Norod78 commited on
Commit
63060b8
1 Parent(s): d1aa0c4

Upload ip_adapter_face_full_sdxl.ipynb

Browse files
Files changed (1) hide show
  1. ip_adapter_face_full_sdxl.ipynb +246 -0
ip_adapter_face_full_sdxl.ipynb ADDED
@@ -0,0 +1,246 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "cells": [
3
+ {
4
+ "cell_type": "code",
5
+ "execution_count": null,
6
+ "metadata": {
7
+ "id": "5rbnEz_FTJKL"
8
+ },
9
+ "outputs": [],
10
+ "source": [
11
+ "\n",
12
+ "######################################################################\n",
13
+ "!pip install -q --upgrade diffusers\n",
14
+ "!pip install -q --upgrade transformers\n",
15
+ "!pip install -q --upgrade tokenizers\n",
16
+ "######################################################################\n",
17
+ "!pip install -q --upgrade peft\n",
18
+ "######################################################################\n",
19
+ "#!pip install diffusers==0.21.1 transformers==4.33.2 tokenizers==0.13.3\n",
20
+ "######################################################################\n",
21
+ "\n",
22
+ "!pip install -q accelerate\n",
23
+ "!pip install -q safetensors\n",
24
+ "!pip install -q einops\n",
25
+ "!pip install -q onnxruntime-gpu\n",
26
+ "!pip install -q dlib\n",
27
+ "!pip install -q opencv-python\n",
28
+ "!pip install -q git+https://github.com/tencent-ailab/IP-Adapter.git"
29
+ ]
30
+ },
31
+ {
32
+ "cell_type": "code",
33
+ "execution_count": null,
34
+ "metadata": {
35
+ "id": "hThiB6kxTJKM"
36
+ },
37
+ "outputs": [],
38
+ "source": [
39
+ "import torch\n",
40
+ "from diffusers.utils import load_image\n",
41
+ "from PIL import Image\n",
42
+ "import os\n"
43
+ ]
44
+ },
45
+ {
46
+ "cell_type": "code",
47
+ "execution_count": null,
48
+ "metadata": {
49
+ "id": "RB9wDFckTJKN"
50
+ },
51
+ "outputs": [],
52
+ "source": [
53
+ "\n",
54
+ "os.makedirs(\"images\", exist_ok=True)\n",
55
+ "!wget --continue https://upload.wikimedia.org/wikipedia/commons/thumb/0/0f/1665_Girl_with_a_Pearl_Earring.jpg/800px-1665_Girl_with_a_Pearl_Earring.jpg -O images/Girl_with_a_Pearl_Earring.jpg\n",
56
+ "\n",
57
+ "!wget --continue https://huggingface.co/datasets/Norod78/ip-adapter-face-full-test/resolve/main/shape_predictor_5_face_landmarks.dat?download=true -O shape_predictor_5_face_landmarks.dat\n",
58
+ "\n",
59
+ "!wget --continue https://huggingface.co/datasets/Norod78/ip-adapter-face-full-test/raw/main/crop_head_dlib5.py -O crop_head_dlib5.py"
60
+ ]
61
+ },
62
+ {
63
+ "cell_type": "code",
64
+ "execution_count": null,
65
+ "metadata": {
66
+ "id": "kfvMkQD6TJKP"
67
+ },
68
+ "outputs": [],
69
+ "source": [
70
+ "import torch\n",
71
+ "from diffusers import StableDiffusionXLPipeline, DDIMScheduler, AutoencoderKL\n",
72
+ "from huggingface_hub import hf_hub_download\n",
73
+ "from diffusers.utils import load_image\n",
74
+ "from ip_adapter import IPAdapterFull\n",
75
+ "from PIL import Image\n",
76
+ "\n",
77
+ "def image_grid(imgs, rows, cols):\n",
78
+ " assert len(imgs) == rows*cols\n",
79
+ "\n",
80
+ " w, h = imgs[0].size\n",
81
+ " grid = Image.new('RGB', size=(cols*w, rows*h))\n",
82
+ " grid_w, grid_h = grid.size\n",
83
+ "\n",
84
+ " for i, img in enumerate(imgs):\n",
85
+ " grid.paste(img, box=(i%cols*w, i//cols*h))\n",
86
+ " return grid\n",
87
+ "\n",
88
+ "noise_scheduler = DDIMScheduler(\n",
89
+ " num_train_timesteps=1000,\n",
90
+ " beta_start=0.00085,\n",
91
+ " beta_end=0.012,\n",
92
+ " beta_schedule=\"scaled_linear\",\n",
93
+ " clip_sample=False,\n",
94
+ " set_alpha_to_one=False,\n",
95
+ " steps_offset=1\n",
96
+ ")\n",
97
+ "\n",
98
+ "base_model_path = \"stabilityai/stable-diffusion-xl-base-1.0\"\n",
99
+ "vae_model_path = \"madebyollin/sdxl-vae-fp16-fix\"\n",
100
+ "ip_ckpt = hf_hub_download(repo_id=\"h94/IP-Adapter\", subfolder=\"sdxl_models\", filename=\"ip-adapter-plus-face_sdxl_vit-h.safetensors\", repo_type=\"model\")\n",
101
+ "image_encoder_path = \"laion/CLIP-ViT-H-14-laion2B-s32B-b79K\"\n",
102
+ "device = \"cuda\"\n",
103
+ "\n",
104
+ "with torch.no_grad():\n",
105
+ " print(\"Loading vae\")\n",
106
+ " vae = AutoencoderKL.from_pretrained(vae_model_path).to(dtype=torch.float16)\n",
107
+ " print(\"Loading pipeline\")\n",
108
+ " pipe = StableDiffusionXLPipeline.from_pretrained(\n",
109
+ " base_model_path,\n",
110
+ " variant=\"fp16\",\n",
111
+ " torch_dtype=torch.float16,\n",
112
+ " scheduler=noise_scheduler,\n",
113
+ " vae=vae,\n",
114
+ " feature_extractor=None,\n",
115
+ " safety_checker=None\n",
116
+ " )\n",
117
+ "print(\"Done\")\n",
118
+ "\n"
119
+ ]
120
+ },
121
+ {
122
+ "cell_type": "code",
123
+ "source": [
124
+ "\n",
125
+ "#lora_weights = hf_hub_download(repo_id=\"Norod78/SDXL-YarnArtStyle-LoRA\", filename=\"SDXL_Yarn_Art_Style.safetensors\", repo_type=\"model\")\n",
126
+ "lora_weights = hf_hub_download(repo_id=\"Norod78/SDXL-Psychemelt-style-LoRA\" , filename=\"SDXL_Psychemelt_style_LoRA-000007.safetensors\", repo_type=\"model\")\n",
127
+ "\n",
128
+ "############################################################################################################################################\n",
129
+ "pipe.load_lora_weights(lora_weights)\n",
130
+ "############################################################################################################################################\n"
131
+ ],
132
+ "metadata": {
133
+ "id": "byQMEGplF0FD"
134
+ },
135
+ "execution_count": null,
136
+ "outputs": []
137
+ },
138
+ {
139
+ "cell_type": "code",
140
+ "execution_count": null,
141
+ "metadata": {
142
+ "id": "ksM5XrlmTJKQ"
143
+ },
144
+ "outputs": [],
145
+ "source": [
146
+ "from ip_adapter import IPAdapterPlusXL\n",
147
+ "# load ip-adapter\n",
148
+ "with torch.no_grad():\n",
149
+ " ip_model = IPAdapterPlusXL(pipe, image_encoder_path, ip_ckpt, device, num_tokens=16)\n"
150
+ ]
151
+ },
152
+ {
153
+ "cell_type": "code",
154
+ "source": [
155
+ "import crop_head_dlib5\n",
156
+ "# Detect and crop the head\n",
157
+ "cropped_head = crop_head_dlib5.detect_and_crop_head(\"images/Girl_with_a_Pearl_Earring.jpg\", factor=4.5)\n",
158
+ "image = cropped_head.resize((224, 224))\n",
159
+ "image"
160
+ ],
161
+ "metadata": {
162
+ "id": "2xz5rU5OIA-R"
163
+ },
164
+ "execution_count": null,
165
+ "outputs": []
166
+ },
167
+ {
168
+ "cell_type": "code",
169
+ "execution_count": null,
170
+ "metadata": {
171
+ "id": "xGjbgEdBTJKQ"
172
+ },
173
+ "outputs": [],
174
+ "source": [
175
+ "prompt = \"full body photo of a woman standing outdoors in a field, wearing a colorful dress in an LSD trip psychemelt style\"\n",
176
+ "negative_prompt = \"monochrome, lowres, bad anatomy, worst quality, low quality, blurry, extra limbs, nude, naked, nsfw\"\n",
177
+ "with torch.no_grad():\n",
178
+ " images = ip_model.generate(\n",
179
+ " prompt=prompt, negative_prompt=negative_prompt, pil_image=image, scale=0.85, guidance_scale=7.5, num_samples=4, width=1024, height=1280, num_inference_steps=30, seed=2024,\n",
180
+ " )\n",
181
+ "\n",
182
+ "grid = image_grid(images, 2, 2)\n",
183
+ "grid\n"
184
+ ]
185
+ },
186
+ {
187
+ "cell_type": "code",
188
+ "source": [
189
+ "!wget --continue https://upload.wikimedia.org/wikipedia/commons/0/0f/SDCC_2015_-_Gal_Gadot_%2819742671671%29_%28cropped%29.jpg -O images/gal_gadot_full_body.jpg\n",
190
+ "cropped_head = crop_head_dlib5.detect_and_crop_head(\"images/gal_gadot_full_body.jpg\", factor=2.5)\n",
191
+ "image = cropped_head.resize((224, 224))\n",
192
+ "image"
193
+ ],
194
+ "metadata": {
195
+ "id": "RePEtucoHcyS"
196
+ },
197
+ "execution_count": null,
198
+ "outputs": []
199
+ },
200
+ {
201
+ "cell_type": "code",
202
+ "source": [
203
+ "\n",
204
+ "with torch.no_grad():\n",
205
+ " images = ip_model.generate(\n",
206
+ " prompt=prompt, negative_prompt=negative_prompt, pil_image=image, scale=0.85, guidance_scale=7.5, num_samples=4, width=1024, height=1280, num_inference_steps=30, seed=2024,\n",
207
+ " )\n",
208
+ "\n",
209
+ "grid = image_grid(images, 2, 2)\n",
210
+ "grid"
211
+ ],
212
+ "metadata": {
213
+ "id": "Ic2r3eJoHn6B"
214
+ },
215
+ "execution_count": null,
216
+ "outputs": []
217
+ }
218
+ ],
219
+ "metadata": {
220
+ "accelerator": "GPU",
221
+ "colab": {
222
+ "provenance": [],
223
+ "machine_shape": "hm",
224
+ "gpuType": "L4",
225
+ "private_outputs": true
226
+ },
227
+ "kernelspec": {
228
+ "display_name": "Python 3",
229
+ "name": "python3"
230
+ },
231
+ "language_info": {
232
+ "codemirror_mode": {
233
+ "name": "ipython",
234
+ "version": 3
235
+ },
236
+ "file_extension": ".py",
237
+ "mimetype": "text/x-python",
238
+ "name": "python",
239
+ "nbconvert_exporter": "python",
240
+ "pygments_lexer": "ipython3",
241
+ "version": "3.11.5"
242
+ }
243
+ },
244
+ "nbformat": 4,
245
+ "nbformat_minor": 0
246
+ }