Type | |
Stats | 1,196 |
Reviews | (207) |
Published | Jul 22, 2023 |
Base Model | |
Usage Tips | Clip Skip: 2 |
Training Images | Download |
Hash | AutoV2 8DB017A8B5 |
This is the LyCORIS version of the previously released LoRA version. While using the same dataset it generally performs better yielding higher quality output while being more flexible with outfits.
How to use
Just prompt away. The dataset is based on official character art with "simple background" and "white background" so use those tags if you want to strengthen the style further. You can also add standard tag for Disgaea (Senkai Makai Disgaea) to further change the style from the softer official character art towards the more "pointy" version provided the model has data on it.
Other things to try for AOM2 is using "3D" and "realistic" as negative tags though that shouldn't be necessary normally.
For more prompt suggestions see the sample images or the zip of training tags.
Note also that the end results depend on the sampler used, some producing more clean and polished results and others more close to drawings, experiment to see what works best with your model and goals.
While it's possible to prompt some characters from the game through tags, especially if you use their name too, there are now specialized LoRAs here on CivitAI for most of them if you want the best accuracy 😊
Don't forget to review and upload your own images, it's always fun to see what you're creating!
Misc
Trained on NAI at 512 with 165 images at 512 resolution. Sample images from AOM2_NSFW
All images are untouched txt2img upscaled, no inpainting or helper embeddings used for quality transparence, e.g. - things should look even better if you do.