use multiple community pipes in a list

[CrazyBoyM]

Is your feature request related to a problem? Please describe.

I find some very great community pipes, such as CLIP Guided Stable Diffusion and Long Prompt Weighting Stable Diffusion. But I don't know how to use them both.

Describe the solution you'd like some code likes:

pipe = DiffusionPipeline.from_pretrained(
    'hakurei/waifu-diffusion',
    custom_pipelines=["lpw_stable_diffusion", "clip_guided_stable_diffusion"],
    revision="fp16",
    torch_dtype=torch.float16
)
pipe=pipe.to("cuda")

simpely change the

custom_pipeline="lpw_stable_diffusion"

to

custom_pipelines=["lpw_stable_diffusion", "clip_guided_stable_diffusion"]

Describe alternatives you've considered a list for custom pipelines

[patrickvonplaten]

Hey @CrazyBoyM,

Note that community pipelines are not always compatible with each other so I'm not sure it's possible to load them into one class. How did you image the design of a "multi-community-pipeline" to look like? :-)

[CrazyBoyM]

Hey @CrazyBoyM,

Note that community pipelines are not always compatible with each other so I'm not sure it's possible to load them into one class. How did you image the design of a "multi-community-pipeline" to look like? :-)

sorry for that i don't have a good idea for it's compatibility now

[adhikjoshi]

Hey @CrazyBoyM,

Note that community pipelines are not always compatible with each other so I'm not sure it's possible to load them into one class. How did you image the design of a "multi-community-pipeline" to look like? :-)

We can pass array of maybe downloaded models, which loads.

If i want to load 3 custom trained models, it's not possible, as far i know (tell me if i can)

If we can do it easily, it would be awesome.

[patrickvonplaten]

Also related: https://github.com/huggingface/diffusers/issues/1046

[github-actions[bot]]

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.