Spaces:
Sleeping
Sleeping
File size: 19,818 Bytes
d392f23 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 | # ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# π§ FUTURE FEATURE: DECLARATIVE JSON RENDERER
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# This module implements a standalone "Video Engine" that renders videos based on a
# JSON specification (similar to Remotion or After Effects Scripting).
#
# NOTE: This is currently EXPERIMENTAL and separate from the main auto-clipping pipeline.
# It is intended for future use cases where precise, programmatic control over
# every frame, text, and transition is required (e.g., frontend-driven editing).
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
import os
import requests
import tempfile
from moviepy.editor import (
VideoFileClip, TextClip, ImageClip, CompositeVideoClip,
ColorClip, AudioFileClip, CompositeAudioClip
)
from pydantic import BaseModel
from typing import List, Optional, Union, Literal
# βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# 1. Define the Schema (The Language of the Engine)
# βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
class Asset(BaseModel):
type: Literal['video', 'image', 'text', 'audio']
src: Optional[str] = None
text: Optional[str] = None
style: Optional[dict] = {} # Font, color, size, bg_color, stroke_color, stroke_width, shadow_color, shadow_offset
class Animation(BaseModel):
type: Literal['fade_in', 'fade_out', 'pop_in', 'scale_in', 'slide_up', 'slide_left']
duration: float = 0.5
class Clip(BaseModel):
asset: Asset
start: float
length: Optional[float] = None
trim_start: float = 0.0
scale: float = 1.0
position: Union[Literal['center', 'top', 'bottom', 'left', 'right'], List[int]] = 'center'
opacity: float = 1.0
volume: float = 1.0
layer: int = 0
animations: List[Animation] = [] # List of animations to apply
class Track(BaseModel):
clips: List[Clip]
class Timeline(BaseModel):
background: str = "#000000"
tracks: List[Track]
class OutputSpec(BaseModel):
format: str = "mp4"
resolution: str = "1080:1920" # width:height
fps: int = 30
class RenderRequest(BaseModel):
timeline: Timeline
output: OutputSpec
# βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# 2. The Engine (JSON -> MoviePy)
# βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
class JSONRenderer:
def __init__(self, output_dir="outputs"):
self.output_dir = output_dir
os.makedirs(output_dir, exist_ok=True)
self.temp_files = []
def _download_asset(self, url):
"""Helper to download assets from URLs"""
if not url.startswith(('http:', 'https:')):
return url
try:
response = requests.get(url, stream=True)
response.raise_for_status()
# Get extension or default
ext = os.path.splitext(url)[1] or ".tmp"
tf = tempfile.NamedTemporaryFile(delete=False, suffix=ext)
for chunk in response.iter_content(chunk_size=8192):
tf.write(chunk)
tf.close()
self.temp_files.append(tf.name)
return tf.name
except Exception as e:
print(f"Failed to download asset: {url} - {e}")
return None
def cleanup(self):
"""Remove temp files"""
for f in self.temp_files:
try:
os.remove(f)
except:
pass
def render(self, request: RenderRequest, output_filename: str):
"""
Takes a JSON spec and renders a video file.
"""
try:
width, height = map(int, request.output.resolution.split(":"))
fps = request.output.fps
# 1. Create Background
final_video_clips = []
max_duration = 0
audio_clips = []
# Background Color
bg_clip = ColorClip(size=(width, height), color=request.timeline.background)
# 2. Process Tracks & Clips
# Flatten all clips and sort by layer
all_clips_spec = []
for track in request.timeline.tracks:
for clip_spec in track.clips:
all_clips_spec.append(clip_spec)
# Sort by layer (ascending)
all_clips_spec.sort(key=lambda x: x.layer)
for clip_spec in all_clips_spec:
clip = self._create_moviepy_clip(clip_spec, width, height)
if clip:
# Apply timing
clip = clip.set_start(clip_spec.start)
# Update max duration
end_time = clip_spec.start + clip.duration
if end_time > max_duration:
max_duration = end_time
# Separate audio/video
if clip_spec.asset.type == 'audio':
audio_clips.append(clip)
else:
final_video_clips.append(clip)
# 3. Final Composition
bg_clip = bg_clip.set_duration(max_duration)
final_video_clips.insert(0, bg_clip)
final_video = CompositeVideoClip(final_video_clips, size=(width, height))
# Handle Audio Mixing
composite_audio_list = []
if final_video.audio:
composite_audio_list.append(final_video.audio)
composite_audio_list.extend(audio_clips)
if composite_audio_list:
final_video.audio = CompositeAudioClip(composite_audio_list)
final_video = final_video.set_duration(max_duration)
# 4. Write File
output_path = os.path.join(self.output_dir, output_filename)
final_video.write_videofile(
output_path,
fps=fps,
codec="libx264",
audio_codec="aac",
