-illustrious

A lora for the character of Pavi, the next Avatar from the upcoming cartoon Avatar : Seven Havens, the follow up to ATLA and TLOK. I'm surprised no one's done her yet.

There are only two "official" pictures of her, one officially released and the other is I think a leak, and I based this lora around the leak's appearance, which I realised but too late that it may be inaccurate regarding her hair, her prosthetic leg and her brooch. Might update it if this model becomes popular. Training the more official design should also prove easier.

This is only my second lora, working with barely two references so it's hard for it to be too accurate. Her missing leg is not always on the right side as well as the slit in her dress and her cape isn't thinned to the right in her back. And weirdly, a "white outline of bad crop residue" generates around her when my dataset had none of it. Also I didn't use Adetailer as to showcase the raw results but her face generates just fine with it.

I add many tags for the training, maybe it's counterproductive. Also the parameters I use make the lora around 300Mo, unlike the usual ~220Mo of SDXL Loras, if anyone got tips.

You can support me on my (NSFW) Patreon if you want to motivate me to make more : https://www.patreon.com/ArtagAI

Don't hesitate to post what you do with it.

評論