app.py

import gradio as gr

import torch
import numpy as np
from PIL import Image
from diffusers import StableDiffusionPipeline, DPMSolverMultistepScheduler

MAX_IMAGES = 1


def generate_images(
    type1: str,
    type2: str,
    hp_num: int,
    attack_num: int,
    defense_num: int,
    sp_attack_num: int,
    sp_defense_num: int,
    speed_num: int,
) -> list:
    """Generates a sprite based on the input stats.

    Parameters
    ----------

    Returns
    -------
    list
        List of PIL images.
    """
    # Initalize the images list
    images_list = []
    # Calculate the base total
    base_total = (
        hp_num + attack_num + defense_num + sp_attack_num + sp_defense_num + speed_num
    )
    # Create the text prompt
    prompt = f"type1: {type1}, type2: {type2}, base_total: {base_total}, hp: {hp_num}, attack: {attack_num}, defense: {defense_num}, sp_attack: {sp_attack_num}, sp_defense: {sp_defense_num}, speed: {speed_num}"
    # Generate the images
    for _ in range(MAX_IMAGES):
        image = pipe(
            prompt,
            height=288,
            width=288,
            num_inference_steps=10,
            guidance_scale=7.5,
            cross_attention_kwargs={"scale": 1.0},
        ).images[0]
        images_list.append(Image.fromarray(np.array(image)))

    return images_list


# Create the demo interface
demo = gr.Blocks()

# Set the models to load
model_base = "stabilityai/stable-diffusion-2-base"
lora_model_path = "michaelriedl/MonsterForgeFusion-sd-2-base"

# Create the pipeline
pipe = StableDiffusionPipeline.from_pretrained(
    model_base, torch_dtype=torch.float32, use_safetensors=False, local_files_only=False
)
pipe.scheduler = DPMSolverMultistepScheduler.from_config(pipe.scheduler.config)
pipe.unet.load_attn_procs(lora_model_path)

# Create the interface
with demo:
    gr.HTML(
        """
        <div style="text-align: center; margin: 0 auto;">
            <p style="margin-bottom: 14px; line-height: 23px;">
                Gradio demo for MonsterForgeFusion models. This was built with LoRA fine-tuning of Stable Diffusion models.
            </p>
        </div>
        """
    )
    with gr.Column():
        with gr.Row():
            gallery = gr.Gallery(
                columns=MAX_IMAGES, preview=True, object_fit="scale-down"
            )
        with gr.Row():
            type1 = gr.Dropdown(
                [
                    "bug",
                    "dark",
                    "dragon",
                    "electric",
                    "fairy",
                    "fighting",
                    "fire",
                    "flying",
                    "ghost",
                    "grass",
                    "ground",
                    "ice",
                    "normal",
                    "poison",
                    "psychic",
                    "rock",
                    "steel",
                    "water",
                ],
                value="steel",
                label="Type 1",
            )
            type2 = gr.Dropdown(
                [
                    "bug",
                    "dark",
                    "dragon",
                    "electric",
                    "fairy",
                    "fighting",
                    "fire",
                    "flying",
                    "ghost",
                    "grass",
                    "ground",
                    "ice",
                    "normal",
                    "poison",
                    "psychic",
                    "rock",
                    "steel",
                    "water",
                ],
                value="fire",
                label="Type 2",
            )
        with gr.Row():
            hp_num = gr.Slider(
                minimum=1,
                maximum=100,
                value=50,
                step=1,
                label="HP",
            )
            attack_num = gr.Slider(
                minimum=1,
                maximum=100,
                value=50,
                step=1,
                label="Attack",
            )
        with gr.Row():
            defense_num = gr.Slider(
                minimum=1,
                maximum=100,
                value=50,
                step=1,
                label="Defense",
            )
            sp_attack_num = gr.Slider(
                minimum=1,
                maximum=100,
                value=50,
                step=1,
                label="Special Attack",
            )
        with gr.Row():
            sp_defense_num = gr.Slider(
                minimum=1,
                maximum=100,
                value=50,
                step=1,
                label="Special Defense",
            )
            speed_num = gr.Slider(
                minimum=1,
                maximum=100,
                value=50,
                step=1,
                label="Speed",
            )
        gen_btn = gr.Button("Generate")
        gen_btn.click(
            fn=generate_images,
            inputs=[
                type1,
                type2,
                hp_num,
                attack_num,
                defense_num,
                sp_attack_num,
                sp_defense_num,
                speed_num,
            ],
            outputs=gallery,
        )
    gr.HTML(
        """
        <div class="footer">
            <div style='text-align: center;'>MonsterForgeFusion by <a href='https://michaelriedl.com/' target='_blank'>Michael Riedl</a></div>
        </div>
        """
    )

# Launch the interface
demo.launch()