[WIP] Allow DDIMInverseScheduler to use same number of noising and denoising steps

[kathath]

For the final DDIMScheduler step, we are already at t=0 and are predicting the sample at
t=0-num_train_steps/num_inference_steps.
The current DDIMInverseScheduler implemetation starts with the original image at t=0. But to match DDIMScheduler, one needs to start at t=0-num_train_steps/num_inference_steps in order to revert all denoising steps.
This can also be seen by the fact that there are num_inference_steps denoising steps but only num_inference_steps-1 noising steps in StableDiffusionPix2PixZeroPipeline and StableDiffusionDiffEditPipeline (those two pipelines are currently using DDIMInverseScheduler).

In contrast to the current DDIMInverseScheduler implementation, in Null-text Inversion for Editing Real Images using Guided Diffusion Models inversion starts at t=0-num_train_steps/num_inference_steps and reverses all denoising steps.
In the description of DDIMInverseScheduler, it is stated that "The implementation is mostly based on the DDIM inversion definition of Null-text Inversion for Editing Real Images using Guided Diffusion Models" - which is currently not quite the case.

This PR allows to use DDIM inversion as implemented in Null-text Inversion for Editing Real Images using Guided Diffusion Models, when setting revert_all_steps=True.

This PR is strongly influenced by https://github.com/google/prompt-to-prompt/#null-text-inversion-for-editing-real-images.

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint.

[patrickvonplaten]

Hmm @clarencechen wdyt here? It essentially reverts the PR here: #2619

[clarencechen]

@patrickvonplaten I wouldn't say it reverts #2619, since the DDIMInverseScheduler implementation before then was also dropping the last (most noisy) timestep, but it does revert the semantic interpretation of the timestamp argument away from "timestamp corresponding to current noise level of input", which I personally strongly prefer. I think I would wait until a fine-tuned model addressing #3475 is out, since the current timestep I'm skipping corresponds to the final steps_offset training timesteps of the DDIMScheduler denoising process.

[bghira]

i have uploaded ptx0/pseudo-journey-v2 which is trained using a patched noise schedule. it has 30,000 steps applied over most of the text encoder for SD 2.1-v over about 30,000 images.

[patrickvonplaten]

Left some comments inline

[patrickvonplaten]

Why do we have to do the renaming here? This is backwards breaking IMO

[patrickvonplaten]

is this ever used?

[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.