app.py

# import gradio as gr
# import torch
# from diffusers import DiffusionPipeline, AutoPipelineForText2Image
# import base64
# from io import BytesIO



# def load_amused_model():
#     # pipeline = DiffusionPipeline.from_pretrained("Bakanayatsu/ponyDiffusion-V6-XL-Turbo-DPO") 
#     # AutoPipelineForText2Image.from_pretrained("stabilityai/sd-turbo")
#     # AutoPipelineForText2Image.from_pretrained("stabilityai/sdxl-turbo")
#     return DiffusionPipeline.from_pretrained("CompVis/stable-diffusion-v1-4",
#         safety_checker = None,
#         requires_safety_checker = False)
     

# # Generate image from prompt using AmusedPipeline
# def generate_image(prompt):
#     try:
#         pipe = load_amused_model()
#         generator = torch.Generator().manual_seed(8)  # Create a generator for reproducibility
#         image = pipe(prompt, generator=generator).images[0]  # Generate image from prompt
#         # image = pipe(prompt=prompt, num_inference_steps=1, guidance_scale=0.0).images[0]
#         return image, None
#     except Exception as e:
#         return None, str(e)

# def inference(prompt):
#     print(f"Received prompt: {prompt}")  # Debugging statement
#     image, error = generate_image(prompt)
#     if error:
#         print(f"Error generating image: {error}")  # Debugging statement
#         return "Error: " + error
    
#     buffered = BytesIO()
#     image.save(buffered, format="PNG")
#     img_str = base64.b64encode(buffered.getvalue()).decode("utf-8")
#     return img_str

# gradio_interface = gr.Interface(
#     fn=inference,
#     inputs="text",
#     outputs="text"  # Change output to text to return base64 string
# )

# if __name__ == "__main__":
#     gradio_interface.launch()



import gradio as gr
from diffusers import DiffusionPipeline, DPMSolverSinglestepScheduler
import torch
import base64
from io import BytesIO


def load_amused_model():
    # pipeline = DiffusionPipeline.from_pretrained("Bakanayatsu/ponyDiffusion-V6-XL-Turbo-DPO") 
    # AutoPipelineForText2Image.from_pretrained("stabilityai/sd-turbo")
    # AutoPipelineForText2Image.from_pretrained("stabilityai/sdxl-turbo")
    return DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float32).to("cpu")


# Generate image from prompt using AmusedPipeline
def generate_image(prompt):
    try:
        pipe = load_amused_model()
        pipe.load_lora_weights(
            "mann-e/Mann-E_Turbo",
            weight_name="manne_turbo.safetensors",
        )
        # This is equivalent to DPM++ SDE Karras, as noted in https://huggingface.co/docs/diffusers/main/en/api/schedulers/overview
        pipe.scheduler = DPMSolverSinglestepScheduler.from_config(pipe.scheduler.config, use_karras_sigmas=True)

        #generator = torch.Generator().manual_seed(8)  # Create a generator for reproducibility
        #image = pipe(prompt, generator=generator).images[0]  # Generate image from prompt
        # image = pipe(prompt=prompt, num_inference_steps=1, guidance_scale=0.0).images[0]
        image = pipe(
            prompt="a cat in a bustling middle eastern city",
            num_inference_steps=8,
            guidance_scale=4,
            width=768,
            height=768,
            clip_skip=1
        ).images[0]
        return image, None
    except Exception as e:
        return None, str(e)


def inference(prompt):
    print(f"Received prompt: {prompt}")  # Debugging statement
    image, error = generate_image(prompt)
    if error:
        print(f"Error generating image: {error}")  # Debugging statement
        return "Error: " + error

    buffered = BytesIO()
    image.save(buffered, format="PNG")
    img_str = base64.b64encode(buffered.getvalue()).decode("utf-8")
    return img_str


gradio_interface = gr.Interface(
    fn=inference,
    inputs="text",
    outputs="text"  # Change output to text to return base64 string
)

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