app.py

from PIL import Image
import torch
import re
import gradio as gr
from diffusers import AutoPipelineForText2Image
from diffusers import AutoPipelineForImage2Image
from diffusers.utils import load_image, make_image_grid
pipeline_text2image = AutoPipelineForText2Image.from_pretrained("stabilityai/sdxl-turbo", torch_dtype=torch.float16, variant="fp16")
pipeline_image2image = AutoPipelineForImage2Image.from_pipe(pipeline_text2image).to("cuda")
pipeline_text2image = pipeline_text2image.to("cuda")


def text2img(prompt = "A cinematic shot of a baby racoon wearing an intricate italian priest robe.",guidance_scale=0.0, num_inference_steps=1):
    image = pipeline_text2image(prompt=prompt, guidance_scale=guidance_scale, num_inference_steps=num_inference_steps).images[0]
    return image

def img2img(image,prompt="A cinematic shot of a baby racoon wearing an intricate italian priest robe.", guidance_scale=0.0, num_inference_steps=1,strength=0.5):
    init_image = load_image(image)
    init_image = init_image.resize((512, 512))
    image = pipeline_image2image(prompt, image=init_image, strength=strength, guidance_scale=guidance_scale, num_inference_steps=num_inference_steps).images[0]
    return image

gradio_app_text2img = gr.Interface(
    fn=text2img,
    inputs=[
        gr.Text(),
        gr.Slider(0.0, 10.0, value=1,step=0.1),
        gr.Slider(0.0, 100.0, value=1,step=1)
    ],
    outputs="image",
)

gradio_app_img2img = gr.Interface(
    fn=img2img,
    inputs=[
        gr.Image(type='filepath'),
        gr.Text(),
        gr.Slider(0.0, 10.0, value=1,step=0.1),
        gr.Slider(0.0, 100.0, value=1,step=1),
        gr.Slider(0.0, 1.0, value=0.5,step=0.05)
    ],
    outputs="image",
)

demo = gr.TabbedInterface([gradio_app_text2img,gradio_app_img2img], ["text2img","img2img"])

if __name__ == "__main__":
    demo.launch()