Commit
·
b27a485
1
Parent(s):
87286e6
Upload HyperNeRF_Render_Video_clean.ipynb
Browse files
HyperNeRF_Render_Video_clean.ipynb
ADDED
@@ -0,0 +1,473 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"cells": [
|
3 |
+
{
|
4 |
+
"cell_type": "markdown",
|
5 |
+
"metadata": {
|
6 |
+
"id": "QMMWf9AQcdlp"
|
7 |
+
},
|
8 |
+
"source": [
|
9 |
+
"# Render a HyperNeRF video!\n",
|
10 |
+
"\n",
|
11 |
+
"**Author**: [Keunhong Park](https://keunhong.com)\n",
|
12 |
+
"\n",
|
13 |
+
"[[Project Page](https://hypernerf.github.io)]\n",
|
14 |
+
"[[Paper](https://arxiv.org/abs/2106.13228)]\n",
|
15 |
+
"[[GitHub](https://github.com/google/hypernerf)]\n",
|
16 |
+
"\n",
|
17 |
+
"This notebook renders a video using the test cameras generated in the capture processing notebook.\n",
|
18 |
+
"\n",
|
19 |
+
"You can also load your own custom cameras by modifying the code slightly.\n",
|
20 |
+
"\n",
|
21 |
+
"### Instructions\n",
|
22 |
+
"\n",
|
23 |
+
"1. Convert a video into our dataset format using the [capture processing notebook](https://colab.sandbox.google.com/github/google/nerfies/blob/main/notebooks/Nerfies_Capture_Processing.ipynb).\n",
|
24 |
+
"2. Train a HyperNeRF model using the [training notebook](https://colab.sandbox.google.com/github/google/hypernerf/blob/main/notebooks/HyperNeRF_Training.ipynb)\n",
|
25 |
+
"3. Run this notebook!\n",
|
26 |
+
"\n",
|
27 |
+
"\n",
|
28 |
+
"### Notes\n",
|
29 |
+
" * Please report issues on the [GitHub issue tracker](https://github.com/google/hypernerf/issues)."
|
30 |
+
]
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"cell_type": "markdown",
|
34 |
+
"metadata": {
|
35 |
+
"id": "gHqkIo4hcGou"
|
36 |
+
},
|
37 |
+
"source": [
|
38 |
+
"## Environment Setup"
|
39 |
+
]
|
40 |
+
},
|
41 |
+
{
|
42 |
+
"cell_type": "code",
|
43 |
+
"execution_count": null,
|
44 |
+
"metadata": {
|
45 |
+
"colab": {
|
46 |
+
"base_uri": "https://localhost:8080/"
|
47 |
+
},
|
48 |
+
"id": "ws81Eje47SuV",
|
49 |
+
"outputId": "2fa89ef5-4030-46d4-e2d9-2eebffd1b0f9"
|
50 |
+
},
|
51 |
+
"outputs": [],
|
52 |
+
"source": [
|
53 |
+
"#!wget https://raw.githubusercontent.com/google/hypernerf/main/requirements.txt\n",
|
54 |
+
"!wget https://raw.githubusercontent.com/xieyizheng/hypernerf/main/requirements.txt\n",
|
55 |
+
"!python --version\n",
|
56 |
+
"!pip install -r requirements.txt\n",
|
57 |
+
"\n",
|
58 |
+
"#if freshly installed, recommend to restart the runtime!"
