For best outputs (both versions):
Sampler: DPM++ 2M Karras.
Steps: 40, cfg 8.5, CLIP skip 2.
VAE: clearvae_v2.3_fp16.safetensors (optional for less saturated outputs).
Pos prompt: include the LoRa "FaceBeauty_qinglong_V3" with a weight of 1 at the end of your prompt (included in my example runs).
Neg prompt: "(lowres, worst quality, bad quality:1.3), ((watermark, out of frame)), disfigured, deformed, distorted, disjointed, contorted, conjoined, mangled, merged, blurred, smudged, panty, bad-hands-5", being the "bad-hands-5" embedding included in my example runs.
What changes from v3.0 to v3.5: input_image feature.
Checkpoint Merge.
.png)
.png)
.png)
.png)
.png)
For best outputs (both versions):
Sampler: DPM++ 2M Karras.
Steps: 40, cfg 8.5, CLIP skip 2.
VAE: clearvae_v2.3_fp16.safetensors (optional for less saturated outputs).
Pos prompt: include the LoRa "FaceBeauty_qinglong_V3" with a weight of 1 at the end of your prompt (included in my example runs).
Neg prompt: "(lowres, worst quality, bad quality:1.3), ((watermark, out of frame)), disfigured, deformed, distorted, disjointed, contorted, conjoined, mangled, merged, blurred, smudged, panty, bad-hands-5", being the "bad-hands-5" embedding included in my example runs.
What changes from v3.0 to v3.5: input_image feature.
Checkpoint Merge.
