Interested why SD 2 was chosen architecture for this model
Congrats on the model this is exactly what Ive been looking for. Curious as to why you chose SD 2 for the base architecture vs another?
Hi! I only have 48GB GPUs, and generating six images at once is somewhat computationally expensive. So there aren't really many options. Another minor reason is to be consistent with Marigold which uses SD2.
Oh nice makes sense! Well super excited to try it out once you release the final weights!
I tried to run it in Google Colab but I encountered the following issue:
AttributeError Traceback (most recent call last)
/tmp/ipython-input-647202517.py in <cell line: 0>()
1 from diffusers import DiffusionPipeline
2
----> 3 pipe = DiffusionPipeline.from_pretrained("KevinHuang/DreamCube")
4
5 prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
5 frames
/usr/local/lib/python3.12/dist-packages/huggingface_hub/utils/_validators.py in _inner_fn(*args, **kwargs)
112 kwargs = smoothly_deprecate_use_auth_token(fn_name=fn.name, has_token=has_token, kwargs=kwargs)
113
--> 114 return fn(*args, **kwargs)
115
116 return _inner_fn # type: ignore
/usr/local/lib/python3.12/dist-packages/diffusers/pipelines/pipeline_utils.py in from_pretrained(cls, pretrained_model_name_or_path, **kwargs)
831 " is neither a valid local path nor a valid repo id. Please check the parameter."
832 )
--> 833 cached_folder = cls.download(
834 pretrained_model_name_or_path,
835 cache_dir=cache_dir,
/usr/local/lib/python3.12/dist-packages/huggingface_hub/utils/_validators.py in _inner_fn(*args, **kwargs)
112 kwargs = smoothly_deprecate_use_auth_token(fn_name=fn.name, has_token=has_token, kwargs=kwargs)
113
--> 114 return fn(*args, **kwargs)
115
116 return _inner_fn # type: ignore
/usr/local/lib/python3.12/dist-packages/diffusers/pipelines/pipeline_utils.py in download(cls, pretrained_model_name, **kwargs)
1546
1547 # retrieve passed components that should not be downloaded
-> 1548 pipeline_class = _get_pipeline_class(
1549 cls,
1550 config_dict,
/usr/local/lib/python3.12/dist-packages/diffusers/pipelines/pipeline_loading_utils.py in _get_pipeline_class(class_obj, config, load_connected_pipeline, custom_pipeline, repo_id, hub_revision, class_name, cache_dir, revision)
481 class_name = class_name[4:] if class_name.startswith("Flax") else class_name
482
--> 483 pipeline_cls = getattr(diffusers_module, class_name)
484
485 if load_connected_pipeline:
/usr/local/lib/python3.12/dist-packages/diffusers/utils/import_utils.py in getattr(self, name)
944 value = getattr(module, name)
945 else:
--> 946 raise AttributeError(f"module {self.name} has no attribute {name}")
947
948 setattr(self, name, value)
AttributeError: module diffusers has no attribute DreamCubeDepthPipeline