|
59 |
+
]
|
60 |
+
},
|
61 |
+
{
|
62 |
+
"cell_type": "code",
|
63 |
+
"execution_count": null,
|
64 |
+
"metadata": {
|
65 |
+
"colab": {
|
66 |
+
"base_uri": "https://localhost:8080/"
|
67 |
+
},
|
68 |
+
"id": "-3T2lBKBcIGP",
|
69 |
+
"outputId": "6bcc5d9c-108a-4c2b-bef5-fe140c87b3fb"
|
70 |
+
},
|
71 |
+
"outputs": [],
|
72 |
+
"source": [
|
73 |
+
"# @title Configure notebook runtime\n",
|
74 |
+
"# @markdown If you would like to use a GPU runtime instead, change the runtime type by going to `Runtime > Change runtime type`. \n",
|
75 |
+
"# @markdown You will have to use a smaller batch size on GPU.\n",
|
76 |
+
"import jax\n",
|
77 |
+
"runtime_type = 'gpu' # @param ['gpu', 'tpu']\n",
|
78 |
+
"if runtime_type == 'tpu':\n",
|
79 |
+
" import jax.tools.colab_tpu\n",
|
80 |
+
" jax.tools.colab_tpu.setup_tpu()\n",
|
81 |
+
"\n",
|
82 |
+
"print('Detected Devices:', jax.devices())"
|
83 |
+
]
|
84 |
+
},
|
85 |
+
{
|
86 |
+
"cell_type": "code",
|
87 |
+
"execution_count": null,
|
88 |
+
"metadata": {
|
89 |
+
"colab": {
|
90 |
+
"base_uri": "https://localhost:8080/"
|
91 |
+
},
|
92 |
+
"id": "82kU-W1NcNTW",
|
93 |
+
"outputId": "08a21bab-c3cc-43a0-9f1f-fb7e843a8aaa"
|
94 |
+
},
|
95 |
+
"outputs": [],
|
96 |
+
"source": [
|
97 |
+
"# @title Mount Google Drive\n",
|
98 |
+
"# @markdown Mount Google Drive onto `/content/gdrive`. You can skip this if running locally.\n",
|
99 |
+
"\n",
|
100 |
+
"from google.colab import drive\n",
|
101 |
+
"drive.mount('/content/gdrive')"
|
102 |
+
]
|
103 |
+
},
|
104 |
+
{
|
105 |
+
"cell_type": "code",
|
106 |
+
"execution_count": null,
|
107 |
+
"metadata": {
|
108 |
+
"id": "YIDbV769cPn1"
|
109 |
+
},
|
110 |
+
"outputs": [],
|
111 |
+
"source": [
|
112 |
+
"# @title Define imports and utility functions.\n",
|
113 |
+
"\n",
|
114 |
+
"import jax\n",
|
115 |
+
"from jax.config import config as jax_config\n",
|
116 |
+
"import jax.numpy as jnp\n",
|
117 |
+
"from jax import grad, jit, vmap\n",
|
118 |
+
"from jax import random\n",
|
119 |
+
"\n",
|
120 |
+
"import flax\n",
|
121 |
+
"import flax.linen as nn\n",
|
122 |
+
"from flax import jax_utils\n",
|
123 |
+
"from flax import optim\n",
|
124 |
+
"from flax.metrics import tensorboard\n",
|
125 |
+
"from flax.training import checkpoints\n",
|
126 |
+
"\n",
|
127 |
+
"from absl import logging\n",
|
128 |
+
"from io import BytesIO\n",
|
129 |
+
"import random as pyrandom\n",
|
130 |
+
"import numpy as np\n",
|
131 |
+
"import PIL\n",
|
132 |
+
"import IPython\n",
|
133 |
+
"import tempfile\n",
|
134 |
+
"import imageio\n",
|
135 |
+
"import mediapy\n",
|
136 |
+
"from IPython.display import display, HTML\n",
|
137 |
+
"from base64 import b64encode\n",
|
138 |
+
"\n",
|
139 |
+
"\n",
|
140 |
+
"# Monkey patch logging.\n",
|
141 |
+
"def myprint(msg, *args, **kwargs):\n",
|
142 |
+
" print(msg % args)\n",
|
143 |
+
"\n",
|
144 |
+
"logging.info = myprint \n",
|
145 |
+
"logging.warn = myprint\n",
|
146 |
+
"logging.error = myprint"
|
147 |
+
]
|
148 |
+
},
|
149 |
+
{
|
150 |
+
"cell_type": "code",
|
151 |
+
"execution_count": null,
|
152 |
+
"metadata": {
|
153 |
+
"colab": {
|
154 |
+
"base_uri": "https://localhost:8080/",
|
155 |
+
"height": 1000
|
156 |
+
},
|
157 |
+
"id": "2QYJ7dyMcw2f",
|
158 |
+
"outputId": "73b49855-05f6-4377-a57a-3a5a061c980a"
|
159 |
+
},
|
160 |
+
"outputs": [],
|
161 |
+
"source": [
|
162 |
+
"# @title Model and dataset configuration\n",
|
163 |
+
"# @markdown Change the directories to where you saved your capture and experiment.\n",
|
164 |
+
"\n",
|
165 |
+
"\n",
|
166 |
+
"from pathlib import Path\n",
|
167 |
+
"from pprint import pprint\n",
|
168 |
+
"import gin\n",
|
169 |
+
"from IPython.display import display, Markdown\n",
|
170 |
+
"\n",
|
171 |
+
"from hypernerf import models\n",
|
172 |
+
"from hypernerf import modules\n",
|
173 |
+
"from hypernerf import warping\n",
|
174 |
+
"from hypernerf import datasets\n",
|
175 |
+
"from hypernerf import configs\n",
|
176 |
+
"\n",
|
177 |
+
"\n",
|
178 |
+
"# @markdown The working directory where the trained model is.\n",
|
179 |
+
"train_dir = '/content/gdrive/My Drive/nerfies/hypernerf_experiments/dvd/exp2' # @param {type: \"string\"}\n",
|
180 |
+
"# @markdown The directory to the dataset capture.\n",
|
181 |
+
"data_dir = '/content/gdrive/My Drive/nerfies/captures/dvd' # @param {type: \"string\"}\n",
|
182 |
+
"\n",
|
183 |
+
"checkpoint_dir = Path(train_dir, 'checkpoints')\n",
|
184 |
+
"checkpoint_dir.mkdir(exist_ok=True, parents=True)\n",
|
185 |
+
"\n",
|
186 |
+
"config_path = Path(train_dir, 'config.gin')\n",
|
187 |
+
"with open(config_path, 'r') as f:\n",
|
188 |
+
" logging.info('Loading config from %s', config_path)\n",
|
189 |
+
" config_str = f.read()\n",
|
190 |
+
"gin.parse_config(config_str)\n",
|
191 |
+
"\n",
|
192 |
+
"config_path = Path(train_dir, 'config.gin')\n",
|
193 |
+
"with open(config_path, 'w') as f:\n",
|
194 |
+
" logging.info('Saving config to %s', config_path)\n",
|
195 |
+
" f.write(config_str)\n",
|
196 |
+
"\n",
|
197 |
+
"exp_config = configs.ExperimentConfig()\n",
|
198 |
+
"train_config = configs.TrainConfig()\n",
|
199 |
+
"eval_config = configs.EvalConfig()\n",
|
200 |
+
"\n",
|
201 |
+
"display(Markdown(\n",
|
202 |
+
" gin.config.markdown(gin.config_str())))"
|
203 |
+
]
|
204 |
+
},
|
205 |
+
{
|
206 |
+
"cell_type": "code",
|
207 |
+
"execution_count": null,
|
208 |
+
"metadata": {
|
209 |
+
"cellView": "form",
|
210 |
+
"colab": {
|
211 |
+
"base_uri": "https://localhost:8080/",
|
212 |
+
"height": 439
|
213 |
+
},
|
214 |
+
"id": "6T7LQ5QSmu4o",
|
215 |
+
"outputId": "399c441e-b125-4a99-b36e-7b58e0256858"
|
216 |
+
},
|
217 |
+
"outputs": [],
|
218 |
+
"source": [
|
219 |
+
"# @title Create datasource and show an example.\n",
|
220 |
+
"\n",
|
221 |
+
"from hypernerf import datasets\n",
|
222 |
+
"from hypernerf import image_utils\n",
|
223 |
+
"\n",
|
224 |
+
"dummy_model = models.NerfModel({}, 0, 0)\n",
|
225 |
+
"datasource = exp_config.datasource_cls(\n",
|
226 |
+
" image_scale=exp_config.image_scale,\n",
|
227 |
+
" random_seed=exp_config.random_seed,\n",
|
228 |
+
" # Enable metadata based on model needs.\n",
|
229 |
+
" use_warp_id=dummy_model.use_warp,\n",
|
230 |
+
" use_appearance_id=(\n",
|
231 |
+
" dummy_model.nerf_embed_key == 'appearance'\n",
|
232 |
+
" or dummy_model.hyper_embed_key == 'appearance'),\n",
|
233 |
+
" use_camera_id=dummy_model.nerf_embed_key == 'camera',\n",
|
234 |
+
" use_time=dummy_model.warp_embed_key == 'time')\n",
|
235 |
+
"\n",
|
236 |
+
"mediapy.show_image(datasource.load_rgb(datasource.train_ids[0]))"
|
237 |
+
]
|
238 |
+
},
|
239 |
+
{
|
240 |
+
"cell_type": "code",
|
241 |
+
"execution_count": null,
|
242 |
+
"metadata": {
|
243 |
+
"colab": {
|
244 |
+
"base_uri": "https://localhost:8080/"
|
245 |
+
},
|
246 |
+
"id": "jEO3xcxpnCqx",
|
247 |
+
"outputId": "15e2646e-cf00-4c86-f110-86e21b686813"
|
248 |
+
},
|
249 |
+
"outputs": [],
|
250 |
+
"source": [
|
251 |
+
"# @title Load model\n",
|
252 |
+
"# @markdown Defines the model and initializes its parameters.\n",
|
253 |
+
"\n",
|
254 |
+
"from flax.training import checkpoints\n",
|
255 |
+
"from hypernerf import models\n",
|
256 |
+
"from hypernerf import model_utils\n",
|
257 |
+
"from hypernerf import schedules\n",
|
258 |
+
"from hypernerf import training\n",
|
259 |
+
"\n",
|
260 |
+
"rng = random.PRNGKey(exp_config.random_seed)\n",
|
261 |
+
"np.random.seed(exp_config.random_seed + jax.process_index())\n",
|
262 |
+
"devices_to_use = jax.devices()\n",
|
263 |
+
"\n",
|
264 |
+
"learning_rate_sched = schedules.from_config(train_config.lr_schedule)\n",
|
265 |
+
"nerf_alpha_sched = schedules.from_config(train_config.nerf_alpha_schedule)\n",
|
266 |
+
"warp_alpha_sched = schedules.from_config(train_config.warp_alpha_schedule)\n",
|
267 |
+
"elastic_loss_weight_sched = schedules.from_config(\n",
|
268 |
+
"train_config.elastic_loss_weight_schedule)\n",
|
269 |
+
"hyper_alpha_sched = schedules.from_config(train_config.hyper_alpha_schedule)\n",
|
270 |
+
"hyper_sheet_alpha_sched = schedules.from_config(\n",
|
271 |
+
" train_config.hyper_sheet_alpha_schedule)\n",
|
272 |
+
"\n",
|
273 |
+
"rng, key = random.split(rng)\n",
|
274 |
+
"params = {}\n",
|
275 |
+
"model, params['model'] = models.construct_nerf(\n",
|
276 |
+
" key,\n",
|
277 |
+
" batch_size=train_config.batch_size,\n",
|
278 |
+
" embeddings_dict=datasource.embeddings_dict,\n",
|
279 |
+
" near=datasource.near,\n",
|
280 |
+
" far=datasource.far)\n",
|
281 |
+
"\n",
|
282 |
+
"optimizer_def = optim.Adam(learning_rate_sched(0))\n",
|
283 |
+
"optimizer = optimizer_def.create(params)\n",
|
284 |
+
"\n",
|
285 |
+
"state = model_utils.TrainState(\n",
|
286 |
+
" optimizer=optimizer,\n",
|
287 |
+
" nerf_alpha=nerf_alpha_sched(0),\n",
|
288 |
+
" warp_alpha=warp_alpha_sched(0),\n",
|
289 |
+
" hyper_alpha=hyper_alpha_sched(0),\n",
|
290 |
+
" hyper_sheet_alpha=hyper_sheet_alpha_sched(0))\n",
|
291 |
+
"scalar_params = training.ScalarParams(\n",
|
292 |
+
" learning_rate=learning_rate_sched(0),\n",
|
293 |
+
" elastic_loss_weight=elastic_loss_weight_sched(0),\n",
|
294 |
+
" warp_reg_loss_weight=train_config.warp_reg_loss_weight,\n",
|
295 |
+
" warp_reg_loss_alpha=train_config.warp_reg_loss_alpha,\n",
|
296 |
+
" warp_reg_loss_scale=train_config.warp_reg_loss_scale,\n",
|
297 |
+
" background_loss_weight=train_config.background_loss_weight,\n",
|
298 |
+
" hyper_reg_loss_weight=train_config.hyper_reg_loss_weight)\n",
|
299 |
+
"\n",
|
300 |
+
"logging.info('Restoring checkpoint from %s', checkpoint_dir)\n",
|
301 |
+
"state = checkpoints.restore_checkpoint(checkpoint_dir, state)\n",
|
302 |
+
"step = state.optimizer.state.step + 1\n",
|
303 |
+
"state = jax_utils.replicate(state, devices=devices_to_use)\n",
|
304 |
+
"del params"
|
305 |
+
]
|
306 |
+
},
|
307 |
+
{
|
308 |
+
"cell_type": "code",
|
309 |
+
"execution_count": null,
|
310 |
+
"metadata": {
|
311 |
+
"cellView": "form",
|
312 |
+
"id": "2KYhbpsklwAy"
|
313 |
+
},
|
314 |
+
"outputs": [],
|
315 |
+
"source": [
|
316 |
+
"# @title Define pmapped render function.\n",
|
317 |
+
"\n",
|
318 |
+
"import functools\n",
|
319 |
+
"from hypernerf import evaluation\n",
|
320 |
+
"\n",
|
321 |
+
"devices = jax.devices()\n",
|
322 |
+
"\n",
|
323 |
+
"\n",
|
324 |
+
"def _model_fn(key_0, key_1, params, rays_dict, extra_params):\n",
|
325 |
+
" out = model.apply({'params': params},\n",
|
326 |
+
" rays_dict,\n",
|
327 |
+
" extra_params=extra_params,\n",
|
328 |
+
" rngs={\n",
|
329 |
+
" 'coarse': key_0,\n",
|
330 |
+
" 'fine': key_1\n",
|
331 |
+
" },\n",
|
332 |
+
" mutable=False)\n",
|
333 |
+
" return jax.lax.all_gather(out, axis_name='batch')\n",
|
334 |
+
"\n",
|
335 |
+
"pmodel_fn = jax.pmap(\n",
|
336 |
+
" # Note rng_keys are useless in eval mode since there's no randomness.\n",
|
337 |
+
" _model_fn,\n",
|
338 |
+
" in_axes=(0, 0, 0, 0, 0), # Only distribute the data input.\n",
|
339 |
+
" devices=devices_to_use,\n",
|
340 |
+
" axis_name='batch',\n",
|
341 |
+
")\n",
|
342 |
+
"\n",
|
343 |
+
"render_fn = functools.partial(evaluation.render_image,\n",
|
344 |
+
" model_fn=pmodel_fn,\n",
|
345 |
+
" device_count=len(devices),\n",
|
346 |
+
" chunk=eval_config.chunk)"
|
347 |
+
]
|
348 |
+
},
|
349 |
+
{
|
350 |
+
"cell_type": "code",
|
351 |
+
"execution_count": null,
|
352 |
+
"metadata": {
|
353 |
+
"colab": {
|
354 |
+
"base_uri": "https://localhost:8080/"
|
355 |
+
},
|
356 |
+
"id": "73Fq0kNcmAra",
|
357 |
+
"outputId": "01f7bcee-833f-47fb-d2ab-0a9a2c15837f"
|
358 |
+
},
|
359 |
+
"outputs": [],
|
360 |
+
"source": [
|
361 |
+
"# @title Load cameras.\n",
|
362 |
+
"\n",
|
363 |
+
"from hypernerf import utils\n",
|
364 |
+
"\n",
|
365 |
+
"camera_path = 'camera-paths/orbit-mild' # @param {type: 'string'}\n",
|
366 |
+
"\n",
|
367 |
+
"camera_dir = Path(data_dir, camera_path)\n",
|
368 |
+
"print(f'Loading cameras from {camera_dir}')\n",
|
369 |
+
"test_camera_paths = datasource.glob_cameras(camera_dir)\n",
|
370 |
+
"test_cameras = utils.parallel_map(datasource.load_camera, test_camera_paths, show_pbar=True)"
|
371 |
+
]
|
372 |
+
},
|
373 |
+
{
|
374 |
+
"cell_type": "code",
|
375 |
+
"execution_count": null,
|
376 |
+
"metadata": {
|
377 |
+
"colab": {
|
378 |
+
"base_uri": "https://localhost:8080/",
|
379 |
+
"height": 1000
|
380 |
+
},
|
381 |
+
"id": "aP9LjiAZmoRc",
|
382 |
+
"outputId": "811dfbc3-ccbc-4748-dee8-92281ea01b2c"
|
383 |
+
},
|
384 |
+
"outputs": [],
|
385 |
+
"source": [
|
386 |