threads=4,
preset="medium"
)
return output_path
finally:
self.cleanup()
def _create_moviepy_clip(self, clip_spec: Clip, screen_w, screen_h):
asset = clip_spec.asset
clip = None
try:
src_path = asset.src
if src_path and src_path.startswith(('http', 'https')):
src_path = self._download_asset(src_path)
# Check file existence for local files
if src_path and not os.path.exists(src_path) and not src_path.startswith(('http', 'https')):
# Try relative to project root if absolute fails
if os.path.exists(os.path.abspath(src_path)):
src_path = os.path.abspath(src_path)
# --- Video ---
if asset.type == 'video':
if not src_path: return None
clip = VideoFileClip(src_path)
if clip_spec.length:
end = clip_spec.trim_start + clip_spec.length
clip = clip.subclip(clip_spec.trim_start, min(end, clip.duration))
else:
clip = clip.subclip(clip_spec.trim_start)
# Resize video
if clip_spec.scale != 1.0:
clip = clip.resize(clip_spec.scale)
# Audio Volume
if clip.audio:
clip = clip.volumex(clip_spec.volume)
# --- Image ---
elif asset.type == 'image':
if not src_path: return None
clip = ImageClip(src_path)
if clip_spec.length:
clip = clip.set_duration(clip_spec.length)
if clip_spec.scale != 1.0:
clip = clip.resize(clip_spec.scale)
# --- Text ---
elif asset.type == 'text':
if not asset.text: return None
fontsize = asset.style.get('fontSize', 70)
color = asset.style.get('color', 'white')
font = asset.style.get('font', 'Arial')
bg_color = asset.style.get('backgroundColor', None)
stroke_color = asset.style.get('stroke_color', None)
stroke_width = asset.style.get('stroke_width', 1)
# TextClip wrapper
# Note: You need ImageMagick installed for TextClip
clip = TextClip(
asset.text,
fontsize=fontsize,
color=color,
font=font,
bg_color=bg_color,
stroke_color=stroke_color,
stroke_width=stroke_width,
method='caption',
size=(int(screen_w * 0.9), None) # Auto-wrap
)
if clip_spec.length:
clip = clip.set_duration(clip_spec.length)
# --- Audio ---
elif asset.type == 'audio':
if not src_path: return None
clip = AudioFileClip(src_path)
if clip_spec.length:
end = clip_spec.trim_start + clip_spec.length
clip = clip.subclip(clip_spec.trim_start, min(end, clip.duration))
clip = clip.volumex(clip_spec.volume)
return clip
# --- Common Visual Props ---
if clip:
# 1. Apply Position first
pos = clip_spec.position
if isinstance(pos, list):
pos = tuple(pos)
clip = clip.set_position(pos)
# 2. Apply Opacity
if clip_spec.opacity < 1.0:
clip = clip.set_opacity(clip_spec.opacity)
# 3. Apply Animations
for anim in clip_spec.animations:
clip = self._apply_animation(clip, anim, screen_w, screen_h)
return clip
except Exception as e:
print(f"Error creating clip for asset {asset}: {e}")
return None
def _create_text_clip_from_style(self, text, style, screen_w):
"""Helper to create a TextClip with full styling support"""
try:
fontsize = style.get('fontSize', 70)
color = style.get('color', 'white')
font = style.get('font', 'Arial')
bg_color = style.get('backgroundColor', None)
stroke_color = style.get('stroke_color', None)
stroke_width = style.get('stroke_width', 0)
# Shadow implementation (simple drop shadow via composition if needed,
# but TextClip has limited shadow support directly.
# We can simulate it by creating a black copy behind.)
shadow_color = style.get('shadow_color', None)
shadow_offset = style.get('shadow_offset', (2, 2))
# Main Text
txt_clip = TextClip(
text,
fontsize=fontsize,
color=color,
font=font,
bg_color=bg_color,
stroke_color=stroke_color,
stroke_width=stroke_width,
method='caption',
align='center',
size=(int(screen_w * 0.9), None) # Auto-wrap
)
if shadow_color:
# Create shadow layer
shadow_clip = TextClip(
text,
fontsize=fontsize,
color=shadow_color,
font=font,
method='caption',
align='center',
size=(int(screen_w * 0.9), None)
).set_position(lambda t: (shadow_offset[0], shadow_offset[1])) # Offset relative to parent
# Composite shadow + text
# We need a CompositeVideoClip that fits both
w, h = txt_clip.size
composite = CompositeVideoClip(
[shadow_clip, txt_clip.set_position('center')],
size=(w + abs(shadow_offset[0])*2, h + abs(shadow_offset[1])*2)
)
return composite
return txt_clip
except Exception as e:
print(f"Error creating text clip: {e}")
return None
def _apply_animation(self, clip, anim: Animation, w, h):
"""Apply MoviePy transformations for animations"""
d = anim.duration
if anim.type == 'fade_in':
return clip.fadein(d)
elif anim.type == 'fade_out':
return clip.fadeout(d)
elif anim.type == 'pop_in':
# Scale from 0 to 1 with a slight bounce effect could be complex,
# simple linear scale 0->1 for now
return clip.resize(lambda t: min(1, t / d) if t < d else 1)
elif anim.type == 'scale_in':
# Zoom from 0.8 to 1.0
return clip.resize(lambda t: 0.8 + 0.2 * (t / d) if t < d else 1)
elif anim.type == 'slide_up':
