app.py

import gradio as gr
from gradio_client import Client

def get_caption(image_in):
    # Correct indentation before client initialization
    client = Client("https://vikhyatk-moondream1.hf.space/")
    
    # Align the call to 'client.predict' with the line above
    # Also, the parameter 'prompt' is not defined in 'get_caption'. 
    # It should be 'image_in' to match the function signature.
    results = client.predict(
        image_in,  # Use the 'image_in' parameter as input to the model
        "Describe the image",  # Assuming this is a constant parameter for the model
        api_name="/answer_question"
    )
    # Print and return the results assuming it contains the description
    print(results)
    return results

def get_lcm(prompt):
    client = Client("https://latent-consistency-lcm-lora-for-sdxl.hf.space/")
    
    # Consistent indentation for 'client.predict' call
    results = client.predict(
        prompt,  # 'parameter_5' 텍스트박스 컴포넌트의 문자열
        0.3,     # 'Guidance' 슬라이더 컴포넌트의 0.0과 5 사이의 부동소수점 값
        8,       # 'Steps' 슬라이더 컴포넌트의 2와 10 사이의 부동소수점 값
        0,       # 'Seed' 슬라이더 컴포넌트의 0과 12013012031030 사이의 부동소수점 값
        True,    # 'Randomize' 체크박스 컴포넌트의 불리언 값
        api_name="/predict"
    )
    
    # Process results and possibly call predict multiple times if multiple images are needed
    print(results)
    return results  # 실제 API 응답에 기반한 반환 구문

def infer(image_in):
    caption = get_caption(image_in)
    img_var = get_lcm(caption)
    return img_var

# Create an Interface object with proper parameters
interface = gr.Interface(
    title="ArXivGPT Image",
    description="Image to Image variation, using LCM SDXL & Moondream1",
    fn=infer,
    inputs=gr.Image(type="filepath", label="Image input"),
    outputs=gr.Image(label="Image variation")
)

# Launch the interface
interface.queue(max_size=25).launch()