Update core/visual_engine.py
Browse files- core/visual_engine.py +207 -539
core/visual_engine.py
CHANGED
@@ -2,16 +2,13 @@
|
|
2 |
from PIL import Image, ImageDraw, ImageFont, ImageOps
|
3 |
# --- MONKEY PATCH FOR Image.ANTIALIAS ---
|
4 |
try:
|
5 |
-
if hasattr(Image, 'Resampling') and hasattr(Image.Resampling, 'LANCZOS'):
|
6 |
-
if not hasattr(Image, 'ANTIALIAS'):
|
7 |
-
|
8 |
-
|
9 |
-
if not hasattr(Image, 'ANTIALIAS'):
|
10 |
-
Image.ANTIALIAS = Image.LANCZOS
|
11 |
elif not hasattr(Image, 'ANTIALIAS'):
|
12 |
-
|
13 |
-
except Exception as e_mp:
|
14 |
-
print(f"WARNING: ANTIALIAS monkey-patch error: {e_mp}")
|
15 |
# --- END MONKEY PATCH ---
|
16 |
|
17 |
from moviepy.editor import (ImageClip, VideoFileClip, concatenate_videoclips, TextClip,
|
@@ -30,30 +27,20 @@ logger = logging.getLogger(__name__)
|
|
30 |
logger.setLevel(logging.INFO)
|
31 |
|
32 |
# --- ElevenLabs Client Import ---
|
33 |
-
ELEVENLABS_CLIENT_IMPORTED = False
|
34 |
-
ElevenLabsAPIClient = None
|
35 |
-
Voice = None
|
36 |
-
VoiceSettings = None
|
37 |
try:
|
38 |
from elevenlabs.client import ElevenLabs as ImportedElevenLabsClient
|
39 |
from elevenlabs import Voice as ImportedVoice, VoiceSettings as ImportedVoiceSettings
|
40 |
-
ElevenLabsAPIClient = ImportedElevenLabsClient
|
41 |
-
|
42 |
-
|
43 |
-
ELEVENLABS_CLIENT_IMPORTED = True
|
44 |
-
logger.info("ElevenLabs client components imported.")
|
45 |
-
except Exception as e_eleven:
|
46 |
-
logger.warning(f"ElevenLabs client import failed: {e_eleven}. Audio disabled.")
|
47 |
|
48 |
# --- RunwayML Client Import (Placeholder) ---
|
49 |
-
RUNWAYML_SDK_IMPORTED = False
|
50 |
-
RunwayMLClient = None
|
51 |
try:
|
52 |
logger.info("RunwayML SDK import is a placeholder.")
|
53 |
-
except ImportError:
|
54 |
-
|
55 |
-
except Exception as e_runway_sdk:
|
56 |
-
logger.warning(f"Error importing RunwayML SDK (placeholder): {e_runway_sdk}. RunwayML disabled.")
|
57 |
|
58 |
|
59 |
class VisualEngine:
|
@@ -65,8 +52,7 @@ class VisualEngine:
|
|
65 |
self.font_filename,
|
66 |
f"/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf",
|
67 |
f"/usr/share/fonts/truetype/liberation/LiberationSans-Bold.ttf",
|
68 |
-
f"/System/Library/Fonts/Supplemental/Arial.ttf",
|
69 |
-
f"C:/Windows/Fonts/arial.ttf",
|
70 |
f"/usr/local/share/fonts/truetype/mycustomfonts/arial.ttf"
|
71 |
]
|
72 |
self.font_path_pil = next((p for p in font_paths_to_try if os.path.exists(p)), None)
|
@@ -76,177 +62,93 @@ class VisualEngine:
|
|
76 |
self.video_overlay_font = 'DejaVu-Sans-Bold'
|
77 |
|
78 |
try:
|
79 |
-
if self.font_path_pil
|
80 |
-
|
81 |
-
|
82 |
-
|
83 |
-
self.font = ImageFont.load_default()
|
84 |
-
logger.warning("Using default Pillow font.")
|
85 |
-
self.font_size_pil = 10
|
86 |
-
except IOError as e_font:
|
87 |
-
logger.error(f"Pillow font loading IOError: {e_font}. Using default.")
|
88 |
-
self.font = ImageFont.load_default()
|
89 |
-
self.font_size_pil = 10
|
90 |
|
91 |
-
self.openai_api_key = None
|
92 |
-
self.
