Spaces:
Running on Zero
Running on Zero
File size: 10,843 Bytes
2680bd5 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 | from typing import Dict, List
import numpy as np
from tqdm import tqdm
import math, json
import textwrap
from matplotlib import pyplot as plt
from matplotlib.animation import FuncAnimation
from matplotlib import gridspec
from matplotlib.font_manager import FontProperties
import matplotlib.gridspec as gridspec
from pygments import highlight
from pygments.lexers import JsonLexer
from pygments.formatter import Formatter
from pygments.styles import get_style_by_name
from pygments.token import Token
# NOTE: bihand_mano2mesh (from .mano2mesh) is imported lazily where used (mano viz mode only),
# so skeleton visualization works without the license-gated MANO model files.
from .mesh_visualizer import Mesh_Visualize_Helper
from .skeleton_visualizer import Skeleton_Visualize_Helper
from ..utils import smart_wrap
class MatplotlibFormatter(Formatter):
def __init__(self, **kwargs):
super().__init__(**kwargs)
self.data = []
def format(self, tokensource, outfile):
self.data = []
for ttype, value in tokensource:
style = self.style.style_for_token(ttype)
color = style['color']
if color:
color = f"#{color}"
else:
style_bg_is_dark = self.style.background_color < '#777'
color = '#ffffff' if style_bg_is_dark else '#000000'
self.data.append((value, color, ttype))
class JsonPlotter:
@staticmethod
def plot_json(
data_dict: dict | list,
ax: plt.Axes,
width: int = 80,
style_name: str = 'monokai',
font_family: str = 'monospace',
font_size: int = 10
):
json_str = json.dumps(data_dict, indent=4, ensure_ascii=False)
lexer = JsonLexer()
style = get_style_by_name(style_name)
formatter = MatplotlibFormatter(style=style)
highlight(json_str, lexer, formatter)
ax.set_facecolor(style.background_color)
fig = ax.get_figure()
font = FontProperties(family=font_family, size=font_size)
sample_text_obj = ax.text(0, 0, 'M', fontproperties=font, visible=False)
try:
fig.canvas.draw()
text_bbox = sample_text_obj.get_window_extent()
ax_bbox = ax.get_window_extent()
if ax_bbox.width == 0 or text_bbox.width == 0:
raise ValueError("Canvas has zero width, cannot calculate font metrics.")
char_width = text_bbox.width / ax_bbox.width
line_height = text_bbox.height / ax_bbox.height * 1.6
print(f"Char Width: {char_width}, Line Height: {line_height}")
except Exception:
char_width = 0.006 * (font_size / 10)
line_height = 0.025 * (font_size / 10)
finally:
sample_text_obj.remove()
lines_of_tokens = []
current_line = []
current_char_count = 0
wrapper = textwrap.TextWrapper(
width=width,
break_long_words=True,
break_on_hyphens=False
)
for text, color, ttype in formatter.data:
if text == '\n':
lines_of_tokens.append(current_line)
current_line = []
current_char_count = 0
continue
is_string = ttype in Token.Literal.String
if is_string and (current_char_count + len(text)) > width:
leading_spaces = current_char_count
wrapper.initial_indent = " " * 0
wrapper.subsequent_indent = " " * (leading_spaces + 4)
content = text[1:-1]
wrapped_lines = wrapper.wrap(content)
current_line.append((text[0], color))
for i, line_content in enumerate(wrapped_lines):
if i == 0:
current_line.append((line_content, color))
else:
lines_of_tokens.append(current_line)
current_line = [ (line_content.lstrip(), color) ]
current_line.append((text[-1], color))
else:
current_line.append((text, color))
current_char_count += len(text)
if current_line:
lines_of_tokens.append(current_line)
x_pos, y_pos = 0.02, 0.98
left_margin = x_pos
for line_tokens in lines_of_tokens:
x_pos = left_margin
for text, color in line_tokens:
if not text: continue
ax.text(x_pos, y_pos, text, color=color, fontproperties=font, ha='left', va='top')
x_pos += len(text) * char_width
y_pos -= line_height
ax.axis('off')
ax.set_ylim(0, 1)
ax.set_xlim(0, 1)
class Visualize_Helper:
@staticmethod
def initialize():
fig = plt.figure(figsize=(10, 10), layout='constrained')
gs = gridspec.GridSpec(2, 2, figure=fig)
skeleton_ax = fig.add_subplot(gs[0, 0], projection='3d')
mano_ax = fig.add_subplot(gs[0, 1], projection='3d')
annotation_ax = fig.add_subplot(gs[1, :])
annotation_ax.axis("off")
return fig, skeleton_ax, mano_ax, annotation_ax
@staticmethod
def format_plain_string(text:str, width:int=90):
