File size: 1,588 Bytes
e48f8e5 98f4d1a a38c0be 98f4d1a 22670e0 e56e26b 98f4d1a 375492b 98f4d1a fae4bf4 7e96635 98f4d1a 1d73717 98f4d1a 1d73717 98f4d1a 413c1cd 98f4d1a fae4bf4 a6ac1c5 46649eb 4e4f2f6 262b18a 413c1cd 46649eb |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 |
from diffusers import StableDiffusionXLInpaintPipeline
import gradio as gr
import numpy as np
import imageio
from PIL import Image
import torch
import modin.pandas as pd
device = "cuda" if torch.cuda.is_available() else "cpu"
pipe = StableDiffusionXLInpaintPipeline.from_pretrained("stabilityai/sdxl-turbo", safety_checker=None)
pipe = pipe.to(device)
def resize(value,img):
img = Image.open(img)
img = img.resize((value,value))
return img
def img2img(source_img, prompt, strength):
imageio.imwrite("data.png", source_img['image'])
imageio.imwrite("data_mask.png", source_img["mask"])
src = resize(768, "data.png")
src.save("src.png")
mask = resize(768, "data_mask.png")
mask.save("mask.png")
image = pipe(prompt=prompt, image=src, mask_image=mask, num_inference_steps=6, strength=strength, guidance_scale=0.0).images[0]
return image
title="SDXL Turbo Inpainting CPU"
description="Inpainting with SDXL Turbo <br><br> <b>Please use square .png image as input, 512x512, 768x768, or 1024x1024</b>"
gr.Interface(fn=img2img, inputs=[gr.Image(source="upload", tool="sketch", label="Source Image"),
gr.Textbox(label='What you want the AI to Generate, 77 Token limit'),
gr.Slider(minimum=.5, maximum=1, value=.75, step=.025, label='Strength')],
outputs='image',
title=title,
description=description,
article = "Code Monkey: <a href=\"https://huggingface.co/Manjushri\">Manjushri</a>").launch(max_threads=True, debug=True) |