It works alright? I am still not satisfied with the style loras I have made so far with anima. I want to be as close as possible. If anyone has any pointers in certain settings that gives better result, please let me know.
THIS IS ANIMA, NOT ILLUSTRIOUS OR NOOBAI! It is a different model and it won't work on those checkpoints/models!
Description
FAQ
Comments (8)
He uses AI too, but I still don't know what machine and Lora he uses to make it look like an original painting.
yeah, I really liked the style of it. It might be NovelAI but that is just a guess. It could be something else.
@AkioAI Yes, I think so too. From the style of the AI images created previously, he used NovelAI. The problem is that I have never used NovelAi because it is a paid service. 😭
Ah man, I was so excited thinking it was for IL :( Looks absolutely great, but I've no clue what Anima is... Now I need to look this new thing up haha
In the meantime any chance for IL? If not, thanks anyways! :D
I can try training it on IL or Noobai but I mainly want to try training on anima for a moment. Anima is a model that can do text (not perfect) and it can use natural and danbooru tags as well: https://huggingface.co/circlestone-labs/Anima
(It only just got released recently)
There are some people who have trained some styles on it but not many. It is still a preview model so the full model isn't released yet. I have also have personal stuff going on so it may take a while (a few weeks maybe or more, depends) until I decide to train it on noobai or illustrious. At the moment I am just training styles whenever I have free time. You can try using comfyui to run anima but if you don't want to use comfyui, I think Neo forge or whatever it is called supports it.
How is Anima? I haven't switched because it doesn't seem to have many community loras or checkpoints out yet, and the ones there are seem heavily female-focused (of course).
Anima is better at poses, concepts and text compared to illustrious/noobai because it can do natural language tagging and danbooru tagging (meaning it can do '1boy, solo' and 'a boy by himself' or something. It knows more characters because it has been trained until Sep 2025 so I am sure the full model would be way better. Completely switching right now? I am not sure, it knows artists but in my opinion, when prompting an artist with the preview model, it isn't that strong compared to Noob and illustrious. I will be training more styles of gay artists anyways so...
I honestly am not sure of switching right now or not. You can have a try at using it if you haven't already. I am just messing around with it because so far I like it when using style loras on it.
Hey guys if you want to try this model online and free go to tensor art I add it there free 50 points every day I can't send the link bc they banned me