+
"# @title Render video frames.\n",
|
387 |
+
"from hypernerf import visualization as viz\n",
|
388 |
+
"\n",
|
389 |
+
"\n",
|
390 |
+
"rng = rng + jax.process_index() # Make random seed separate across hosts.\n",
|
391 |
+
"keys = random.split(rng, len(devices))\n",
|
392 |
+
"\n",
|
393 |
+
"results = []\n",
|
394 |
+
"for i in range(len(test_cameras)):\n",
|
395 |
+
" print(f'Rendering frame {i+1}/{len(test_cameras)}')\n",
|
396 |
+
" camera = test_cameras[i]\n",
|
397 |
+
" batch = datasets.camera_to_rays(camera)\n",
|
398 |
+
" batch['metadata'] = {\n",
|
399 |
+
" 'appearance': jnp.zeros_like(batch['origins'][..., 0, jnp.newaxis], jnp.uint32),\n",
|
400 |
+
" 'warp': jnp.zeros_like(batch['origins'][..., 0, jnp.newaxis], jnp.uint32),\n",
|
401 |
+
" }\n",
|
402 |
+
" #these two are the \"ambient dimensions\" or \"time axis\" for rendering\n",
|
403 |
+
" batch['metadata']['appearance'] += i\n",
|
404 |
+
" batch['metadata']['warp'] += i\n",
|
405 |
+
"\n",
|
406 |
+
" render = render_fn(state, batch, rng=rng)\n",
|
407 |
+
" rgb = np.array(render['rgb'])\n",
|
408 |
+
" depth_med = np.array(render['med_depth'])\n",
|
409 |
+
" results.append((rgb, depth_med))\n",
|
410 |
+
" depth_viz = viz.colorize(depth_med.squeeze(), cmin=datasource.near, cmax=datasource.far, invert=True)\n",
|
411 |
+
" mediapy.show_images([rgb, depth_viz])"
|
412 |
+
]
|
413 |
+
},
|
414 |
+
{
|
415 |
+
"cell_type": "code",
|
416 |
+
"execution_count": null,
|
417 |
+
"metadata": {
|
418 |
+
"id": "_5hHR9XVm8Ix"
|
419 |
+
},
|
420 |
+
"outputs": [],
|
421 |
+
"source": [
|
422 |
+
"# @title Show rendered video.\n",
|
423 |
+
"\n",
|
424 |
+
"fps = 30 # @param {type:'number'}\n",
|
425 |
+
"\n",
|
426 |
+
"frames = []\n",
|
427 |
+
"for rgb, depth in results:\n",
|
428 |
+
" depth_viz = viz.colorize(depth.squeeze(), cmin=datasource.near, cmax=datasource.far, invert=True)\n",
|
429 |
+
" frame = np.concatenate([rgb, depth_viz], axis=1)\n",
|
430 |
+
" frames.append(image_utils.image_to_uint8(frame))\n",
|
431 |
+
"\n",
|
432 |
+
"mediapy.show_video(frames, fps=fps)"
|
433 |
+
]
|
434 |
+
},
|
435 |
+
{
|
436 |
+
"cell_type": "code",
|
437 |
+
"execution_count": null,
|
438 |
+
"metadata": {
|
439 |
+
"id": "WW32AVGR0Vwh"
|
440 |
+
},
|
441 |
+
"outputs": [],
|
442 |
+
"source": []
|
443 |
+
}
|
444 |
+
],
|
445 |
+
"metadata": {
|
446 |
+
"accelerator": "GPU",
|
447 |
+
"colab": {
|
448 |
+
"gpuType": "T4",
|
449 |
+
"machine_shape": "hm",
|
450 |
+
"provenance": []
|
451 |
+
},
|
452 |
+
"gpuClass": "standard",
|
453 |
+
"kernelspec": {
|
454 |
+
"display_name": "Python 3 (ipykernel)",
|
455 |
+
"language": "python",
|
456 |
+
"name": "python3"
|
457 |
+
},
|
458 |
+
"language_info": {
|
459 |
+
"codemirror_mode": {
|
460 |
+
"name": "ipython",
|
461 |
+
"version": 3
|
462 |
+
},
|
463 |
+
"file_extension": ".py",
|
464 |
+
"mimetype": "text/x-python",
|
465 |
+
"name": "python",
|
466 |
+
"nbconvert_exporter": "python",
|
467 |
+
"pygments_lexer": "ipython3",
|
468 |
+
"version": "3.10.10"
|
469 |
+
}
|
470 |
+
},
|
471 |
+
"nbformat": 4,
|
472 |
+
"nbformat_minor": 1
|
473 |
+
}
|