update for 5-2-1

Author: shunk031

notebooks/5-2-2_controlnet.ipynb

@@ -8,6 +8,16 @@
     "## 準備"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "116b9549-ae46-4794-85df-0c4109daaaf1",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!pip install -qq py-img-gen[controlnet]"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 1,
@@ -961,7 +971,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.10.12"
+   "version": "3.11.10"
   }
  },
  "nbformat": 4,

scripts/5-2-1_attend-and-excite.py

@@ -0,0 +1,143 @@
+# ---
+# jupyter:
+#   jupytext:
+#     text_representation:
+#       extension: .py
+#       format_name: percent
+#       format_version: '1.3'
+#       jupytext_version: 1.16.7
+#   kernelspec:
+#     display_name: Python 3 (ipykernel)
+#     language: python
+#     name: python3
+# ---
+
+# %% [markdown]
+# # Attend-and-Excite の実装
+
+# %% [markdown]
+# [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/py-img-gen/python-image-generation/blob/main/notebooks/5-2-1_attend-and-excite.ipynb)
+
+# %% [markdown]
+# ## 準備
+
+# %%
+# !pip install -qq py-img-gen
+
+# %%
+import warnings
+
+import torch
+
+device = torch.device(
+    "cuda" if torch.cuda.is_available() else "cpu"
+)
+dtype = torch.float16
+seed = 42
+
+warnings.simplefilter("ignore")
+
+# %% [markdown]
+# ## オリジナルの StableDiffusionPipeline の読み込み
+
+# %%
+from diffusers import StableDiffusionPipeline
+
+model_id = "stable-diffusion-v1-5/stable-diffusion-v1-5"
+
+pipe_sd = StableDiffusionPipeline.from_pretrained(
+    model_id, torch_dtype=dtype
+)
+pipe_sd = pipe_sd.to(device)
+
+# %% [markdown]
+# ## Attend and Excite を実装した StableDiffusionAttendAndExcitePipeline の読み込み
+
+# %%
+from diffusers import StableDiffusionAttendAndExcitePipeline
+
+pipe_ae = (
+    StableDiffusionAttendAndExcitePipeline.from_pretrained(
+        model_id, torch_dtype=dtype
+    )
+)
+pipe_ae = pipe_ae.to(device)
+
+# %% [markdown]
+# ## StableDiffusion での画像生成
+
+# %%
+from diffusers.utils import make_image_grid
+
+prompt = "A horse and a dog"
+
+images_sd = pipe_sd(
+    prompt,
+    num_images_per_prompt=2,
+    generator=torch.manual_seed(seed),
+).images
+
+# %%
+gen_result_sd = make_image_grid(
+    images=images_sd, rows=1, cols=2
+)
+gen_result_sd
+
+# %% [markdown]
+# ## Attend and Excite を適用した Stable Diffusion での画像生成
+
+# %%
+# `get_indices` 関数を使用して、対象のトークン（horse と dog）のインデックスを調べる
+# 2 と 5 がそれぞれ horse と dog であることを確認
+print(f"Indicies: {pipe_ae.get_indices(prompt)}")
+
+# %%
+# 上記で調べたトークンのインデックスを指定
+token_indices = [2, 5]
+
+# Attend-and-Excite パイプラインによって画像を生成
+images_ae = pipe_ae(
+    prompt,
+    num_images_per_prompt=2,
+    generator=torch.manual_seed(seed),
+    #
+    # Additional arguments for Attend-and-Excite
+    # 対象のトークンを指定
+    #
+    token_indices=token_indices,
+).images
+
+# %%
+gen_result_ae = make_image_grid(
+    images=images_ae, rows=1, cols=2
+)
+gen_result_ae
+
+# %% [markdown]
+# ## 生成結果の比較
+
+# %%
+import matplotlib.pyplot as plt
+from mpl_toolkits.axes_grid1 import ImageGrid
+
+fig = plt.figure(figsize=(20, 5))
+grid = ImageGrid(
+    fig,
+    rect=111,
+    nrows_ncols=(1, 2),
+    axes_pad=0.1,
+)
+fig.suptitle(f"Prompt: {prompt}")
+
+images = [
+    gen_result_sd,
+    gen_result_ae,
+]
+titles = [
+    r"Stable Diffusion ${\it without}$ Attend-and-Excite",
+    r"Stable Diffusion ${\it with}$ Attend-and-Excite",
+]
+for i, (image, title) in enumerate(zip(images, titles)):
+    grid[i].imshow(image)
+    grid[i].axis("off")
+    grid[i].set_title(title)