Style lora for Neone Genesis Evangelion. V1 can do Rebuild of Evangelion movies while V2 can do both movies and the TV series.

Almost all characters were tagged. But only the ones with lots of images work reliably

Characters that work well. You can get their likeness without any effort: souryuu asuka langley, ayanami rei, makinami mari illustrious, ikari shinji , katsuragi misato.

Characters that work somewhat. You need to describe their appearance in detail to get them: akagi ritsuko, ibuki maya, kitakami midori, horaki hikari, nagisa kaworu.

The model managed to learn how to do interior of the EVA cockpit, which is nice.

Note: The images in the showcase do contain prompt and metadata. Civitai just doesn't show it. Download the image and view them with metadata viewer of some sort if you want the prompts.

V3:
Includes TV series, moves 1, 2, 3, 3+1 and EoE (1174 images) and other Eva artowrk (281 images). Also includes images from other anime to stabilize general anime style: Konosuba (50 images), Monogatari (93 images) and yozakura quartet (44 images).

All Evangelion images were tagged as "evangelion", Different Evangelion series were tagged as "evatv", "reb1", "reb2", "reb3", "reb4" and "endev". All anime screenshots were tagged as "anime screenshot".

Larger quantity of varied characters were included.

Trained on illustriousXL v1.0 for 14 epochs.
V3 was trained as a LoCon.
Training data is available in the file downloads.

V2:
Includes TV series (574 images), moves 1 (180 images), 2 (118 images), 3 (96 images), and 3+1 (213 images). Trained on illustriousXL v0.1 for 9 epochs.
Training data is available in the file downloads.

They were tagged as "anime screenshot", "evatv", "reb1", "reb2", "reb3", and "reb3_1".

Larger quantity of varied characters were included.

V1:
Includes moves 1 (180 images), 2 (118 images), 3 (96 images), and 3+1 (213 images).

Useful keywords:

Each movie had separate tag in captions. You can use "eva_1", "eva_2", "eva_3" and "eva3_1" to slightly nudge the style/character to better align with specific movie.

Including "(anime screencap:0.7)" helps in avoiding situation where you get face in style of the movies but body in a general style.

All images were tagged as "evangelion" but that doesn't seem to be helpful keyword to use.

Training details:

600 screenshots. Trained for 10 epochs on illustriousXL_v01.