|
93 |
-
self.dalle_model = "dall-e-3"
|
94 |
-
self.image_size_dalle3 = "1792x1024"
|
95 |
self.video_frame_size = (1280, 720)
|
96 |
-
self.elevenlabs_api_key = None
|
97 |
-
self.USE_ELEVENLABS = False
|
98 |
-
self.elevenlabs_client = None
|
99 |
self.elevenlabs_voice_id = default_elevenlabs_voice_id
|
100 |
-
if VoiceSettings and ELEVENLABS_CLIENT_IMPORTED:
|
101 |
-
|
102 |
-
|
103 |
-
|
104 |
-
style=0.15,
|
105 |
-
use_speaker_boost=True
|
106 |
-
)
|
107 |
-
else:
|
108 |
-
self.elevenlabs_voice_settings = None
|
109 |
-
self.pexels_api_key = None
|
110 |
-
self.USE_PEXELS = False
|
111 |
-
self.runway_api_key = None
|
112 |
-
self.USE_RUNWAYML = False
|
113 |
-
self.runway_client = None
|
114 |
logger.info("VisualEngine initialized.")
|
115 |
|
116 |
-
def set_openai_api_key(self, k)
|
117 |
-
|
118 |
-
self.
|
119 |
-
|
120 |
-
|
121 |
-
def set_elevenlabs_api_key(self, api_key, voice_id_from_secret=None):
|
122 |
-
self.elevenlabs_api_key = api_key
|
123 |
-
if voice_id_from_secret:
|
124 |
-
self.elevenlabs_voice_id = voice_id_from_secret
|
125 |
if api_key and ELEVENLABS_CLIENT_IMPORTED and ElevenLabsAPIClient:
|
126 |
-
try:
|
127 |
-
|
128 |
-
|
129 |
-
|
130 |
-
except Exception as e:
|
131 |
-
logger.error(f"ElevenLabs client init error: {e}. Disabled.", exc_info=True)
|
132 |
-
self.USE_ELEVENLABS = False
|
133 |
-
else:
|
134 |
-
self.USE_ELEVENLABS = False
|
135 |
-
logger.info("ElevenLabs Disabled (no key or SDK).")
|
136 |
-
|
137 |
-
def set_pexels_api_key(self, k):
|
138 |
-
self.pexels_api_key = k
|
139 |
-
self.USE_PEXELS = bool(k)
|
140 |
-
logger.info(f"Pexels Search {'Ready.' if k else 'Disabled.'}")
|
141 |
-
|
142 |
def set_runway_api_key(self, k):
|
143 |
self.runway_api_key = k
|
144 |
-
if k and RUNWAYML_SDK_IMPORTED and RunwayMLClient:
|
145 |
-
try:
|
146 |
-
|
147 |
-
|
148 |
-
|
149 |
-
logger.error(f"RunwayML client (Placeholder SDK) init error: {e}. Disabled.", exc_info=True)
|
150 |
-
self.USE_RUNWAYML = False
|
151 |
-
elif k:
|
152 |
-
self.USE_RUNWAYML = True
|
153 |
-
logger.info("RunwayML API Key set (direct API or placeholder).")
|
154 |
-
else:
|
155 |
-
self.USE_RUNWAYML = False
|
156 |
-
logger.info("RunwayML Disabled (no API key).")
|
157 |
|
158 |
def _get_text_dimensions(self, text_content, font_obj):
|
159 |
default_line_height = getattr(font_obj, 'size', self.font_size_pil)
|
160 |
-
if not text_content:
|
161 |
-
return 0, default_line_height
|
162 |
try:
|
163 |
if hasattr(font_obj, 'getbbox'):
|
164 |
-
bbox = font_obj.getbbox(text_content)
|
165 |
-
width = bbox[2] - bbox[0]
|
166 |
-
height = bbox[3] - bbox[1]
|
167 |
return width, height if height > 0 else default_line_height
|
168 |
elif hasattr(font_obj, 'getsize'):
|
169 |
width, height = font_obj.getsize(text_content)
|
170 |
return width, height if height > 0 else default_line_height
|
171 |
-
else:
|
172 |
-
|
173 |
-
except Exception as e:
|
174 |
-
logger.warning(f"Error in _get_text_dimensions for '{text_content[:20]}...': {e}")
|
175 |
-
return int(len(text_content) * self.font_size_pil * 0.6), int(self.font_size_pil * 1.2)
|
176 |
|
177 |
def _create_placeholder_image_content(self, text_description, filename, size=None):
|
178 |
-
if size is None:
|
179 |
-
|
180 |
-
|
181 |
-
|
182 |
-
|
183 |
-
max_text_width = size[0] - (2 * padding)
|
184 |
-
lines = []
|
185 |
-
if not text_description:
|
186 |
-
text_description = "(Placeholder: No text description provided)"
|
187 |
-
words = text_description.split()
|
188 |
-
current_line = ""
|
189 |
for word in words:
|
190 |
-
test_line = current_line + word + " "
|
191 |
-
line_width_test
|
192 |
-
if line_width_test <= max_text_width:
|
193 |
-
current_line = test_line
|
194 |
else:
|
195 |
-
if current_line.strip():
|
196 |
-
lines.append(current_line.strip())
|
197 |
word_width, _ = self._get_text_dimensions(word, self.font)
|
198 |
if word_width > max_text_width:
|
199 |
avg_char_w = self._get_text_dimensions("A", self.font)[0] or 10
|
200 |
chars_that_fit = int(max_text_width / avg_char_w) if avg_char_w > 0 else 10
|
201 |
-
if len(word) > chars_that_fit
|
202 |
-
lines.append(word[:chars_that_fit-3] + "...")
