bigslime/StableDiffusion-Img2Img

1import gradio as gr2import torch3import streamlit as st4from PIL import Image5import numpy as np6from io import BytesIO7from diffusers import StableDiffusionImg2ImgPipeline8 9device="cpu"10 11pipe = StableDiffusionImg2ImgPipeline.from_pretrained("CompVis/stable-diffusion-v1-4", use_auth_token=st.secrets['USER_TOKEN'])12pipe.to(device)13 14def resize(w_val,l_val,img):15  img = Image.open(img)16  img = img.resize((w_val,l_val), Image.Resampling.LANCZOS)17  #img = img.resize((value,value), Image.Resampling.LANCZOS)18  return img19 20 21def infer(source_img, prompt, guide, steps, seed, Strength): 22    generator = torch.Generator('cpu').manual_seed(seed)     23    source_image = resize(768, 512, source_img)24    source_image.save('source.png')25    image_list = pipe([prompt], init_image=source_image, strength=Strength, guidance_scale=guide, num_inference_steps=steps)26    images = []27    safe_image = Image.open(r"unsafe.png")28    for i, image in enumerate(image_list["sample"]):29        if(image_list["nsfw_content_detected"][i]):30            images.append(safe_image)31        else:32            images.append(image)    33    return image34 35gr.Interface(fn=infer, inputs=[gr.Image(source="upload", type="filepath", label="Raw Image"), gr.Textbox(label = 'Prompt Input Text'),36    gr.Slider(2, 15, value = 7, label = 'Guidence Scale'),37    gr.Slider(10, 50, value = 25, step = 1, label = 'Number of Iterations'),38    gr.Slider(39        label = "Seed",40        minimum = 0,41        maximum = 2147483647,42        step = 1,43        randomize = True), gr.Slider(label='Strength', minimum = 0, maximum = 1, step = .05, value = .5)44    ], outputs='image').queue(max_size=10).launch(enable_queue=True)45