v1.6
外形の学習を強化し、キャラクターが崩壊しにくくしました。
We have strengthened the learning of external shapes and made it difficult for characters to collapse.
*********************************************
基本的にガチャ。
アニメモデルをベースにしているので、多くのアニメモデルで使用出来ます。
一部の2.5D、及び、リアルモデルでの出力も出来ますが、様々な調整が必要になると思います。
Basically a gacha.
Since it is based on an anime model, it can be used with many anime models.
Although it is possible to output some 2.5D and real models, various adjustments may be required.
*********************************************
アクションを追加する場合や、重量を下げる場合、キャラクター成分が弱まるので、"トリガーワード"に記載している外形のプロンプトから、弱くなった部分を抜粋して適用して下さい。
When adding actions or reducing weight, the character components will be weakened, so please extract and apply the weakened parts from the external form prompts listed in "Trigger Words".
*********************************************
好きな衣装や、背景、ポーズを組み合わせて下さい。
Please combine your favorite costumes, backgrounds, and poses.
*********************************************
weight 0.4～0.9、ADetailerを使用しface_yolov8nで顔修正を行っています。
高解像度補助:R-ESRGAN 4x+ Anime6B,倍率2倍
Weight 0.4 to 0.9 ADetailer is used and face_yolov8n is used for face correction.
Hires.fix:R-ESRGAN 4x+ Anime6B,2x magnification
*********************************************
"Detail Slider - LoRA"
https://civarchive.com/models/153562?modelVersionId=171989
"Detail Tweaker LoRA (细节调整LoRA)"
https://civarchive.com/models/58390?modelVersionId=62833
"noline"
https://civarchive.com/models/81309/noline
"boldline"
https://civarchive.com/models/81313/boldline
"Zoom Slider - LoRA"
https://civarchive.com/models/114460?modelVersionId=123732
"ADetailer"
(インストールは自己責任でお願いします)(Please install at your own risk.)
顔の詳細を修正する追加機能
Additional features to correct facial details
https://github.com/Bing-su/adetailer
Embeddings
"Fast Negative Embedding (+ FastNegativeV2)"
https://civarchive.com/models/71961?modelVersionId=94057
"EasyNegative"
https://civarchive.com/models/7808/easynegative
"negative_hand Negative Embedding"
https://civarchive.com/models/56519/negativehand-negative-embedding
**********************************************
@KXRO
"Project KR4X - 2.5D / AaYMix"
https://civarchive.com/models/75088/project-kr4x-25d-aaymix
**********************************************
v1.6
外形の学習を強化し、キャラクターが崩壊しにくくしました。
v1.5
アクション追加で、容易にキャラクターが崩壊する不具合があります、ダウンロードする必要は有りません。
v1.1 to v1.2
一般モデルに対応する為と、バグ除去の為の途中経過です、ダウンロードする必要は有りません。
v1
非常に高画質で顔のアップにも非常にクリアに対応出来ますが、再生環境は非常に限定的で"kantanmix"か、2.5Dモデルでの出力にしか対応していません
v1.6
We have strengthened the learning of external shapes and made it difficult for characters to collapse.
v1.5
There is a problem where the character easily collapses when adding an action. There is no need to download.
v1.1 to v1.2
This is a work in progress to support general models and to remove bugs, so there is no need to download it.
v1
It has very high image quality and can handle close-ups of faces very clearly, but the playback environment is very limited and only supports output in "kantanmix" or 2.5D models.
Description
画像をさらに排除
烏排除
一部トリガー排除
FAQ
Comments (4)
Just a question: why this LORA works so well with semirealistic checkpoints?
Did you changed your workflow, training it?
I'm pretty happy with Kurama while i'm using it with a semirealistic model. But there is the problem about all other charaters from Urusei Yatsura are "anime" looking. So now the style of Kurama didnt' match with them.
May be you will separate the experiment, keeping a "semirealistic Kurama" and also a "anime Kurama" version?
@settima_ai
Currently, along with bug removal in KuramaLORA, we are looking for improvements to make it compatible with various general models. If this works well, it may be possible to support various models in future learning.
現在、KuramaLORAのバグ除去に合わせて、一般的な様々なモデルに適合する様に改良を模索中です。
これが上手く機能した場合、今後の学習では、様々なモデルに対応出来る可能性があります。
Advice:
since the beginnig the version 001 of Kurama was incontrollable, but also very powerfull and creative. It was refusing to perform simple requests but had the tendency to "invent" a lot of strange details. It was like it has a devil inside!
While using it, I called it "the MidJourney" of the LORA, because of it's simbolic power!
Unfortunately it was uncontrollable.
The version 003 is much more controllable, but it retains a sort of rich immagination, sometime.
Dont' eliminate this, if possible! : )
@settima_ai
v1.2 has a fatal flaw in character generation, so I think it will be fixed in a later version, but I will leave v1.2 as is without deleting it.
v1.2は、キャラクター生成に致命的な欠陥が有るので、その後のバージョンで修正されると思いますが、ｖ1.2は消さずにそのまま置いておきます。
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.