Windows - It breaks after 'dream'

[HQuasar]

I followed the Windows installation guide and up until executing the 'dream' command all was fine. I also moved the contents of the 'scripts' folder to the main folder as suggested.

There's a small mistake in the guide, the pre-release weight folder should be named "text2img-large" instead of "text2img.large", since it keeps raising an exception 'no such file or directory' if I name it "text2img.large".

Anyway, after invoking "dream.py -l" for the first time, the initialization is successful. After inputting a simple prompt, however, I get this:

• Initialization done! Awaiting your command (-h for help, q to quit)...
  dream> an astronaut riding a horse
  Traceback (most recent call last):
  File "C:\Users\xxx\stable-diffusion\dream.py", line 276, in
  main()
  File "C:\Users\xxx\stable-diffusion\dream.py", line 67, in main
  main_loop(t2i,cmd_parser,log)
  File "C:\Users\xxx\stable-diffusion\dream.py", line 114, in main_loop
  results = t2i.txt2img(**vars(opt))
  File "C:\Users\xxx\stable-diffusion\ldm\simplet2i.py", line 154, in txt2img
  model = self.load_model() # will instantiate the model or return it from cache
  File "C:\Users\xxx\stable-diffusion\ldm\simplet2i.py", line 375, in load_model
  config = OmegaConf.load(self.config)
  File "C:\Users\xxx\Conda\envs\ldm\lib\site-packages\omegaconf\omegaconf.py", line 183, in load
  with io.open(os.path.abspath(file_), "r", encoding="utf-8") as f:
  FileNotFoundError: [Errno 2] No such file or directory: 'C:\Users\xxx\configs\latent-diffusion\txt2img-1p4B-eval.yaml'

I tried creating the folder (and the .yaml file) myself. I get this error now:

• Initialization done! Awaiting your command (-h for help, q to quit)...
  dream> an astronaut riding a horse
  Loading model from models/ldm/text2img-large/model.ckpt
  Traceback (most recent call last):
  File "C:\Users\xxx\stable-diffusion\dream.py", line 276, in
  main()
  File "C:\Users\xxx\stable-diffusion\dream.py", line 67, in main
  main_loop(t2i,cmd_parser,log)
  File "C:\Users\xxx\stable-diffusion\dream.py", line 114, in main_loop
  results = t2i.txt2img(**vars(opt))
  File "C:\Users\xxx\stable-diffusion\ldm\simplet2i.py", line 154, in txt2img
  model = self.load_model() # will instantiate the model or return it from cache
  File "C:\Users\xxx\stable-diffusion\ldm\simplet2i.py", line 377, in load_model
  model = self._load_model_from_config(config,self.weights)
  File "C:\Users\xxx\stable-diffusion\ldm\simplet2i.py", line 396, in _load_model_from_config
  pl_sd = torch.load(ckpt, map_location="cpu")
  File "C:\Users\xxx\Conda\envs\ldm\lib\site-packages\torch\serialization.py", line 699, in load
  with _open_file_like(f, 'rb') as opened_file:
  File "C:\Users\xxx\Conda\envs\ldm\lib\site-packages\torch\serialization.py", line 231, in _open_file_like
  return _open_file(name_or_buffer, mode)
  File "C:\Users\xxx\Conda\envs\ldm\lib\site-packages\torch\serialization.py", line 212, in init
  super(_open_file, self).init(open(name, mode))
  FileNotFoundError: [Errno 2] No such file or directory: 'models/ldm/text2img-large/model.ckpt'

even though such folder (and file) exists.

[lstein]

I think the issue is that the files are identified using relative directory paths. When you launch the script, make sure your current directory is the same one that main.py and the subdirectories scripts, ldm, and models are in. Try "python scripts\dream.py -l"

When you created the .yaml file, you copied one from one of the stable-diffusion repositories, right?

[HQuasar]

Alright, it's working! I guess I made a noob mistake, I had to change path from the main Conda installation folder (C:\Users\xxx) to the 'stable-diffusion' folder (C:\Users\xxx\stable-diffusion) to launch the script, otherwise it was trying to find folders where they shouldn't exist. Also moving dreams.py to the 'scripts' subfolder doesn't matter, it works either way.

[lstein]

I'm glad you figured it out. I'll tweak the README to make it easier for people to follow.