Settings?
#18
by
brennanisgreat
- opened
I'm still downloading the model, but has anyone had any noteworthy success with specific settings re: sampler, scheduler, steps, etc.? I gave up on SD 3 because I could never find a good combination of settings that delivered what SD promised, and no one else I talked to did either. I've been using Ideogram for ages and love it, but I'm happy to give SD another try.
I run SD3.5 and Flux locally on my own machine. Not using Comfy or other software. I find many of the samplers / schedulers made to work for Transformer models like SD3.5 & FLUX work quite well.
Generally my inference setup:
with torch.inference_mode():
image = pipe(
prompt=text_prompt,
negative_prompt=negative_prompt,
width = 1024,
height = 1024,
guidance_scale=3.5,
generator=generator,
max_sequence_length=512,
num_inference_steps=28).images[0]
As someone posted on Reddit, Flux / SD3.5 aspect ratios to use:
aspect_ratio == '1:1 1024 x 1024':
aspect_ratio == '1:1 1408 x 1408':
aspect_ratio == '3:2 1728 x 1152':
aspect_ratio == '4:3 1664 x 1216':
aspect_ratio == '16:9 1920 x 1088':
aspect_ratio == '21:9 2176 x 960':
aspect_ratio == '2:3 1152 x 1728':
aspect_ratio == '3:4 1216 x 1664':
aspect_ratio == '9:16 1088 x 1920':