|
203 |
-
else:
|
204 |
-
lines.append(word)
|
205 |
current_line = ""
|
206 |
-
else:
|
207 |
-
|
208 |
-
if current_line.strip():
|
209 |
-
lines.append(current_line.strip())
|
210 |
if not lines and text_description:
|
211 |
-
avg_char_w = self._get_text_dimensions("A", self.font)[0] or 10
|
212 |
-
chars_that_fit
|
213 |
-
|
214 |
-
|
215 |
-
|
216 |
-
|
217 |
-
|
218 |
-
|
219 |
-
_, single_line_height = self._get_text_dimensions("Ay", self.font)
|
220 |
-
single_line_height = single_line_height if single_line_height > 0 else (self.font_size_pil + 2)
|
221 |
-
line_spacing = 2
|
222 |
-
max_lines_to_display = min(len(lines), (size[1] - (2 * padding)) // (single_line_height + line_spacing)) if single_line_height > 0 else 1
|
223 |
-
if max_lines_to_display <= 0:
|
224 |
-
max_lines_to_display = 1
|
225 |
-
total_text_block_height = max_lines_to_display * single_line_height + (max_lines_to_display - 1) * line_spacing
|
226 |
-
y_text_start = padding + (size[1] - (2 * padding) - total_text_block_height) / 2.0
|
227 |
-
current_y = y_text_start
|
228 |
for i in range(max_lines_to_display):
|
229 |
-
line_content = lines[i]
|
230 |
-
|
231 |
-
x_text =
|
232 |
-
|
233 |
-
|
234 |
-
|
235 |
-
ellipsis_width, _ = self._get_text_dimensions("...", self.font)
|
236 |
-
x_ellipsis = max(padding, (size[0] - ellipsis_width) / 2.0)
|
237 |
-
draw.text((x_ellipsis, current_y), "...", font=self.font, fill=(200, 200, 180))
|
238 |
-
break
|
239 |
filepath = os.path.join(self.output_dir, filename)
|
240 |
-
try:
|
241 |
-
|
242 |
-
return filepath
|
243 |
-
except Exception as e:
|
244 |
-
logger.error(f"Error saving placeholder image {filepath}: {e}", exc_info=True)
|
245 |
-
return None
|
246 |
|
247 |
def _search_pexels_image(self, query, output_filename_base):
|
248 |
-
if not self.USE_PEXELS or not self.pexels_api_key:
|
249 |
-
return None
|
250 |
headers = {"Authorization": self.pexels_api_key}
|
251 |
params = {"query": query, "per_page": 1, "orientation": "landscape", "size": "large2x"}
|
252 |
base_name, _ = os.path.splitext(output_filename_base)
|
@@ -275,79 +177,53 @@ class VisualEngine:
|
|
275 |
else:
|
276 |
logger.info(f"No photos found on Pexels for query: '{effective_query}'")
|
277 |
return None
|
278 |
-
except requests.exceptions.RequestException as e_req:
|
279 |
-
|
280 |
-
except
|
281 |
-
logger.error(f"Pexels JSON decode error for query '{query}': {e_json}", exc_info=True)
|
282 |
-
except Exception as e:
|
283 |
-
logger.error(f"General Pexels error for query '{query}': {e}", exc_info=True)
|
284 |
return None
|
285 |
|
286 |
def _generate_video_clip_with_runwayml(self, pt, iip, sifnb, tds=5):
|
287 |
-
if not self.USE_RUNWAYML or not self.runway_api_key:
|
288 |
-
|
289 |
-
return None
|
290 |
-
if not iip or not os.path.exists(iip):
|
291 |
-
logger.error(f"Runway Gen-4 needs input image. Path invalid: {iip}")
|
292 |
-
return None
|
293 |
runway_dur = 10 if tds > 7 else 5
|
294 |
-
ovfn = sifnb.replace(".png", f"_runway_gen4_d{runway_dur}s.mp4")
|
295 |
ovfp = os.path.join(self.output_dir, ovfn)
|
296 |
logger.info(f"Runway Gen-4 (Placeholder) img: {os.path.basename(iip)}, motion: '{pt[:100]}...', dur: {runway_dur}s")
|
297 |
logger.warning("Using PLACEHOLDER video for Runway Gen-4.")
|
298 |
-
img_clip = None
|
299 |
-
txt_c = None
|
300 |
-
final_ph_clip = None
|
301 |
try:
|
302 |
img_clip = ImageClip(iip).set_duration(runway_dur)
|
303 |
txt = f"Runway Gen-4 Placeholder\nInput: {os.path.basename(iip)}\nMotion: {pt[:50]}..."
