"Torch not compiled with CUDA enabled"

[gerroon]

Hi

I believe I installed according to the doc but hitting a dead road here when I try to generate something :(

Error: Python: Traceback (most recent call last):
  File "C:\Users\USER\AppData\Roaming\Blender Foundation\Blender\4.0\scripts\addons\Pallaidium\__init__.py", line 2765, in execute
    sequencer.generate_movie()
  File "c:\blender-4.0\4.0\scripts\modules\bpy\ops.py", line 109, in __call__
    ret = _op_call(self.idname_py(), kw)
RuntimeError: Error: Python: Traceback (most recent call last):
  File "C:\Users\USER\AppData\Roaming\Blender Foundation\Blender\4.0\scripts\addons\Pallaidium\__init__.py", line 1653, in execute
    video_frames = upscale(
  File "c:\blender-4.0\4.0\python\lib\site-packages\torch\utils\_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "C:\Users\USER\AppData\Roaming\Python\Python310\site-packages\diffusers\pipelines\text_to_video_synthesis\pipeline_text_to_video_synth_img2img.py", line 648, in __call__
    prompt_embeds, negative_prompt_embeds = self.encode_prompt(
  File "C:\Users\USER\AppData\Roaming\Python\Python310\site-packages\diffusers\pipelines\text_to_video_synthesis\pipeline_text_to_video_synth_img2img.py", line 328, in encode_prompt
    text_input_ids.to(device),
  File "c:\blender-4.0\4.0\python\lib\site-packages\torch\cuda\__init__.py", line 239, in _lazy_init
    raise AssertionError("Torch not compiled with CUDA enabled")
AssertionError: Error: Python: Traceback (most recent call last):
  File "C:\Users\USER\AppData\Roaming\Blender Foundation\Blender\4.0\scripts\addons\Pallaidium\__init__.py", line 2765, in execute
    sequencer.generate_movie()
  File "c:\blender-4.0\4.0\scripts\modules\bpy\ops.py", line 109, in __call__
    ret = _op_call(self.idname_py(), kw)
RuntimeError: Error: Python: Traceback (most recent call last):
  File "C:\Users\USER\AppData\Roaming\Blender Foundation\Blender\4.0\scripts\addons\Pallaidium\__init__.py", line 1653, in execute
    video_frames = upscale(
  File "c:\blender-4.0\4.0\python\lib\site-packages\torch\utils\_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "C:\Users\USER\AppData\Roaming\Python\Python310\site-packages\diffusers\pipelines\text_to_video_synthesis\pipeline_text_to_video_synth_img2img.py", line 648, in __call__
    prompt_embeds, negative_prompt_embeds = self.encode_prompt(
  File "C:\Users\USER\AppData\Roaming\Python\Python310\site-packages\diffusers\pipelines\text_to_video_synthesis\pipeline_text_to_video_synth_img2img.py", line 328, in encode_prompt
    text_input_ids.to(device),
  File "c:\blender-4.0\4.0\python\lib\site-packages\torch\cuda\__init__.py", line 239, in _lazy_init
    raise AssertionError("Torch not compiled with CUDA enabled")
AssertionError: Torch not compiled with CUDA enabled
Location: c:\blender-4.0\4.0\scripts\modules\bpy\ops.py:109
Location: c:\blender-4.0\4.0\scripts\modules\bpy\ops.py:109

[tin2tin]

You properly had another version of Torch installed. The first thing you can try is to click the Uninstall Dependencies button, restart Blender, and the click Install. For other things to try, read this thread: #40

Let me know what/if it works for you.

[gerroon]

I have torch installed for other purposes but I am sure they are CUDA enabled.

[tin2tin]

The needed python modules can't find Torch with CUDA, so unless you try to update it, I'm out of suggestions.

It is super tricky to install Torch. I spend two weeks to get it working on my laptop when I started out with this add-on. So, you'll have to start digging into a solution which will work on your setup. The add-on mentioned on the main page to install and remove python models is very convenient.

[gerroon]

Ok it looks like uninstall dependencies and reinstalling them worked. Thanks for the help.

[tin2tin]

Oh, that's great news! Thank you for letting me know.

If you make something cool with the add-on, I would love to see it.

[gerroon]

I am loving it thanks. Couple wishes so far

Possible to replace the current frame of a strip with a newly generated image?

Ability to add missing frames, as in AI supported optical flow thing

thanks