app.py

import subprocess
import os
import gradio as gr
import torch
from PIL import Image, ImageEnhance
from pygltflib import GLTF2
from pygltflib.utils import ImageFormat, Texture, Material, Image as GLTFImage
import spaces


if torch.cuda.is_available():
    device = "cuda"
    print("Using GPU")
else:
    device = "cpu"
    print("Using CPU")


subprocess.run(["git", "clone", "https://github.com/Nick088Official/Stable_Diffusion_Finetuned_Minecraft_Skin_Generator.git"])


@spaces.GPU()
def run_inference(prompt, stable_diffusion_model, num_inference_steps, guidance_scale, model_precision_type, seed, output_image_name, verbose):

    os.chdir("Stable_Diffusion_Finetuned_Minecraft_Skin_Generator")
    
    if stable_diffusion_model == '2':
        sd_model = "minecraft-skins"
    else:
        sd_model = "minecraft-skins-sdxl"

    inference_command = f"python Python_Scripts/{sd_model}.py '{prompt}' {num_inference_steps} {guidance_scale} {model_precision_type} {seed} {output_image_name} {'--verbose' if verbose else ''}"
    
    os.system(inference_command)
    
    os.chdir("..")
    
    to3d_model_command = f"sh 64x32to64x64skin3dmodel.sh Stable_Diffusion_Finetuned_Minecraft_Skin_Generator/output_minecraft_skins/{output_image_name}"
    
    os.system(to3d_model_command)

    
    filename = "3d_model_player.glb"
    gltf = GLTF2().load(filename)

    # Step 1: Find the index of the existing texture you want to replace
    # Let's assume the texture you want to replace is at index 1 (you need to replace 1 with the actual index)
    existing_texture_index = 0

    # Check if the existing_texture_index is valid
    if existing_texture_index < len(gltf.textures):
        # Step 2: Remove the old texture and its associated image from the GLB
        # Remove the texture
        gltf.textures.pop(existing_texture_index)

        # Remove the image associated with the texture
        existing_image_index = gltf.materials[0].pbrMetallicRoughness.baseColorTexture.index
        gltf.images.pop(existing_image_index)


        # Step 3: Add the new image and texture to the GLB
        # Create and add a new image to the glTF (same as before)
        new_image = GLTFImage()
        new_image.uri = os.path.join(f"Stable_Diffusion_Finetuned_Minecraft_Skin_Generator/output_minecraft_skins/{output_image_name}-converted.png")
        gltf.images.append(new_image)



        # Create a new texture and associate it with the added image
        new_texture = Texture()
        new_texture.source = len(gltf.images) - 1  # Index of the newly added image
        new_texture.sampler = 0
        # set to nearest neighbor

        gltf.textures.append(new_texture)

        # Step 4: Assign the new texture to the appropriate material(s) or mesh(es)
        # Assuming you have a mesh/primitive that was using the old texture and you want to apply the new texture to it, you need to set the material index for that mesh/primitive.
        # Replace 0 with the actual index of the mesh/primitive you want to update.
        gltf.materials[0].pbrMetallicRoughness.baseColorTexture.index = len(gltf.textures) - 1


        # Now you can convert the images to data URIs and save the updated GLB
        gltf.convert_images(ImageFormat.DATAURI)

        output_3d_model = "output_3d_model.glb"
        gltf.save(output_3d_model)
    else:
        print("Invalid existing_texture_index")


    # Return the generated image and the processed model
    return os.path.join(f"Stable_Diffusion_Finetuned_Minecraft_Skin_Generator/output_minecraft_skins/{output_image_name}"), output_3d_model

    
    
# Define Gradio UI components
prompt = gr.Textbox(label="Your Prompt", info="What the Minecraft Skin should look like")
stable_diffusion_model = gr.Dropdown(['2', 'xl'], value="xl", label="Stable Diffusion Model", info="Choose which Stable Diffusion Model to use, xl understands prompts better")
num_inference_steps = gr.Number(label="Number of Inference Steps", precision=0, value=25)
guidance_scale = gr.Number(minimum=0.1, value=7.5, label="Guidance Scale", info="The number of denoising steps of the image. More denoising steps usually lead to a higher quality image at the cost of slower inference")
model_precision_type = gr.Dropdown(["fp16", "fp32"], value="fp16", label="Model Precision Type", info="The precision type to load the model, like fp16 which is faster, or fp32 which gives better results")
seed = gr.Number(value=42, label="Seed", info="A starting point to initiate generation, put 0 for a random one")
output_image_name = gr.Textbox(label="Output Image Name", info="The name of the file of the output image skin, keep the .png", value="output-skin.png")
verbose = gr.Checkbox(label="Verbose Output", info="Produce more detailed output while running", value=False)


# Create the Gradio interface
gr.Interface(
    fn=run_inference,
    inputs=[
        prompt,
        stable_diffusion_model,
        num_inference_steps,
        guidance_scale,
        model_precision_type,
        seed,
        output_image_name,
        verbose
    ],
    outputs=[
        gr.Image(label="Generated Minecraft Skin Image Asset"),
        gr.Model3D(clear_color=[0.0, 0.0, 0.0, 0.0],  label="3D Model")
    ],
    title="Minecraft Skin Generator",
    description="Make AI generated Minecraft Skins by a Finetuned Stable Diffusion Version!<br>Model used: https://github.com/Nick088Official/Stable_Diffusion_Finetuned_Minecraft_Skin_Generator<br>Hugging Face Space made by [Nick088](https://linktr.ee/Nick088)",
).launch(show_api=False, share=True)