Anima Tuned Model (Patreon)

Okay, friends, I just finished an incredibly thorough training session; I think it's the most complex one I've ever done. I used a dataset of almost 10,000 images to refine the Anima model. The goal was to give the prompts more consistency and improve the lighting and expressiveness. Anima has a tendency to change atmospheric lighting from one image to another, which is also something I tried to improve, although I haven't done too many tests. Here's a small sample showing the difference. Both images have the same prompt and seed; the first is with the base style and model I've been using, and the second is with the new training. I look forward to your feedback.