Is the v1.1 8-step LoRA actually v1.1 4-step?
It looks and feel as though it was intended for 4-step generation (or at least meant to be replacing the 4-step LoRA).
A close-up portrait of a dog with black, brown, and white fur, a white stripe on its forehead, and brown and black markings on its ears, is looking directly at the camera with a serious expression. The dog has brown eyes with black pupils and a black nose, and its ears are large and pointed. The background is blurred and appears to be an outdoor setting with green and brown grass and a light grey sky.
v1.0 4-step lora | 4 steps vs. 8 steps vs. 16 steps | [1536x1536, no shift, lora strength 1, deis, kl_optimal, seed 187]
v1.0 8-step lora | 4 steps vs. 8 steps vs. 16 steps | [1536x1536, no shift, lora strength 1, deis, kl_optimal, seed 187]
v1.1 8-step lora | 4 steps vs. 8 steps vs. 16 steps | [1536x1536, no shift, lora strength 1, deis, kl_optimal, seed 187]
Hi, 8steps V1.1 can work with 4-step inference but it's not expected to generate better results in 4-step setting.
In our settings, the 8step lora works better with 8-step inference.
May you use the workflow here for inference? Just select the lora_path to 8-step V1.1 model and change the number of steps, leaving other settings unchanged, including shift value (3.0), scheduler (euler).