|
304 |
-
txt_c = TextClip(
|
305 |
-
txt,
|
306 |
-
fontsize=24,
|
307 |
-
color='white',
|
308 |
-
font=self.video_overlay_font,
|
309 |
-
bg_color='rgba(0,0,0,0.5)',
|
310 |
-
size=(self.video_frame_size[0] * 0.8, None),
|
311 |
-
method='caption'
|
312 |
-
).set_duration(runway_dur).set_position('center')
|
313 |
final_ph_clip = CompositeVideoClip([img_clip, txt_c], size=img_clip.size)
|
314 |
-
final_ph_clip.write_videofile(ovfp,
|
315 |
-
logger.info(f"Runway Gen-4 placeholder video: {ovfp}")
|
316 |
-
|
317 |
-
except Exception as e:
|
318 |
-
logger.error(f"Runway Gen-4 placeholder error: {e}", exc_info=True)
|
319 |
-
return None
|
320 |
finally:
|
321 |
-
if img_clip and hasattr(img_clip,
|
322 |
-
|
323 |
-
if
|
324 |
-
txt_c.close()
|
325 |
-
if final_ph_clip and hasattr(final_ph_clip, 'close'):
|
326 |
-
final_ph_clip.close()
|
327 |
|
328 |
-
def _create_placeholder_video_content(self, text_description, filename, duration=4, size=None):
|
329 |
if size is None:
|
330 |
size = self.video_frame_size
|
331 |
filepath = os.path.join(self.output_dir, filename)
|
332 |
-
txt_clip = None
|
333 |
try:
|
334 |
-
txt_clip = TextClip(
|
335 |
-
|
336 |
-
|
337 |
-
|
338 |
-
|
339 |
-
|
340 |
-
|
341 |
-
|
342 |
-
|
343 |
-
|
344 |
-
|
345 |
-
|
346 |
-
|
347 |
-
|
348 |
-
logger=None,
|
349 |
-
threads=2
|
350 |
-
)
|
351 |
logger.info(f"Generic placeholder video created successfully: {filepath}")
|
352 |
return filepath
|
353 |
except Exception as e:
|
@@ -364,20 +240,10 @@ class VisualEngine:
|
|
364 |
scene_data, scene_identifier_filename_base,
|
365 |
generate_as_video_clip=False, runway_target_duration=5):
|
366 |
base_name = scene_identifier_filename_base
|
367 |
-
asset_info = {
|
368 |
-
'path': None,
|
369 |
-
'type': 'none',
|
370 |
-
'error': True,
|
371 |
-
'prompt_used': image_generation_prompt_text,
|
372 |
-
'error_message': 'Generation not attempted'
|
373 |
-
}
|
374 |
input_image_for_runway_path = None
|
375 |
image_filename_for_base = base_name + "_base_image.png"
|
376 |
-
temp_image_asset_info = {
|
377 |
-
'error': True,
|
378 |
-
'prompt_used': image_generation_prompt_text,
|
379 |
-
'error_message': 'Base image generation not attempted'
|
380 |
-
}
|
381 |
|
382 |
if self.USE_AI_IMAGE_GENERATION and self.openai_api_key:
|
383 |
max_r, att_n = 2, 0
|
@@ -386,349 +252,151 @@ class VisualEngine:
|
|
386 |
img_fp_dalle = os.path.join(self.output_dir, image_filename_for_base)
|
387 |
logger.info(f"Attempt {att_n+1} DALL-E (base img): {image_generation_prompt_text[:100]}...")
|
388 |
cl = openai.OpenAI(api_key=self.openai_api_key, timeout=90.0)
|
389 |
-
r = cl.images.generate(
|
390 |
-
|
391 |
-
|
392 |
-
|
393 |
-
|
394 |
-
|
395 |
-
|
396 |
-
style="vivid"
|
397 |
-
)
|
398 |
-
iu = r.data[0].url
|
399 |
-
rp = getattr(r.data[0], 'revised_prompt', None)
|
400 |
-
if rp:
|
401 |
-
logger.info(f"DALL-E revised: {rp[:100]}...")
|
402 |
-
ir = requests.get(iu, timeout=120)
|
403 |
-
ir.raise_for_status()
|
404 |
-
id_img = Image.open(io.BytesIO(ir.content))
|
405 |
-
if id_img.mode != 'RGB':
|
406 |
-
id_img = id_img.convert('RGB')
|
407 |
-
id_img.save(img_fp_dalle)
|
408 |
-
logger.info(f"DALL-E base image: {img_fp_dalle}")
|
409 |
input_image_for_runway_path = img_fp_dalle
|
410 |
-
temp_image_asset_info = {
|
411 |
-
'path': img_fp_dalle,
|
412 |
-
'type': 'image',
|
413 |
-
'error': False,
|
414 |
-
'prompt_used': image_generation_prompt_text,
|
415 |
-
'revised_prompt': rp
|
416 |
-
}
|
417 |
break
|
418 |
-
except openai.RateLimitError as e:
|
419 |
-
|
420 |
-
|
421 |
-
|
422 |
-
except Exception as e:
|
423 |
-
logger.error(f"DALL-E error: {e}", exc_info=True)
|
424 |
-
temp_image_asset_info['error_message'] = str(e)
|
425 |
-
break
|
426 |
-
if temp_image_asset_info['error']:
|
427 |
-
logger.warning(f"DALL-E failed after {att_n+1} attempts for base image.")
