Possible bug in pipelines that use denoising_start and denoising_end

[a-r-r-o-w]

Here are a few examples of what I think is a possible bug.

diffusers/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_img2img.py

Line 1277 in 6ae7e81

denoising_start=self.denoising_start if denoising_value_valid else None,

diffusers/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_inpaint.py

Line 1516 in 6ae7e81

denoising_start=self.denoising_start if denoising_value_valid else None,

diffusers/examples/community/lpw_stable_diffusion_xl.py

Line 1777 in 6ae7e81

denoising_start=self.denoising_start if denoising_value_valid else None,

We can see that denoising_value_valid is defined, but not called, and is used in determining whether self.denoising_start or None is passed into self.get_timesteps.

# ...
        timesteps, num_inference_steps = self.get_timesteps(
            num_inference_steps,
            strength,
            device,
            denoising_start=self.denoising_start if denoising_value_valid else None,
        )

However, it is not called which evaluates condition to True always. This does not seem like the intended behaviour, no?

@yiyixuxu @patrickvonplaten

[LWprogramming]

appears to be mentioned in #5685 (comment) as well but not sure if a PR for that was opened or not. as of now it looks unfixed (permalink at time of writing)

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.

Please note that issues that do not follow the contributing guidelines are likely to be ignored.

[yiyixuxu]

I believe this is a bug - do you want to open a PR to fix it?

[a-r-r-o-w]

Yes, it's been here for a while as mentioned in other issues as well. Will open one shortly