For best outputs:
Sampler: DPM++ SDE Karras.
Steps: 30, cfg 7, CLIP skip 2.
VAE: vae-ft-mse-840000-ema-pruned_fp16.safetensors
Pos prompt: include the LoRa's "mix4" with a weight of 1 and "more_details" with a weight of 0.8 at the end of your prompt (included in my example runs).
Neg prompt: "(worst quality, low quality, cgi, bad eye, worst eye, illustration, cartoon), deformed, distorted, disfigured, poorly drawn, bad anatomy, wrong anatomy, (normal quality:2), lowres, bad hands, normal quality, ((monochrome)), ((grayscale)), EasyNegativeV2, paintings, sketches, bad-hands-5", being the "EasyNegativeV2" and "bad-hands-5" embeddings included in my example runs.
What changes from v1.0 to v2.0: input_image feature.
Checkpoint Merge.
For best outputs:
Sampler: DPM++ SDE Karras.
Steps: 30, cfg 7, CLIP skip 2.
VAE: vae-ft-mse-840000-ema-pruned_fp16.safetensors
Pos prompt: include the LoRa's "mix4" with a weight of 1 and "more_details" with a weight of 0.8 at the end of your prompt (included in my example runs).
Neg prompt: "(worst quality, low quality, cgi, bad eye, worst eye, illustration, cartoon), deformed, distorted, disfigured, poorly drawn, bad anatomy, wrong anatomy, (normal quality:2), lowres, bad hands, normal quality, ((monochrome)), ((grayscale)), EasyNegativeV2, paintings, sketches, bad-hands-5", being the "EasyNegativeV2" and "bad-hands-5" embeddings included in my example runs.
What changes from v1.0 to v2.0: input_image feature.
Checkpoint Merge.
