from huggingface_hub import from_pretrained_keras
from keras_cv import models
import gradio as gr
from tensorflow import keras

keras.mixed_precision.set_global_policy("mixed_float16")

sd_dreambooth_model = models.StableDiffusion(
    img_width=512, img_height=512, jit_compile=True
)
db_diffusion_model = from_pretrained_keras("keras-dreambooth/dreambooth_dosa_v2")
sd_dreambooth_model._diffusion_model = db_diffusion_model

# generate images
def generate_images(prompt, negative_prompt, num_imgs_to_gen, num_steps, guidance_scale):
    generated_images = sd_dreambooth_model.text_to_image(
        prompt,
        negative_prompt=negative_prompt,
        batch_size=num_imgs_to_gen,
        num_steps=num_steps,
        unconditional_guidance_scale=guidance_scale
    )
    return generated_images

with gr.Blocks() as demo:
    gr.HTML("<h2 style=\"font-size: 2em; font-weight: bold\" align=\"center\">Keras Dreambooth - The Humble Dosa</h2>")
    gr.HTML("<p style=\"font-size: 14; font-weight: normal\" align=\"left\">This model has been fine-tuned to learn the concept of a dosa.<br>To use this demo, insert the string <q>bhr dosa</q> in your prompt</p>")    
    with gr.Row():
        with gr.Column():
            prompt = gr.Textbox(label="Prompt", lines=1, value="bhr dosa")
            negative_prompt = gr.Textbox(label="Negative Prompt", lines=1, value="deformed")
            samples = gr.Slider(label="Number of Images", minimum=1, maximum=4, value=1, step=1)
            num_steps = gr.Slider(label="Inference Steps", minimum=25, maximum=100, value=50, step=1)
            guidance_scale = gr.Slider(label="Guidance Scale", minimum=1, maximum=12, value=7.5, step=0.5)
            run = gr.Button(value="Run")
        with gr.Column():
            gallery = gr.Gallery(label="Outputs").style(grid=(2,2))
    
    run.click(fn=generate_images, inputs=[prompt, negative_prompt, samples, num_steps, guidance_scale], outputs=gallery)

    gr.Examples([["realistic picture of a bhr dosa in a restaurant", "sambar", 1, 50, 7.5]],
                [prompt, negative_prompt, samples, num_steps, guidance_scale], gallery, generate_images)
    gr.Markdown('Demo created by [Bharat Raghunathan](https://huggingface.co/bharat-raghunathan/)')
# pass function, input type for prompt, the output for multiple images
demo.queue().launch()