|
428 |
-
|
429 |
if temp_image_asset_info['error'] and self.USE_PEXELS:
|
430 |
-
pqt = scene_data.get('pexels_search_query_감독',
|
431 |
-
f"{scene_data.get('emotional_beat','')} {scene_data.get('setting_description','')}")
|
432 |
pp = self._search_pexels_image(pqt, image_filename_for_base)
|
433 |
-
if pp:
|
434 |
-
|
435 |
-
temp_image_asset_info = {
|
436 |
-
'path': pp,
|
437 |
-
'type': 'image',
|
438 |
-
'error': False,
|
439 |
-
'prompt_used': f"Pexels: {pqt}"
|
440 |
-
}
|
441 |
-
else:
|
442 |
-
current_em = temp_image_asset_info.get('error_message', "")
|
443 |
-
temp_image_asset_info['error_message'] = (current_em + " Pexels failed.").strip()
|
444 |
|
445 |
if temp_image_asset_info['error']:
|
446 |
logger.warning("Base image (DALL-E/Pexels) failed. Placeholder base image.")
|
447 |
ppt = temp_image_asset_info.get('prompt_used', image_generation_prompt_text)
|
448 |
-
php = self._create_placeholder_image_content(f"[Base Img Placeholder] {ppt[:100]}...", image_filename_for_base)
|
449 |
-
if php:
|
450 |
-
|
451 |
-
|
452 |
-
'path': php,
|
453 |
-
'type': 'image',
|
454 |
-
'error': False,
|
455 |
-
'prompt_used': ppt
|
456 |
-
}
|
457 |
-
else:
|
458 |
-
current_em = temp_image_asset_info.get('error_message', "")
|
459 |
-
temp_image_asset_info['error_message'] = (current_em + " Base placeholder failed.").strip()
|
460 |
-
|
461 |
if generate_as_video_clip:
|
462 |
if self.USE_RUNWAYML and input_image_for_runway_path:
|
463 |
-
video_path = self._generate_video_clip_with_runwayml(
|
464 |
-
motion_prompt_text_for_video,
|
465 |
-
input_image_for_runway_path,
|
466 |
-
base_name,
|
467 |
-
runway_target_duration
|
468 |
-
)
|
469 |
if video_path and os.path.exists(video_path):
|
470 |
-
return {
|
471 |
-
|
472 |
-
|
473 |
-
|
474 |
-
|
475 |
-
'base_image_path': input_image_for_runway_path
|
476 |
-
}
|
477 |
-
else:
|
478 |
-
asset_info = temp_image_asset_info
|
479 |
-
asset_info['error'] = True
|
480 |
-
asset_info['error_message'] = "RunwayML video gen failed; using base image."
|
481 |
-
asset_info['type'] = 'image'
|
482 |
-
return asset_info
|
483 |
-
elif not self.USE_RUNWAYML:
|
484 |
-
asset_info = temp_image_asset_info
|
485 |
-
asset_info['error_message'] = "RunwayML disabled; using base image."
|
486 |
-
asset_info['type'] = 'image'
|
487 |
-
return asset_info
|
488 |
-
else:
|
489 |
-
asset_info = temp_image_asset_info
|
490 |
-
asset_info['error_message'] = (asset_info.get('error_message', "") + " Base image failed, Runway video not attempted.").strip()
|
491 |
-
asset_info['type'] = 'image'
|
492 |
-
return asset_info
|
493 |
-
else:
|
494 |
-
return temp_image_asset_info
|
495 |
|
496 |
def generate_narration_audio(self, ttn, ofn="narration_overall.mp3"):
|
497 |
-
if not self.USE_ELEVENLABS or not self.elevenlabs_client or not ttn:
|
498 |
-
|
499 |
-
|
500 |
-
|
501 |
-
|
502 |
-
|
503 |
-
|
504 |
-
|
505 |
-
|
506 |
-
|
507 |
-
elif hasattr(self.
|
508 |
-
|
509 |
-
|
510 |
-
|
511 |
-
|
512 |
-
|
513 |
-
|
514 |
-
|
515 |
-
with open(afp, "wb") as f:
|
516 |
-
f.write(ab)
|
517 |
-
logger.info(f"11L audio (non-stream): {afp}")
|
518 |
-
return afp
|
519 |
-
else:
|
520 |
-
logger.error("No 11L audio method.")
