Works best with SD 1.5 based checkpoints or as a standalone LoRA . Recommended Settings: Steps: 20-30 CFG Scale: 5-7.5 Sampler: DPM++ 2M Karras Clip Skip: 2
Just pushed the latest PyTorch weights for the Berry-Real project. Berry-Real-3.0.pt
Start your prompt with photo of [subject], berry-real style for best results. Works best with SD 1
Move the file into your /stable-diffusion-webui/models/Stable-diffusion folder (or your LoRA folder if specified). here is a quick setup guide:
Handles complex shadows and HDR-like environments much more effectively.
For those asking how to get the most out of the new Berry-Real-3.0.pt file, here is a quick setup guide: