Update

.pre-commit-config.yaml

@@ -0,0 +1,46 @@
+exclude: ^stylegan3
+repos:
+- repo: https://github.com/pre-commit/pre-commit-hooks
+  rev: v4.2.0
+  hooks:
+  - id: check-executables-have-shebangs
+  - id: check-json
+  - id: check-merge-conflict
+  - id: check-shebang-scripts-are-executable
+  - id: check-toml
+  - id: check-yaml
+  - id: double-quote-string-fixer
+  - id: end-of-file-fixer
+  - id: mixed-line-ending
+    args: ['--fix=lf']
+  - id: requirements-txt-fixer
+  - id: trailing-whitespace
+- repo: https://github.com/myint/docformatter
+  rev: v1.4
+  hooks:
+  - id: docformatter
+    args: ['--in-place']
+- repo: https://github.com/pycqa/isort
+  rev: 5.10.1
+  hooks:
+    - id: isort
+- repo: https://github.com/pre-commit/mirrors-mypy
+  rev: v0.812
+  hooks:
+    - id: mypy
+      args: ['--ignore-missing-imports']
+- repo: https://github.com/google/yapf
+  rev: v0.32.0
+  hooks:
+  - id: yapf
+    args: ['--parallel', '--in-place']
+- repo: https://github.com/kynan/nbstripout
+  rev: 0.5.0
+  hooks:
+    - id: nbstripout
+      args: ['--extra-keys', 'metadata.interpreter metadata.kernelspec cell.metadata.pycharm']
+- repo: https://github.com/nbQA-dev/nbQA
+  rev: 1.3.1
+  hooks:
+    - id: nbqa-isort
+    - id: nbqa-yapf

.style.yapf

@@ -0,0 +1,5 @@
+[style]
+based_on_style = pep8
+blank_line_before_nested_class_or_def = false
+spaces_before_comment = 2
+split_before_logical_operator = true

app.py

@@ -3,172 +3,127 @@
 from __future__ import annotations
 
 import argparse
-import functools
-import os
-import pickle
-import sys
 
 import gradio as gr
 import numpy as np
-import torch
-import torch.nn as nn
-from huggingface_hub import hf_hub_download
 
-sys.path.insert(0, 'stylegan3')
+from model import Model
 
-TITLE = 'NVlabs/stylegan3'
-DESCRIPTION = '''This is an unofficial demo for https://github.com/NVlabs/stylegan3.
+TITLE = '# NVlabs/stylegan3'
+DESCRIPTION = '''This is an unofficial demo for [https://github.com/NVlabs/stylegan3](https://github.com/NVlabs/stylegan3).
 
 Expected execution time on Hugging Face Spaces: 50s
 '''
-SAMPLE_IMAGE_DIR = 'https://huggingface.co/spaces/hysts/StyleGAN3/resolve/main/samples'
-ARTICLE = f'''## Generated images
-- truncation: 0.7
-### AFHQv2
-- size: 512x512
-- seed: 0-99
-![AFHQv2 samples]({SAMPLE_IMAGE_DIR}/afhqv2.jpg)
-### FFHQ
-- size: 1024x1024
-- seed: 0-99
-![FFHQ samples]({SAMPLE_IMAGE_DIR}/ffhq.jpg)
-### FFHQ-U
-- size: 1024x1024
-- seed: 0-99
-![FFHQ-U samples]({SAMPLE_IMAGE_DIR}/ffhq-u.jpg)
-### MetFaces
-- size: 1024x1024
-- seed: 0-99
-![MetFaces samples]({SAMPLE_IMAGE_DIR}/metfaces.jpg)
-### MetFaces-U
-- size: 1024x1024
-- seed: 0-99
-![MetFaces-U samples]({SAMPLE_IMAGE_DIR}/metfaces-u.jpg)
-
-<center><img src="https://visitor-badge.glitch.me/badge?page_id=hysts.stylegan3" alt="visitor badge"/></center>
-'''
-
-TOKEN = os.environ['TOKEN']
+FOOTER = '<img id="visitor-badge" alt="visitor badge" src="https://visitor-badge.glitch.me/badge?page_id=hysts.stylegan3" />'
 
 
 def parse_args() -> argparse.Namespace:
     parser = argparse.ArgumentParser()
     parser.add_argument('--device', type=str, default='cpu')
     parser.add_argument('--theme', type=str)
-    parser.add_argument('--live', action='store_true')
     parser.add_argument('--share', action='store_true')
     parser.add_argument('--port', type=int)
     parser.add_argument('--disable-queue',
                         dest='enable_queue',
                         action='store_false')
-    parser.add_argument('--allow-flagging', type=str, default='never')
     return parser.parse_args()
 
 
-def make_transform(translate: tuple[float, float], angle: float) -> np.ndarray:
-    mat = np.eye(3)
-    sin = np.sin(angle / 360 * np.pi * 2)
-    cos = np.cos(angle / 360 * np.pi * 2)
-    mat[0][0] = cos
-    mat[0][1] = sin
-    mat[0][2] = translate[0]
-    mat[1][0] = -sin
-    mat[1][1] = cos
-    mat[1][2] = translate[1]
-    return mat
-
-
-def generate_z(z_dim: int, seed: int, device: torch.device) -> torch.Tensor:
-    return torch.from_numpy(np.random.RandomState(seed).randn(
-        1, z_dim)).to(device).float()
-
-
-@torch.inference_mode()
-def generate_image(model_name: str, seed: int, truncation_psi: float,
-                   tx: float, ty: float, angle: float,
-                   model_dict: dict[str, nn.Module],
-                   device: torch.device) -> np.ndarray:
-    model = model_dict[model_name]
-    seed = int(np.clip(seed, 0, np.iinfo(np.uint32).max))
-
-    z = generate_z(model.z_dim, seed, device)
-    label = torch.zeros([1, model.c_dim], device=device)
-
-    mat = make_transform((tx, ty), angle)
-    mat = np.linalg.inv(mat)
-    model.synthesis.input.transform.copy_(torch.from_numpy(mat))
-
-    out = model(z, label, truncation_psi=truncation_psi)
-    out = (out.permute(0, 2, 3, 1) * 127.5 + 128).clamp(0, 255).to(torch.uint8)
-    return out[0].cpu().numpy()
+def get_sample_image_url(name: str) -> str:
+    sample_image_dir = 'https://huggingface.co/spaces/hysts/StyleGAN3/resolve/main/samples'
+    return f'{sample_image_dir}/{name}.jpg'
 
 
-def load_model(file_name: str, device: torch.device) -> nn.Module:
-    path = hf_hub_download('hysts/StyleGAN3',
-                           f'models/{file_name}',
-                           use_auth_token=TOKEN)
-    with open(path, 'rb') as f:
-        model = pickle.load(f)['G_ema']
-    model.eval()
-    model.to(device)
-    with torch.inference_mode():
-        z = torch.zeros((1, model.z_dim)).to(device)
-        label = torch.zeros([1, model.c_dim], device=device)
-        model(z, label)
-    return model
+def get_sample_image_markdown(name: str) -> str:
+    url = get_sample_image_url(name)
+    size = 512 if name == 'afhqv2' else 1024
+    seed = '0-99'
+    return f'''
+    - size: {size}x{size}
+    - seed: {seed}
+    - truncation: 0.7
+    ![sample images]({url})'''
 
 
 def main():
     args = parse_args()
-    device = torch.device(args.device)
-
-    model_names = {
-        'AFHQv2-512-R': 'stylegan3-r-afhqv2-512x512.pkl',
-        'FFHQ-1024-R': 'stylegan3-r-ffhq-1024x1024.pkl',
-        'FFHQ-U-256-R': 'stylegan3-r-ffhqu-256x256.pkl',
-        'FFHQ-U-1024-R': 'stylegan3-r-ffhqu-1024x1024.pkl',
-        'MetFaces-1024-R': 'stylegan3-r-metfaces-1024x1024.pkl',
-        'MetFaces-U-1024-R': 'stylegan3-r-metfacesu-1024x1024.pkl',
-        'AFHQv2-512-T': 'stylegan3-t-afhqv2-512x512.pkl',
-        'FFHQ-1024-T': 'stylegan3-t-ffhq-1024x1024.pkl',
-        'FFHQ-U-256-T': 'stylegan3-t-ffhqu-256x256.pkl',
-        'FFHQ-U-1024-T': 'stylegan3-t-ffhqu-1024x1024.pkl',
-        'MetFaces-1024-T': 'stylegan3-t-metfaces-1024x1024.pkl',
-        'MetFaces-U-1024-T': 'stylegan3-t-metfacesu-1024x1024.pkl',
-    }
-
-    model_dict = {
-        name: load_model(file_name, device)
-        for name, file_name in model_names.items()
-    }
-
-    func = functools.partial(generate_image,
-                             model_dict=model_dict,
-                             device=device)
-    func = functools.update_wrapper(func, generate_image)
-
-    gr.Interface(
-        func,
-        [
-            gr.inputs.Radio(list(model_names.keys()),
-                            type='value',
-                            default='FFHQ-1024-R',
-                            label='Model'),
-            gr.inputs.Number(default=0, label='Seed'),
-            gr.inputs.Slider(
-                0, 2, step=0.05, default=0.7, label='Truncation psi'),
-            gr.inputs.Slider(-1, 1, step=0.05, default=0, label='Translate X'),
-            gr.inputs.Slider(-1, 1, step=0.05, default=0, label='Translate Y'),
-            gr.inputs.Slider(-180, 180, step=5, default=0, label='Angle'),
-        ],
-        gr.outputs.Image(type='numpy', label='Output'),
-        title=TITLE,
-        description=DESCRIPTION,
-        article=ARTICLE,
-        theme=args.theme,
-        allow_flagging=args.allow_flagging,
-        live=args.live,
-    ).launch(
+    model = Model(args.device)
+
+    with gr.Blocks(theme=args.theme, css='style.css') as demo:
+        gr.Markdown(TITLE)
+        gr.Markdown(DESCRIPTION)
+
+        with gr.Tabs():
+            with gr.TabItem('App'):
+                with gr.Row():
+                    with gr.Column():
+                        with gr.Group():
+                            model_name = gr.Dropdown(list(
+                                model.MODEL_NAME_DICT.keys()),
+                                                     value='FFHQ-1024-R',
+                                                     label='Model')
+                            seed = gr.Slider(0,
+                                             np.iinfo(np.uint32).max,
+                                             step=1,
+                                             value=0,
+                                             label='Seed')
+                            psi = gr.Slider(0,
+                                            2,
+                                            step=0.05,
+                                            value=0.7,
+                                            label='Truncation psi')
+                            tx = gr.Slider(-1,
+                                           1,
+                                           step=0.05,
+                                           value=0,
+                                           label='Translate X')
+                            ty = gr.Slider(-1,
+                                           1,
+                                           step=0.05,
+                                           value=0,
+                                           label='Translate Y')
+                            angle = gr.Slider(-180,
+                                              180,
+                                              step=5,
+                                              value=0,
+                                              label='Angle')
+                            run_button = gr.Button('Run')
+                    with gr.Column():
+                        result = gr.Image(label='Result', elem_id='result')
+
+            with gr.TabItem('Sample Images'):
+                with gr.Row():
+                    model_name2 = gr.Dropdown([
+                        'afhqv2',
+                        'ffhq',
+                        'ffhq-u',
+                        'metfaces',
+                        'metfaces-u',
+                    ],
+                                              value='afhqv2',
+                                              label='Model')
+                with gr.Row():
+                    text = get_sample_image_markdown(model_name2.value)
+                    sample_images = gr.Markdown(text)
+
+        gr.Markdown(FOOTER)
+
+        model_name.change(fn=model.set_model, inputs=model_name, outputs=None)
+        run_button.click(fn=model.set_model_and_generate_image,
+                         inputs=[
+                             model_name,
+                             seed,
+                             psi,
+                             tx,
+                             ty,
+                             angle,
+                         ],
+                         outputs=result)
+        model_name2.change(fn=get_sample_image_markdown,
+                           inputs=model_name2,
+                           outputs=sample_images)
+
+    demo.launch(
         enable_queue=args.enable_queue,
         server_port=args.port,
         share=args.share,

model.py

@@ -0,0 +1,112 @@
+from __future__ import annotations
+
+import os
+import pathlib
+import pickle
+import sys
+
+import numpy as np
+import torch
+import torch.nn as nn
+from huggingface_hub import hf_hub_download
+
+current_dir = pathlib.Path(__file__).parent
+submodule_dir = current_dir / 'stylegan3'
+sys.path.insert(0, submodule_dir.as_posix())
+
+HF_TOKEN = os.environ['HF_TOKEN']
+
+
+class Model:
+    MODEL_NAME_DICT = {
+        'AFHQv2-512-R': 'stylegan3-r-afhqv2-512x512.pkl',
+        'FFHQ-1024-R': 'stylegan3-r-ffhq-1024x1024.pkl',
+        'FFHQ-U-256-R': 'stylegan3-r-ffhqu-256x256.pkl',
+        'FFHQ-U-1024-R': 'stylegan3-r-ffhqu-1024x1024.pkl',
+        'MetFaces-1024-R': 'stylegan3-r-metfaces-1024x1024.pkl',
+        'MetFaces-U-1024-R': 'stylegan3-r-metfacesu-1024x1024.pkl',
+        'AFHQv2-512-T': 'stylegan3-t-afhqv2-512x512.pkl',
+        'FFHQ-1024-T': 'stylegan3-t-ffhq-1024x1024.pkl',
+        'FFHQ-U-256-T': 'stylegan3-t-ffhqu-256x256.pkl',
+        'FFHQ-U-1024-T': 'stylegan3-t-ffhqu-1024x1024.pkl',
+        'MetFaces-1024-T': 'stylegan3-t-metfaces-1024x1024.pkl',
+        'MetFaces-U-1024-T': 'stylegan3-t-metfacesu-1024x1024.pkl',
+    }
+
+    def __init__(self, device: str | torch.device):
+        self.device = torch.device(device)
+        self._download_all_models()
+        self.model_name = 'FFHQ-1024-R'
+        self.model = self._load_model(self.model_name)
+
+    def _load_model(self, model_name: str) -> nn.Module:
+        file_name = self.MODEL_NAME_DICT[model_name]
+        path = hf_hub_download('hysts/StyleGAN3',
+                               f'models/{file_name}',
+                               use_auth_token=HF_TOKEN)
+        with open(path, 'rb') as f:
+            model = pickle.load(f)['G_ema']
+        model.eval()
+        model.to(self.device)
+        return model
+
+    def set_model(self, model_name: str) -> None:
+        if model_name == self.model_name:
+            return
+        self.model_name = model_name
+        self.model = self._load_model(model_name)
+
+    def _download_all_models(self):
+        for name in self.MODEL_NAME_DICT.keys():
+            self._load_model(name)
+
+    @staticmethod
+    def make_transform(translate: tuple[float, float],
+                       angle: float) -> np.ndarray:
+        mat = np.eye(3)
+        sin = np.sin(angle / 360 * np.pi * 2)
+        cos = np.cos(angle / 360 * np.pi * 2)
+        mat[0][0] = cos
+        mat[0][1] = sin
+        mat[0][2] = translate[0]
+        mat[1][0] = -sin
+        mat[1][1] = cos
+        mat[1][2] = translate[1]
+        return mat
+
+    def generate_z(self, seed: int) -> torch.Tensor:
+        seed = int(np.clip(seed, 0, np.iinfo(np.uint32).max))
+        z = np.random.RandomState(seed).randn(1, self.model.z_dim)
+        return torch.from_numpy(z).float().to(self.device)
+
+    def postprocess(self, tensor: torch.Tensor) -> np.ndarray:
+        tensor = (tensor.permute(0, 2, 3, 1) * 127.5 + 128).clamp(0, 255).to(
+            torch.uint8)
+        return tensor.cpu().numpy()
+
+    def set_transform(self, tx: float, ty: float, angle: float) -> None:
+        mat = self.make_transform((tx, ty), angle)
+        mat = np.linalg.inv(mat)
+        self.model.synthesis.input.transform.copy_(torch.from_numpy(mat))
+
+    @torch.inference_mode()
+    def generate(self, z: torch.Tensor, label: torch.Tensor,
+                 truncation_psi: float) -> torch.Tensor:
+        return self.model(z, label, truncation_psi=truncation_psi)
+
+    def generate_image(self, seed: int, truncation_psi: float, tx: float,
+                       ty: float, angle: float) -> np.ndarray:
+        self.set_transform(tx, ty, angle)
+
+        z = self.generate_z(seed)
+        label = torch.zeros([1, self.model.c_dim], device=self.device)
+
+        out = self.generate(z, label, truncation_psi)
+        out = self.postprocess(out)
+        return out[0]
+
+    def set_model_and_generate_image(self, model_name: str, seed: int,
+                                     truncation_psi: float, tx: float,
+                                     ty: float, angle: float) -> np.ndarray:
+        self.set_model(model_name)
+        return self.generate_image(seed, truncation_psi, tx, ty, angle)

style.css

@@ -0,0 +1,11 @@
+h1 {
+  text-align: center;
+}
+div#result {
+  max-width: 600px;
+  max-height: 600px;
+}
+img#visitor-badge {
+  display: block;
+  margin: auto;
+}