Spaces:

AnTrc2
/

Try-On

Runtime error

Try-On / app.py

Upload app.py

fa5f529 verified almost 2 years ago

1.46 kB

import torch
from diffusers import StableDiffusionPipeline
import gradio as gr
from PIL import Image

# Load the model and pipeline
model_id = "ares1123/virtual-dress-try-on"
pipeline = StableDiffusionPipeline.from_pretrained(model_id)
pipeline.to("cuda" if torch.cuda.is_available() else "cpu")

def virtual_try_on(image, clothing_image):
    # Convert images to proper format and get dimensions
    width, height = image.size
    
    # Ensure dimensions are multiples of 8
    width = (width // 8) * 8
    height = (height // 8) * 8

    # Resize images to fit the model's expected input
    image = image.resize((width, height))
    clothing_image = clothing_image.resize((width, height))

    # Define a prompt describing what you want the model to do
    prompt = "A person wearing new clothes"

    # Process the images using the model
    result = pipeline(prompt=prompt, image=image, conditioning_image=clothing_image)
    try_on_image = result.images[0]
    
    return try_on_image

# Set up a simple Gradio interface for testing
interface = gr.Interface(
    fn=virtual_try_on,
    inputs=[gr.Image(type="pil", label="User Image"), 
            gr.Image(type="pil", label="Clothing Image")],
    outputs=gr.Image(type="pil"),
    title="Virtual Dress Try-On",
    description="Upload an image of yourself and a clothing image to try it on virtually!"
)

# Launch the interface
interface.launch(share=True)

	import torch
	from diffusers import StableDiffusionPipeline
	import gradio as gr
	from PIL import Image

	# Load the model and pipeline
	model_id = "ares1123/virtual-dress-try-on"
	pipeline = StableDiffusionPipeline.from_pretrained(model_id)
	pipeline.to("cuda" if torch.cuda.is_available() else "cpu")

	def virtual_try_on(image, clothing_image):
	# Convert images to proper format and get dimensions
	width, height = image.size

	# Ensure dimensions are multiples of 8
	width = (width // 8) * 8
	height = (height // 8) * 8

	# Resize images to fit the model's expected input
	image = image.resize((width, height))
	clothing_image = clothing_image.resize((width, height))

	# Define a prompt describing what you want the model to do
	prompt = "A person wearing new clothes"

	# Process the images using the model
	result = pipeline(prompt=prompt, image=image, conditioning_image=clothing_image)
	try_on_image = result.images[0]

	return try_on_image

	# Set up a simple Gradio interface for testing
	interface = gr.Interface(
	fn=virtual_try_on,
	inputs=[gr.Image(type="pil", label="User Image"),
	gr.Image(type="pil", label="Clothing Image")],
	outputs=gr.Image(type="pil"),
	title="Virtual Dress Try-On",
	description="Upload an image of yourself and a clothing image to try it on virtually!"
	)

	# Launch the interface
	interface.launch(share=True)