app.py

import gradio as gr
#import torch
#from torch import autocast // only for GPU

from PIL import Image
import numpy as np
from io import BytesIO
import os
MY_SECRET_TOKEN=os.environ.get('HF_TOKEN_SD')

from diffusers import StableDiffusionImg2ImgPipeline

print("hello sylvain")

YOUR_TOKEN=MY_SECRET_TOKEN

device="cpu"

#prompt_pipe = StableDiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", use_auth_token=YOUR_TOKEN)
#prompt_pipe.to(device)

img_pipe = StableDiffusionImg2ImgPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", use_auth_token=YOUR_TOKEN)
img_pipe.to(device)

source_img = gr.Image(source="upload", type="filepath", label="init_img | 512*512 px")
gallery = gr.Gallery(label="Generated images", show_label=False, elem_id="gallery").style(grid=[2], height="auto")

def resize(value,img):
  #baseheight = value
  img = Image.open(img)
  #hpercent = (baseheight/float(img.size[1]))
  #wsize = int((float(img.size[0])*float(hpercent)))
  #img = img.resize((wsize,baseheight), Image.Resampling.LANCZOS)
  img = img.resize((value,value), Image.Resampling.LANCZOS)
  return img


def infer(prompt, source_img): 
         
    source_image = resize(512, source_img)
    source_image.save('source.png')
    images_list = img_pipe([prompt] * 2, init_image=source_image, strength=0.75)
    images = []
    safe_image = Image.open(r"unsafe.png")
    for i, image in enumerate(images_list["sample"]):
        if(images_list["nsfw_content_detected"][i]):
            images.append(safe_image)
        else:
            images.append(image)    
    return images

print("Great sylvain ! Everything is working fine !")

title="Img2Img Stable Diffusion CPU"
description="Img2Img Stable Diffusion example using CPU and HF token. <br />Warning: Slow process... ~5/10 min inference time. <b>NSFW filter enabled.</b>" 

gr.Interface(fn=infer, inputs=["text", source_img], outputs=gallery,title=title,description=description).queue(max_size=100).launch(enable_queue=True)
#from torch import autocast
#import requests
#import torch
#from PIL import Image
#from io import BytesIO
#import os
#MY_SECRET_TOKEN = os.environ.get('HF_TOKEN_SD')

#from diffusers import StableDiffusionImg2ImgPipeline

#YOUR_TOKEN = MY_SECRET_TOKEN
# load the pipeline
#device = "cuda"
#model_id_or_path = "CompVis/stable-diffusion-v1-4"

# pipe = StableDiffusionImg2ImgPipeline.from_pretrained("CompVis/stable-diffusion-v1-4", use_auth_token = YOUR_TOKEN)

#pipe = StableDiffusionImg2ImgPipeline.from_pretrained(
#    model_id_or_path,
#    revision="fp16", 
#    torch_dtype=torch.float16,
#    use_auth_token=YOUR_TOKEN
#)
# or download via git clone https://huggingface.co/CompVis/stable-diffusion-v1-4
# and pass `model_id_or_path="./stable-diffusion-v1-4"` without having to use `use_auth_token=True`.
#pipe = pipe.to(device)

# let's download an initial image
#url = "https://raw.githubusercontent.com/CompVis/stable-diffusion/main/assets/stable-samples/img2img/sketch-mountains-input.jpg"

#response = requests.get(url)
#init_image = Image.open(BytesIO(response.content)).convert("RGB")
#init_image = init_image.resize((768, 512))

#prompt = "Lively, illustration of a [[[<king::4>]]], portrait, fantasy, intricate, Scenic, hyperdetailed, hyper realistic <king-hearthstone>, unreal engine, 4k, smooth, sharp focus, intricate, cinematic lighting, highly detailed, octane, digital painting, artstation, concept art, vibrant colors, Cinema4D, WLOP, 3d render, in the style of hearthstone::5 art by Artgerm and greg rutkowski and magali villeneuve, martina jackova, Giger"

#with autocast("cuda"):
#    images = pipe(prompt=prompt, init_image=init_image, strength=0.75, guidance_scale=7.5).images

#images[0].save("fantasy_landscape.png")