|
521 |
-
return None
|
522 |
-
|
523 |
-
if asm:
|
524 |
-
vps = {"voice_id": str(self.elevenlabs_voice_id)}
|
525 |
-
if self.elevenlabs_voice_settings:
|
526 |
-
if hasattr(self.elevenlabs_voice_settings, 'model_dump'):
|
527 |
-
vps["voice_settings"] = self.elevenlabs_voice_settings.model_dump()
|
528 |
-
elif hasattr(self.elevenlabs_voice_settings, 'dict'):
|
529 |
-
vps["voice_settings"] = self.elevenlabs_voice_settings.dict()
|
530 |
-
else:
|
531 |
-
vps["voice_settings"] = self.elevenlabs_voice_settings
|
532 |
-
adi = asm(text=ttn, model_id="eleven_multilingual_v2", **vps)
|
533 |
-
with open(afp, "wb") as f:
|
534 |
-
for c in adi:
|
535 |
-
if c:
|
536 |
-
f.write(c)
|
537 |
-
logger.info(f"11L audio (stream): {afp}")
|
538 |
-
return afp
|
539 |
-
except Exception as e:
|
540 |
-
logger.error(f"11L audio error: {e}", exc_info=True)
|
541 |
-
return None
|
542 |
|
543 |
def assemble_animatic_from_assets(self, asset_data_list, overall_narration_path=None, output_filename="final_video.mp4", fps=24):
|
544 |
-
if not asset_data_list:
|
545 |
-
|
546 |
-
return None
|
547 |
-
processed_clips = []
|
548 |
-
narration_clip = None
|
549 |
-
final_clip = None
|
550 |
logger.info(f"Assembling from {len(asset_data_list)} assets. Frame: {self.video_frame_size}.")
|
551 |
|
552 |
for i, asset_info in enumerate(asset_data_list):
|
553 |
-
asset_path = asset_info.get('path')
|
554 |
-
|
555 |
-
scene_dur = asset_info.get('duration', 4.5)
|
556 |
-
scene_num = asset_info.get('scene_num', i + 1)
|
557 |
-
key_action = asset_info.get('key_action', '')
|
558 |
logger.info(f"S{scene_num}: Path='{asset_path}', Type='{asset_type}', Dur='{scene_dur}'s")
|
559 |
|
560 |
-
if not (asset_path and os.path.exists(asset_path)):
|
561 |
-
|
562 |
-
continue
|
563 |
-
if scene_dur <= 0:
|
564 |
-
logger.warning(f"S{scene_num}: Invalid duration ({scene_dur}s). Skip.")
|
565 |
-
continue
|
566 |
|
567 |
current_scene_mvpy_clip = None
|
568 |
try:
|
569 |
if asset_type == 'image':
|
570 |
-
pil_img = Image.open(asset_path)
|
571 |
-
logger.debug(f"S{scene_num}: Loaded img. Mode:{pil_img.mode}, Size:{pil_img.size}")
|
572 |
img_rgba = pil_img.convert('RGBA') if pil_img.mode != 'RGBA' else pil_img.copy()
|
573 |
-
thumb = img_rgba.copy()
|
574 |
-
|
575 |
-
|
576 |
-
|
577 |
-
|
578 |
-
|
579 |
-
|
580 |
-
final_rgb_pil = Image.new("RGB", self.video_frame_size, (0, 0, 0))
|
581 |
-
final_rgb_pil.paste(cv_rgba, mask=cv_rgba.split()[3])
|
582 |
-
dbg_path = os.path.join(self.output_dir, f"debug_PRE_NUMPY_S{scene_num}.png")
|
583 |
-
final_rgb_pil.save(dbg_path)
|
584 |
-
logger.info(f"DEBUG: Saved PRE_NUMPY_S{scene_num} to {dbg_path}")
|
585 |
-
frame_np = np.array(final_rgb_pil, dtype=np.uint8)
|
586 |
-
if not frame_np.flags['C_CONTIGUOUS']:
|
587 |
-
frame_np = np.ascontiguousarray(frame_np, dtype=np.uint8)
|
588 |
logger.debug(f"S{scene_num}: NumPy for MoviePy. Shape:{frame_np.shape}, DType:{frame_np.dtype}, C-Contig:{frame_np.flags['C_CONTIGUOUS']}")
|
589 |
-
if frame_np.size
|
590 |
-
|
591 |
-
|
592 |
-
clip_base = ImageClip(frame_np, transparent=False).set_duration(scene_dur)
|
593 |
-
mvpy_dbg_path = os.path.join(self.output_dir, f"debug_MOVIEPY_FRAME_S{scene_num}.png")
|
594 |
-
clip_base.save_frame(mvpy_dbg_path, t=0.1)
|
595 |
-
logger.info(f"DEBUG: Saved MOVIEPY_FRAME_S{scene_num} to {mvpy_dbg_path}")
|
596 |
clip_fx = clip_base
|
597 |
-
try:
|
598 |
-
|
599 |
-
clip_fx = clip_base.fx(
|
600 |
-
vfx.resize,
|
601 |
-
lambda t: 1 + (es - 1) * (t / scene_dur) if scene_dur > 0 else 1
|
602 |
-
).set_position('center')
|
603 |
-
except Exception as e:
|
604 |
-
logger.error(f"S{scene_num} Ken Burns error: {e}", exc_info=False)
|
605 |
current_scene_mvpy_clip = clip_fx
|
606 |
elif asset_type == 'video':
|
607 |
-
src_clip
|
608 |
try:
|
609 |
-
src_clip
|
610 |
-
|
611 |
-
|
612 |
-
|
613 |
-
)
|
614 |
-
tmp_clip = src_clip
|
615 |
-
if src_clip.duration != scene_dur:
|
616 |
-
if src_clip.duration > scene_dur:
|
617 |
-
tmp_clip = src_clip.subclip(0, scene_dur)
|
618 |
else:
|
619 |
-
if scene_dur
|
620 |
-
|
621 |
-
|
622 |
-
|
623 |
-
|
624 |
-
current_scene_mvpy_clip = tmp_clip.set_duration(scene_dur)
|
625 |
-
if current_scene_mvpy_clip.size != list(self.video_frame_size):
|
626 |
-
current_scene_mvpy_clip = current_scene_mvpy_clip.resize(self.video_frame_size)
|
627 |
-
except Exception as e:
|
628 |
-
logger.error(f"S{scene_num} Video load error '{asset_path}':{e}", exc_info=True)
|
629 |
-
continue
|
630 |
finally:
|
631 |
-
if src_clip and src_clip is not current_scene_mvpy_clip and hasattr(src_clip,
|
632 |
-
|
633 |
-
else:
|
634 |
-
logger.warning(f"S{scene_num} Unknown asset type '{asset_type}'. Skip.")
|
635 |
-
continue
|
636 |
-
|
637 |
if current_scene_mvpy_clip and key_action:
|
638 |
try:
|
639 |
-
to_dur
|
640 |
-
to_start
|
641 |
-
txt_c
|
642 |
-
|
643 |
-
|
644 |
-
|
645 |
-
|
646 |
-
bg_color='rgba(10,10,20,0.7)',
|
647 |
-
method='caption',
|
648 |
-
align='West',
|
649 |
-
size=(self.video_frame_size[0] * 0.9, None),
|
650 |
-
kerning=-1,
|
651 |
-
stroke_color='black',
|
652 |
-
stroke_width=1.5
|
653 |
-
).set_duration(to_dur).set_start(to_start).set_position(('center', 0.92), relative=True)
|
654 |
-
current_scene_mvpy_clip = CompositeVideoClip(
|
655 |
-
[current_scene_mvpy_clip, txt_c],
|
656 |
-
size=self.video_frame_size,
|
657 |
-
use_bgclip=True
|
658 |
-
)
|
659 |
-
except Exception as e:
|
660 |
-
logger.error(f"S{scene_num} TextClip error:{e}. No text.", exc_info=True)
|
661 |
-
|
662 |
-
if current_scene_mvpy_clip:
|
663 |
-
processed_clips.append(current_scene_mvpy_clip)
|
664 |
-
logger.info(f"S{scene_num} Processed. Dur:{current_scene_mvpy_clip.duration:.2f}s.")
|
665 |
-
except Exception as e:
|
666 |
-
logger.error(f"MAJOR Error S{scene_num} ({asset_path}):{e}", exc_info=True)
|
667 |
finally:
|
668 |
-
if current_scene_mvpy_clip and hasattr(current_scene_mvpy_clip,
|
669 |
-
try:
|
670 |
-
|
671 |
-
except:
|
672 |
-
pass
|
673 |
|
674 |
-
if not processed_clips:
|
675 |
-
|
676 |
-
return None
|
677 |
-
td = 0.75
|
678 |
try:
|
679 |
-
logger.info(f"Concatenating {len(processed_clips)} clips.")
|
680 |
-
if len(processed_clips) >
|
681 |
-
|
682 |
-
|
683 |
-
final_clip = processed_clips[0]
|
684 |
-
if not final_clip:
|
685 |
-
logger.error("Concatenation failed.")
|
686 |
-
return None
|
687 |
logger.info(f"Concatenated dur:{final_clip.duration:.2f}s")
|
688 |
-
if td
|
689 |
-
if final_clip.duration
|
690 |
-
|
691 |
-
|
692 |
-
|
693 |
-
|
694 |
-
|
695 |
-
|
696 |
-
|
697 |
-
|
698 |
-
|
699 |
-
|
700 |
-
|
701 |
-
logger.warning("Video no duration. No audio.")
