This is an experimental LoRA trained for the Anima model. I trained it using a mixed dataset of anime-style artwork and synthetically generated photorealistic. The resulting LoRA can generate both anime-style and realistic images.
However, the realistic outputs may not always look fully photorealistic. This LoRA is primarily an interesting experimental project rather than a model specifically optimized for photorealism.
To generate realistic images, use the following trigger words:
realistic, photorealistic, photo background
To generate anime-style images, remove the tags above. You can also specify @azzai or @orico to guide the anime style.

