Model Description / モデル詳細
This model is a Anima realistic model.
I am training the model to generate Japanese people.
The trigger word is "jpl".Whether or not to include it depends on the situation.
It is currently impossible to prevent smooth skin.
日本人を出力するための学習をしています。
トリガーワードはjplです。つけるかつけないかは、状況によって変わります。
お肌ツルツル対策は今のところ無理
個人的なおすすめプロンプト前文 / Personal Recommendation Prompt Preamble
masterpiece, best quality, (realism, photorealism, amateur photo, detailed), jpl, real skin, detailed skin,
個人的なおすすめネガティブプロンプト / Personal Recommended Negative Prompt
worst quality, bad quality, low quality, score_1, score_2, score_3, blurry, jpeg artifacts, sepia, deviantart, ye-pop, bad anatomy, bad hands, fewer digits, ribs, motion lines, speed lines, watermark, signature, artist name, plump, fat, obese, monochrome, greyscale, high contrast, (anime, animation, disney), (@3dcg:0.8), mirrored, blur, AI generated
※anima純正やそれ以外もturbo loraの使用は、大きく結果が異なるためおすすめしません。
We do not recommend using anima genuine or other turbo lora as the results will be very different.
For reference only(This setting is an example)
Steps: 25, Sampler: ER SDE, Schedule type: Beta, CFG scale: 5, Shift: 12
My sample used HiresFix(Essential if making it realistic.)
HiresFixなどのアップスケールがないと2.5Dのような感じになります。
Sampling method:4x-UltraSharpV2
Hires steps:15
Denoising strength:0.15
Upscale by:1.2
About License
https://huggingface.co/circlestone-labs/Anima/blob/main/LICENSE.md
Description
model merge
FAQ
Comments (3)
V2 has a very interesting style. It features sharp lines and good prompt adherence, though the execution differs from other anima checkpoints. For my taste, it is a bit too bright and leans too heavily into the Japanese aesthetic, but that is clearly intentional. It is a very interesting checkpoint for certain use cases.
That understanding is correct.
I make everything for my own personal use and to suit my own tastes.
These are essentially just things I'm sharing with others.
I like this implementation of the prompt; it’s different but accurate. You get what you prompt for, but from a different perspective than with other checkpoints, and I think that’s good.