Claymore Style版本v1.0 (ID: 291589)

Armored girls in the style of the Claymore manga by Norihiro Yagi.
LyCORIS version of an old LoRA which may work better/worse/differently depending on what you're prompting.

How to use
No trigger word, just use prompt away. Tags that work well are, well, anything to do with girls in armor.
For anything else you likely have to adjust the LyCORIS strength to find a good balance between flexibility and quality. There was a lot of "power poses" in the dataset so hands tend to be all over unless you prompt their place.
On AOM2 I found the best results generally came in txt2img using Euler A at 30 steps and 768*512. DPM++ 2M Karras work too and tend to produce outputs leaning more towards drawings than Euler A.
For more prompt tips please see the sample images and uploaded zip with training tags.

Misc
Dataset of greyscale 79 images from Claymore Memorabilia Artbook.
Trained at 512 resolution on NAI.
All sample images except main are raw txt2img upscaled, no helper embeddings/negatives used. E.g. - things should look even better if you do.
The training was done on a small subset of images, as such the output of this LyCORIS is not an accurate representation neither in quality or style of the source material. Go buy the manga and other merch if you want the good stuff.

描述:

Initial release.

训练词语: