Sign In

Heart of Apple XL

646
3.4k
32
0
Verified:
SafeTensor
Type
Checkpoint Trained
Stats
1,822
Uploaded
Feb 10, 2024
Base Model
SDXL 1.0
Training
Epochs: 400
Hash
AutoV2
CA2F93E76D
0
0
0
0
0

Introduction to

This is a model based on Animagine XL V3, which has relatively good performance capabilities. It is trained using roughly the same methods as Animagine XL V3, such as arranging tags. Compared to Animagine XL 3, Heart of Apple XL is more diversified in terms of Animagine XL 3.0 style performance, with specific training for some game characters and artists.

Heart of Apple XL has been trained for 8 days on a100 40g server. It will continue to be updated in the future. Thank you again to CagliostroLab for their contributions to the community. I also want to thank 阿萨姆奶茶,小苹果,青秋,zwh20081,原物质,琥珀青叶,灾厄之轮,hans,星河主炮向前,alens and others for their technical and financial assistance, as well as my partners who have encouraged me.

User's Guide

Like Animagine XL V3, you need to arrange the tags. For example, if you want to achieve a better effect with multiple artist prompts, you can try placing them at the end. And we have added several negative words such as quality bad, hands bad, eyes bad, face bad, normal quality,

Recommended Prompts

negative prompts:lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, quality bad, hands bad, eyes bad, face bad,normal quality, jpeg artifacts, signature, watermark, username, blurry, artist name

Recommended artists prompts

wlop,anmi,tianliang duohe fangdongye,ningen mame,mignon,nai3,konya karasue ,ciloranko,sho_(sho_lwlw) etc

V2 New artist style tags: ask (askzy), omone hokoma agm, fajyobore, chen bin, kani biimu, mikozin, kedama milk, rurudo, tab head, ru zai, kgt (pixiv12957613), ogipote,

New unique style: mmd style

The same tag combination may have significant differences from v1

Special effects prompts

god light

QQ group 244078420

This is a newly created QQ group for users to discuss and provide feedback to improve the use of the model.

介绍

这是一款基于Animagine XL V3的模型,拥有较为不错的表现能力,使用和Animagine XL V3大致相同的方法进行训练,如对tag进行排列。相比于Animagine XL 3 ,Heart of Apple XL在Animagine XL 3.0风格表现方面更加多元化,对一些游戏角色和画师进行了专向训练。

Heart of Apple XL目前在a100 40g的服务器上训练了8天。未来还会继续更新的,再次感谢CagliostroLab对社区的付出,同样我也要感谢阿萨姆奶茶,小苹果,青秋,zwh20081,alens,原物质,琥珀青叶,灾厄之轮,hans,星河主炮向前,等人在技术和资金或数据集帮助,还有那些鼓励我的伙伴们。

使用指南

和Animagine XL V3一样你需要对标签进行排列,像是多个艺术家prompts你想实现较好的效果你可以尝试放在最后面。并在额外增加了几个负面词quality bad, hands bad, eyes bad, face bad,normal quality,

推荐的prompts

负面 prompts:lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, quality bad, hands bad, eyes bad, face bad,normal quality, jpeg artifacts, signature, watermark, username, blurry, artist name

推荐的艺术家prompts

wlop,anmi,tianliang duohe fangdongye,ningen mame,mignon,nai3,konya karasue ,ciloranko,sho_(sho_lwlw)等

V2新增艺术家风格tag:ask (askzy),omone hokoma agm,fajyobore,chen bin,kani biimu,mikozin,kedama milk,rurudo,tab head,ru zhai,kgt (pixiv12957613),ogipote,

新的独特风格:mmd style

相同tag组合可能和v1有较大区别

特效prompts

god light

qq群244078420

这是我新建的qq群用于使用者讨论,和提出意见来改善模型的使用。

爱发电: https://afdian.net/a/willy23333123