xywwww's picture
Update app.py
3dc891c verified
raw
history blame
3.21 kB
import torch
import random
import numpy as np
import gradio as gr
from pytorch_lightning import seed_everything
from annotator.util import resize_image, HWC3
from diffusers import StableDiffusionControlNetPipeline, ControlNetModel
# # Load the controlnet model
# controlnet = ControlNetModel.from_pretrained("CompVis/controlnet")
# # Load the pipeline
# pipe = StableDiffusionControlNetPipeline.from_pretrained(
# "CompVis/stable-diffusion-v1-4",
# controlnet=controlnet
# ).to("cuda")
controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-canny", torch_dtype=torch.float16)
pipe = StableDiffusionControlNetPipeline.from_pretrained(
"runwayml/stable-diffusion-v1-5", controlnet=controlnet, torch_dtype=torch.float16
)
def process(input_image, prompt, a_prompt, n_prompt, num_samples, image_resolution, ddim_steps, guess_mode, strength, scale, seed, eta, low_threshold, high_threshold):
with torch.no_grad():
img = resize_image(HWC3(input_image), image_resolution)
if seed == -1:
seed = random.randint(0, 65535)
seed_everything(seed)
# Generate images using the pipeline
generator = torch.Generator("cuda").manual_seed(seed)
images = pipe(prompt=prompt + ', ' + a_prompt, num_inference_steps=ddim_steps, guidance_scale=scale, generator=generator, num_images_per_prompt=num_samples).images
results = [np.array(image) for image in images]
return results
block = gr.Blocks().queue()
with block:
with gr.Row():
gr.Markdown("## Scene Diffusion with ControlNet")
with gr.Row():
with gr.Column():
input_image = gr.Image(label="Image")
prompt = gr.Textbox(label="Prompt")
a_prompt = gr.Textbox(label="Additional Prompt")
n_prompt = gr.Textbox(label="Negative Prompt")
num_samples = gr.Slider(label="Images", minimum=1, maximum=12, value=1, step=1)
image_resolution = gr.Slider(label="Image Resolution", minimum=256, maximum=768, value=512, step=64)
ddim_steps = gr.Slider(label="Steps", minimum=1, maximum=100, value=20, step=1)
guess_mode = gr.Checkbox(label='Guess Mode', value=False)
strength = gr.Slider(label="Strength", minimum=0.0, maximum=1.0, value=0.5, step=0.1)
scale = gr.Slider(label="Scale", minimum=0.1, maximum=30.0, value=10.0, step=0.1)
seed = gr.Slider(label="Seed", minimum=0, maximum=10000, value=42, step=1)
eta = gr.Slider(label="ETA", minimum=0.0, maximum=1.0, value=0.0, step=0.1)
low_threshold = gr.Slider(label="Canny Low Threshold", minimum=1, maximum=255, value=100, step=1)
high_threshold = gr.Slider(label="Canny High Threshold", minimum=1, maximum=255, value=200, step=1)
submit = gr.Button("Generate")
with gr.Column():
output_image = gr.Gallery(label='Output', show_label=False, elem_id="gallery")
submit.click(fn=process, inputs=[input_image, prompt, a_prompt, n_prompt, num_samples, image_resolution, ddim_steps, guess_mode, strength, scale, seed, eta, low_threshold, high_threshold], outputs=output_image)
demo = block
demo.launch()