return '\n'.join(textwrap.wrap(text, width=width))
@staticmethod
def create_3d_animation(
left_motion:np.ndarray, right_motion:np.ndarray,
left_mano:Dict[str, np.ndarray]|None=None, right_mano:Dict[str, np.ndarray]|None=None,
annotation:str | dict | list="",
save_path:str|None=None, fps:int=30
):
'''
left_motion: (T, J, 3)
right_motion: (T, J, 3)
left_mano:
shape: (T, 10)
pose: (T, 48),
trans: (T, 3)
right_mano: the same as left_mano
'''
fig, skeleton_ax, mano_ax, annotation_ax = Visualize_Helper.initialize()
print(f"Annotation Type: {type(annotation)}")
if isinstance(annotation, dict) or isinstance(annotation, list):
JsonPlotter.plot_json(annotation, annotation_ax)
elif isinstance(annotation, str):
print(f"annotation: {annotation.__repr__()}")
annotation = '\n'.join(textwrap.wrap(annotation, width=90, replace_whitespace=False, tabsize=4))
annotation_ax.text(
x=0.1, y=0.95, s=annotation, ha='left', va='top', fontsize=14
)
frame_num = fig.text(
x=0.5, y=0.95, s="", ha='center', va='top', fontsize=16
)
skeleton_helper = Skeleton_Visualize_Helper(skeleton_ax, left_motion, right_motion)
skeleton_helper.initialize_ax()
if left_mano is not None and right_mano is not None:
from .mano2mesh import bihand_mano2mesh
vertices, faces = bihand_mano2mesh(left_mano, right_mano)
mesh_helper = Mesh_Visualize_Helper(mano_ax, vertices, faces)
mesh_helper.initialize_ax()
def update(frame):
frame_num.set_text(f"Frame: {frame} / {left_motion.shape[0]}")
skeleton_helper.draw(frame)
if left_mano is not None and right_mano is not None:
mesh_helper.draw_mesh(frame)
ani = FuncAnimation(fig, update, frames=left_motion.shape[0])
if save_path:
pbar = tqdm(total=left_motion.shape[0], desc='Exporting animation')
ani.save(save_path, writer='ffmpeg', fps=fps, progress_callback=lambda x, y: pbar.update(1))
plt.close(fig)
class MultiMotionVisualizer:
def initialize(number_of_motions:int):
rows = int(math.floor(math.sqrt(number_of_motions)))
columns = int(math.ceil(number_of_motions / rows))
fig = plt.figure(figsize=(4 * columns, 4 * rows + 8))
gs = gridspec.GridSpec(rows + 1, columns, figure=fig, height_ratios=[4] * rows + [8])
motion_axes = []
for i in range(number_of_motions):
ax = fig.add_subplot(gs[i // columns, i % columns], projection='3d')
motion_axes.append(ax)
motion_axes = np.array(motion_axes)
text_ax = fig.add_subplot(gs[rows, :])
text_ax.axis("off")
return fig, motion_axes, text_ax
@staticmethod
def create_3d_animation(
motions:List[Dict[str, np.ndarray | str | dict]],
text:str="",
save_path:str=None, fps:int=30
):
'''
if motion['type'] == 'skeleton'
motion = {
'type': 'skeleton',
'left_motion': (T, J, 3),
'right_motion': (T, J, 3)
}
elif motion['type'] == 'mano'
motion = {
'type': 'mano',
'left_motion': dict
'right_motion': dict'
}
'''
fig, axes, text_ax = MultiMotionVisualizer.initialize(len(motions))
drawers = []
frame_num = 0
for motion_id, motion in enumerate(motions):
if motion['type'] == 'skeleton':
helper = Skeleton_Visualize_Helper(
axes[motion_id], motion['left_motion'], motion['right_motion'], motion.get('title', None)
)
frame_num = max(frame_num, motion['left_motion'].shape[0])
elif motion['type'] == 'mano':
from .mano2mesh import bihand_mano2mesh
vertices, faces = bihand_mano2mesh(
motion['left_motion'], motion['right_motion']
)
helper = Mesh_Visualize_Helper(
axes[motion_id], vertices, faces, motion.get('title', None)
)
frame_num = max(vertices.shape[0], frame_num)
helper.initialize_ax()
drawers.append(helper)
frame_num_text = fig.text(
x=0.5, y=0.95, s="", ha='center', va='top', fontsize=16
)
# text = '\n'.join(textwrap.wrap(text, width=90))
text = smart_wrap(text, width=90)
text_ax.text(
x=0.01, y=0.5, s=text, ha='left', va='center', fontsize=14
)
def update(frame):
frame_num_text.set_text(f"Frame: {frame} / {frame_num}")
for drawer in drawers:
if isinstance(drawer, Skeleton_Visualize_Helper):
drawer.draw(frame)
elif isinstance(drawer, Mesh_Visualize_Helper):
drawer.draw_mesh(frame)
ani = FuncAnimation(fig, update, frames=frame_num)
if save_path:
name = save_path.split('/')[-1]
pbar = tqdm(total=frame_num, desc=f'Exporting animation to {name}')
ani.save(save_path, writer='ffmpeg', fps=fps, progress_callback=lambda x, y: pbar.update(1))
plt.close(fig)
|