File size: 1,955 Bytes
54ad393 f4ff201 54ad393 3fec1fb f4ff201 1291143 54ad393 1291143 54ad393 f4ff201 3fec1fb 70e3d12 3fec1fb 70e3d12 3fec1fb 70e3d12 548031b 1291143 548031b 3fec1fb 0a14984 3fec1fb 5305898 70e3d12 54ad393 548031b 70e3d12 548031b 3fec1fb 548031b b00a80c |
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 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 |
from diffusers import StableDiffusionXLPipeline
import torch
#from controlnet_aux import OpenposeDetector
#from diffusers.utils import load_image
import gradio as gr
model_base = "stabilityai/stable-diffusion-xl-base-1.0"
pipe = StableDiffusionXLPipeline.from_pretrained(
model_base, torch_dtype=torch.float16
)
pipe = pipe.to("cuda")
css = """
.btn-green {
background-image: linear-gradient(to bottom right, #6dd178, #00a613) !important;
border-color: #22c55e !important;
color: #166534 !important;
}
.btn-green:hover {
background-image: linear-gradient(to bottom right, #6dd178, #6dd178) !important;
}
"""
def generate(prompt, neg_prompt, samp_steps, guide_scale, lora_scale, progress=gr.Progress(track_tqdm=True)):
images = pipe(
prompt,
negative_prompt=neg_prompt,
num_inference_steps=samp_steps,
guidance_scale=guide_scale,
#cross_attention_kwargs={"scale": lora_scale},
num_images_per_prompt=4,
#generator=torch.manual_seed(97),
).images
return [(img, f"Image {i+1}") for i, img in enumerate(images)]
with gr.Blocks(css=css) as demo:
with gr.Column():
prompt = gr.Textbox(label="Prompt")
negative_prompt = gr.Textbox(label="Negative Prompt", value="lowres, bad anatomy, bad hands, cropped, worst quality, disfigured, deformed, extra limbs, asian, filter, render")
submit_btn = gr.Button("Generate", elem_classes="btn-green")
gallery = gr.Gallery(label="Generated images", height=1100)
with gr.Row():
samp_steps = gr.Slider(1, 100, value=25, step=1, label="Sampling steps")
guide_scale = gr.Slider(1, 10, value=6, step=0.5, label="Guidance scale")
lora_scale = gr.Slider(0, 1, value=0.5, step=0.01, label="LoRA power")
submit_btn.click(generate, [prompt, negative_prompt, samp_steps, guide_scale, lora_scale], [gallery], queue=True)
demo.queue(1)
demo.launch(debug=True) |