Models>

judau and ple (GUNDAM ZZ) LORA

masterpiece,2 people,
<lora:judauandple001:0.7>,
ADDBASE,
judauashta,1boy,spoon,holding,
ADDCOL,
elpeople,1loli,closed eyes,open mouth,parfait,
masterpiece,2 people,open kafue,
<lora:judauandple001:0.7>,
ADDBASE,
judauashta,1boy,
ADDCOL,
elpeople,1loli,
masterpiece,2 people,outdoors,
<lora:judauandple001:0.7>,
ADDBASE,
judauashta,1boy,
ADDCOL,
elpeople,1loli,
Quick Generate
Select a style, write what you want to see and click Generate
What you want in the image.
Clear
Random
* This model is available for use and will generate similar character to the ones seen in the images.

Model Description by Creator

テストLORAです、キャラクターの出来は追及していません

キャラクターの身長差を学習出来るかのテスト

結果から言えば可能だが問題多数。

It's a test LORA, I haven't pursued character performance

Test whether you can learn the height difference of the character

Judging from the results, it is possible, but there are many problems.

*******************************

白背景の512x512のキャンバスにfull body, standing,状態

judau高さ500、、ple高さ400とした。

キャラクター外見、衣装は固定として全てのタグを消去、

キャラクター教師画像は双方5枚、close-up,upper body,cowboy shot, full body, おまけ1枚

身長差を考慮したのは full body, のみ。

他数枚、身長差がわかる2人同時の画像をキャラクターフォルダ以外で学習させた。(こちらは固定せず全てのタグを残した)

Full body, standing, on a 512x512 canvas with a white background

The judau height is 500, and the ple height is 400.

Erase all tags with character appearance and costume fixed,

5 character teacher images, close-up, upper body, cowboy shot, full body, 1 bonus

Height difference is taken into account only for full body.

In addition, several images of two people at the same time that show the height difference were learned outside the character folder. (This is not fixed and all tags are left)

*******************************

LORA単体使用だとキャラが混じり再現不能、,Regional Prompter使用前提の手法。

背景のみの画像も、そのフォルダーで同時に学習させ、アニメの様にキャラと背景をAIが合成してくれないか期待したが上手くいかなかった。

キャラクターフォルダ以外で学習させたアクションも、トリガーを使わずに何となく再現してくれるが、トリガーを入れ固定すると画像が崩壊する。

背景を詳細指定しようとすると崩壊、キャラクターにアクションさせようとタグを増やすと崩壊

(この辺は教師画像の少なさに問題があるのかもだが・・・)

学習キャラを増やした場合の挙動が不明

等の問題はあるが、以前に行っていたRegional Prompterで画面を細切れにしてタグを打ち込んでいく作業から比べれば楽に身長差が再現出来ると思われる。

If you use LORA alone, the characters will be mixed and cannot be reproduced.Method based on using Regional Prompter.

I had hoped that the background-only image would also be learned in that folder at the same time, and that the AI would synthesize the character and the background like in anime, but it didn't work.

Actions learned outside of the character folder are also reproduced without using the trigger, but the image collapses when the trigger is inserted and fixed.

Collapses when you try to specify the background in detail, collapses when you add tags to make the character take action

(There may be a problem with the lack of teacher images around here...)

Behavior when increasing the number of learning characters is unknown

Although there are problems such as this, it seems that the height difference can be reproduced easily compared to the work of dividing the screen into small pieces and typing in tags with Regional Prompter, which was done before.

*******************************

weight0.7、ADetailerを使用しface_yolov8nで顔修正を行っています。

以下、トリガー

Weight 0.7, ADetailer is used and face_yolov8n is used for face correction.

Below, trigger,

*******************************

judauashta,1boy,

elpeople,1loli,

Images Generated With This Model

Loading...

Train Models

Train AI Models On Your Own Pictures

You can use your own pictures to train AI models. Reimagine yourself in new ways with our AI-powered image generator.
Create pictures of yourself around the world, in your dream outfit, as a video game character, or in countless art styles. The possibilities are endless.
Train Your Own Model

Generate AI Images. Simple and Easy

AIEasyPic is the #1 AI image generator. Generate images with simple prompts, swap faces, train models to reimagine yourself, and more.