'StableDiffusionPipeline' object has no attribute 'tiled'

[entangledloops]

The requirements.txt seem to be missing an install for package realesrgan. After installing that, I hit the error:

  File "/home/snd/bin/miniconda3/envs/ldm/lib/python3.8/site-packages/stable_diffusion_videos/image_generation.py", line 168, in generate_images
    tiled=pipeline.tiled,
AttributeError: 'StableDiffusionPipeline' object has no attribute 'tiled'

You should probably pin the versions in requirements.txt.

[nateraw]

Looking at your error, it seems that you are using a diffusers StableDiffusionPipeline instead of the StableDiffusionWalkPipeline from this package to call that generate_images fn. That pipeline would indeed not have the tiled attribute, because it only exists in my pipeline here.

---

As for realesrgan install (which is separate issue from this one), it was originally optional, but I got lazy along the way and now it will try to import it either way. I'll just add it to requirements.txt then.

[entangledloops]

Yup, you're right. It was some Jupyter state mixup. I just reran the same code:

from stable_diffusion_videos import StableDiffusionWalkPipeline, Interface
import torch

pipe = StableDiffusionWalkPipeline.from_pretrained(
    "runwayml/stable-diffusion-v1-5", 
    torch_dtype=torch.float16, 
    revision="fp16"
)
device = "cuda"
pipe.safety_checker = lambda images, clip_input: (images, False)
pipe = pipe.to(device)

interface = Interface(pipe)
interface.launch()

And it works this time. Closing the issue.