app.py

import os
import sys

ROOT = os.path.dirname(os.path.abspath(__file__))
sys.path.insert(0, os.path.join(ROOT, "stylegan2"))

import torch
import numpy as np
import gradio as gr
from PIL import Image
from stylegan2 import legacy

# ---------------- CONFIG ----------------
MODEL_PATH = "network-snapshot-000700.pkl"
DEVICE = "cpu"

# ---------------- LOAD MODEL ----------------
with open(MODEL_PATH, "rb") as f:
    G = legacy.load_network_pkl(f)["G_ema"].to(DEVICE)
G.eval()

# ---------------- GENERATION ----------------
@torch.no_grad()
def generate_image(seed, truncation):
    seed = int(seed)
    truncation = float(truncation)

    print(f"Generating image for seed {seed} | trunc={truncation}")

    z = torch.from_numpy(
        np.random.RandomState(seed).randn(1, G.z_dim)
    ).to(DEVICE)

    img = G(
        z,
        None,
        truncation_psi=truncation,
        noise_mode="const"
    )

    img = (img.permute(0, 2, 3, 1) * 127.5 + 128)
    img = img.clamp(0, 255).to(torch.uint8)

    return Image.fromarray(img[0].cpu().numpy(), "RGB")

# ---------------- GRADIO UI ----------------
with gr.Blocks() as demo:
    gr.Markdown("## STYLEGAN2 Anime Image Generator")

    seed = gr.Slider(
        minimum=0,
        maximum=10000,
        value=0,
        step=1,
        label="Seed"
    )

    truncation = gr.Slider(
        minimum=0.3,
        maximum=1.0,
        value=0.7,
        step=0.05,
        label="Truncation (ψ)"
    )

    generate_btn = gr.Button("Generate Image")
    output = gr.Image(type="pil", label="Generated Image")

    generate_btn.click(
        fn=generate_image,
        inputs=[seed, truncation],
        outputs=output
    )

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