The basic model used is Realistic Vision V6.0 B1.99 images were trained. From the loss curve, it seems that the convergence is not very good. We will try to optimize it by adjusting the parameters later.<h3 id="recommended-weight0.8~1-hk52en6vs">Recommended weight：0.8~1</h3>prompt:My_villa, building, no humans, tree, window, glass, light, grass, plant, water, sky, day, scenery, rock, stairs, bush

architecture-villa

architecture V1.0.safetensors

v1.0

It's still not ideal. When the loss curve drops rapidly, I feel bad.I don't quite understand how to control the generalization of the picture, and I don't know how to train the style I want. Please guide me.Recommended weight：1

Files with SHA256: a78bb0b4b5b5046e0247c82d5c4251cfc0993926ade3144fe8464f398cccf8a2