This LoRA can either make a realistic scene into a 3D-Style scene, or make a 3D scene into a photorealistic scene.

Works both in I2I and T2I.

Note: This is a concept LoRA, not a *style* LoRA. The model knows what "cinematic animation" is, the LoRA just plays on these weights. The training dataset does not contain any 3D style images.

The sample weights range from 0 to 2.5 (full effect). The last sample is a 3D-Style image generated with the slider, and reverted back to realism.

Hope you like it!