# Move from bottom to original position
# Note: This overrides static position, so needs care.
# We assume 'pos' was set to the final destination.
# Get final x, y. This is tricky in MoviePy as pos can be strings.
# Simplified: Slide from bottom of screen
def slide(t):
if t >= d: return clip.pos(t) # Stay at final
progress = t / d
x, y = clip.pos(t)
# If y is a string (like 'center'), we can't easily calculate offset without computing logic
# Fallback to simple fade if pos is relative, or implement relative sliding later
return x, y # Placeholder: Real sliding requires resolving 'center' to pixels
# Better approach for slide: CompositeVideoClip handles pos better.
# For now, let's use a simple transform if pos is absolute, else skip
pass
return clip
# βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# 3. Helpers (STT -> Timeline)
# βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
def convert_whisper_to_timeline(
whisper_result: dict,
video_path: str,
max_words_per_line: int = 5,
base_style: dict = {},
highlight_style: dict = {}
) -> Timeline:
"""
Convert Whisper STT output to a renderer Timeline.
Args:
whisper_result: The raw output from Whisper (segments with words).
video_path: Path to the source video.
max_words_per_line: Max words to show at once (auto-segmentation).
base_style: Default text style.
highlight_style: Style for the active word (karaoke effect).
"""
tracks = []
# 1. Video Track (Background)
video_track = Track(clips=[
Clip(
asset=Asset(type='video', src=video_path),
start=0,
layer=0
)
])
tracks.append(video_track)
# 2. Text Track (Captions)
text_clips = []
all_words = []
# Flatten segments into a single list of words
if 'segments' in whisper_result:
for seg in whisper_result['segments']:
if 'words' in seg:
all_words.extend(seg['words'])
# Group words into chunks (lines)
for i in range(0, len(all_words), max_words_per_line):
chunk = all_words[i : i + max_words_per_line]
if not chunk: continue
start_time = chunk[0]['start']
end_time = chunk[-1]['end']
text_content = " ".join([w['word'].strip() for w in chunk])
# Build Word objects with highlight timing
words_objs = []
for w in chunk:
words_objs.append(Word(
text=w['word'].strip(),
start=w['start'],
end=w['end'],
style=highlight_style # Active style
))
text_clips.append(Clip(
asset=Asset(
type='text',
text=text_content,
words=words_objs,
style=base_style
),
start=start_time,
length=end_time - start_time,
position='center', # Default position
layer=1
))
tracks.append(Track(clips=text_clips))
return Timeline(background="#000000", tracks=tracks)
def _apply_animation(self, clip, anim: Animation, w, h):
"""Apply MoviePy transformations for animations"""
d = anim.duration
if anim.type == 'fade_in':
return clip.fadein(d)
elif anim.type == 'fade_out':
return clip.fadeout(d)
elif anim.type == 'pop_in':
# Scale from 0 to 1 with a slight bounce effect could be complex,
# simple linear scale 0->1 for now
return clip.resize(lambda t: min(1, t / d) if t < d else 1)
elif anim.type == 'scale_in':
# Zoom from 0.8 to 1.0
return clip.resize(lambda t: 0.8 + 0.2 * (t / d) if t < d else 1)
elif anim.type == 'slide_up':
# Move from bottom to original position
# Note: This overrides static position, so needs care.
# We assume 'pos' was set to the final destination.
# Get final x, y. This is tricky in MoviePy as pos can be strings.
# Simplified: Slide from bottom of screen
def slide(t):
if t >= d: return clip.pos(t) # Stay at final
progress = t / d
x, y = clip.pos(t)
# If y is a string (like 'center'), we can't easily calculate offset without computing logic
# Fallback to simple fade if pos is relative, or implement relative sliding later
return x, y # Placeholder: Real sliding requires resolving 'center' to pixels
# Better approach for slide: CompositeVideoClip handles pos better.
# For now, let's use a simple transform if pos is absolute, else skip
pass
return clip |