This was trained on AnythingV5 in the "realistic" mode (assumingly it uses the regulation class images of real life people photos), resulting in a closer to 3D look. (Cue the meme about seiyuu being 2.5D)
This was trained on AnythingV5 in the "realistic" mode (assumingly it uses the regulation class images of real life people photos), resulting in a closer to 3D look. (Cue the meme about seiyuu being 2.5D)