Download
1 variant available
Trained on Anime v1.0. Added some other EVA artwork into the training mix.
Trained on 1919 images for 4000 steps, resumed for another 2700 steps. Batch size 4, gradient accumulation step 2 (so effectively 8 images per step).
Training data (images, captions) in zip download.
(Righth after training I noticed that some images were not processed right in the training set. The zip contains fixed images. Not planning to retrain anytime soon though)
Show more

1690 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
1.2K0 1 2 3 4 5 6 7 8 9.0 1 2 3 4 5 6 7 8 9K
License:
AnimaThe Anima Model is licensed by CircleStone Labs LLC. Copyright CircleStone Labs LLC. IN NO EVENT SHALL CIRCLESTONE LABS LLC BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL.
Built on NVIDIA Cosmos
Style lora for Neone Genesis Evangelion.
Almost all characters were tagged. But only the ones with lots of images work reliably.
Different Evangelion series were tagged as "@evatv", "@endev", "@reb1", "@reb2", "@reb3", "@reb4". These can be used to somewhat steer the style. "anime screenshot" is also powerful word to use.
I also tested to see if I could get different plugsuit variants to be learned. So They were all labeled like "plugsuit 00", "plugsuit 01" etc.
It turns out that the model learned to do the label on the plugsuit... So prompting for "plugsuit 03" will get you a plugsuit with "03" on its chest. Not massively useful but it is something...
The model learned to do complex backgrounds. Unfortunately it did not learn to do them well. So backgrounds often get a bit messed up.
This is Anima version.
Older Illustrious version: https://civitai.red/models/1167353/neon-genesis-evangelion-style-and-characters
