app.py

from diffusers import StableDiffusionPipeline
import torch
work_around_for_hugging_face_gradio_sdk_bug = "/blob/main/rem_3k.ckpt"
model_url = "https://huggingface.co/waifu-research-department/Rem" + work_around_for_hugging_face_gradio_sdk_bug
pipeline = StableDiffusionPipeline.from_single_file(
    model_url,
    torch_dtype=torch.float16,
)



import gradio as gr 


description="""

# running stable diffusion from a ckpt file

## NOTICE ⚠️: 
- this space does not work rn because it needs GPU, feel free to **clone this space** and set your own with GPU an meet your waifu **ヽ（≧□≦）ノ**


if you do not have money (just like me **(┬┬﹏┬┬)** ) you can always :
* **run the code in your PC** if you have a good GPU a good internet connection (to download the ai model only a 1 time thing)
* **run the model in the cloud** (colab, and kaggle are good alternatives and they have a pretty good internet connection ) 
### minimalistic code to run a ckpt model 
* enable GPU (click runtime thenchange runtime type) 
* install the following libraries
```
!pip install -q diffusers gradio omegaconf
```
* **restart your kernal** 👈 (click runtime then click restart session)
* run the following code
```python

from diffusers import StableDiffusionPipeline
import torch
pipeline = StableDiffusionPipeline.from_single_file(
    "https://huggingface.co/waifu-research-department/Rem/blob/main/rem_3k.ckpt", # put your model url here
    torch_dtype=torch.float16,
).to("cuda")

postive_prompt = "anime girl prompt here"  # 👈 change this
negative_prompt = "3D" 👈  things you hate here
image = pipeline(postive_prompt,negative_prompt=negative_prompt).images[0]

image # your image is saved in this PIL variable
```
"""


log = "GPU available"
try :
  pipeline.to("cuda")
except : 
  log = "no GPU available"


def text2img(positive_prompt,negative_prompt):
    if log == "no GPU available":
        image = None
        return log,image
    else : 
        image = pipeline(positive_prompt,negative_prompt=negative_prompt).images[0]
        log = {"postive_prompt":positive_prompt,"negative_prompt":negative_prompt}
        return log,image

gr.Interface(text2img,["text","text"],["text","image"],examples=[["rem","3D"]],description=description).launch()