|
702 |
-
if final_clip and final_clip.duration > 0:
|
703 |
-
op = os.path.join(self.output_dir, output_filename)
|
704 |
-
logger.info(f"Writing video:{op} (Dur:{final_clip.duration:.2f}s)")
|
705 |
-
final_clip.write_videofile(
|
706 |
-
op,
|
707 |
-
fps=fps,
|
708 |
-
codec='libx264',
|
709 |
-
preset='medium',
|
710 |
-
audio_codec='aac',
|
711 |
-
temp_audiofile=os.path.join(self.output_dir, f'temp-audio-{os.urandom(4).hex()}.m4a'),
|
712 |
-
remove_temp=True,
|
713 |
-
threads=os.cpu_count() or 2,
|
714 |
-
logger='bar',
|
715 |
-
bitrate="5000k",
|
716 |
-
ffmpeg_params=["-pix_fmt", "yuv420p"]
|
717 |
-
)
|
718 |
-
logger.info(f"Video created:{op}")
|
719 |
-
return op
|
720 |
-
else:
|
721 |
-
logger.error("Final clip invalid. No write.")
|
722 |
-
return None
|
723 |
-
except Exception as e:
|
724 |
-
logger.error(f"Video write error:{e}", exc_info=True)
|
725 |
-
return None
|
726 |
finally:
|
727 |
logger.debug("Closing all MoviePy clips in `assemble_animatic_from_assets` finally block.")
|
728 |
clips_to_close = processed_clips + ([narration_clip] if narration_clip else []) + ([final_clip] if final_clip else [])
|
729 |
for clip_obj in clips_to_close:
|
730 |
if clip_obj and hasattr(clip_obj, 'close'):
|
731 |
-
try:
|
732 |
-
|
733 |
-
except Exception as e_close:
|
734 |
-
logger.warning(f"Ignoring error while closing a clip: {e_close}")
|
|
|
2 |
from PIL import Image, ImageDraw, ImageFont, ImageOps
|
3 |
# --- MONKEY PATCH FOR Image.ANTIALIAS ---
|
4 |
try:
|
5 |
+
if hasattr(Image, 'Resampling') and hasattr(Image.Resampling, 'LANCZOS'): # Pillow 9+
|
6 |
+
if not hasattr(Image, 'ANTIALIAS'): Image.ANTIALIAS = Image.Resampling.LANCZOS
|
7 |
+
elif hasattr(Image, 'LANCZOS'): # Pillow 8
|
8 |
+
if not hasattr(Image, 'ANTIALIAS'): Image.ANTIALIAS = Image.LANCZOS
|
|
|
|
|
9 |
elif not hasattr(Image, 'ANTIALIAS'):
|
10 |
+
print("WARNING: Pillow version lacks common Resampling attributes or ANTIALIAS. Video effects might fail.")
|
11 |
+
except Exception as e_mp: print(f"WARNING: ANTIALIAS monkey-patch error: {e_mp}")
|
|
|
12 |
# --- END MONKEY PATCH ---
|
13 |
|
14 |
from moviepy.editor import (ImageClip, VideoFileClip, concatenate_videoclips, TextClip,
|
|
|
27 |
logger.setLevel(logging.INFO)
|
28 |
|
29 |
# --- ElevenLabs Client Import ---
|
30 |
+
ELEVENLABS_CLIENT_IMPORTED = False; ElevenLabsAPIClient = None; Voice = None; VoiceSettings = None
|
|
|
|
|
|
|
31 |
try:
|
32 |
from elevenlabs.client import ElevenLabs as ImportedElevenLabsClient
|
33 |
from elevenlabs import Voice as ImportedVoice, VoiceSettings as ImportedVoiceSettings
|
34 |
+
ElevenLabsAPIClient = ImportedElevenLabsClient; Voice = ImportedVoice; VoiceSettings = ImportedVoiceSettings
|
35 |
+
ELEVENLABS_CLIENT_IMPORTED = True; logger.info("ElevenLabs client components imported.")
|
36 |
+
except Exception as e_eleven: logger.warning(f"ElevenLabs client import failed: {e_eleven}. Audio disabled.")
|
|
|
|
|
|
|
|
|
37 |
|
38 |
# --- RunwayML Client Import (Placeholder) ---
|
39 |
+
RUNWAYML_SDK_IMPORTED = False; RunwayMLClient = None
|
|
|
40 |
try:
|
41 |
logger.info("RunwayML SDK import is a placeholder.")
|
42 |
+
except ImportError: logger.warning("RunwayML SDK (placeholder) not found. RunwayML disabled.")
|
43 |
+
except Exception as e_runway_sdk: logger.warning(f"Error importing RunwayML SDK (placeholder): {e_runway_sdk}. RunwayML disabled.")
|
|
|
|
|
44 |
|
45 |
|
46 |
class VisualEngine:
|
|
|
52 |
self.font_filename,
|
53 |
f"/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf",
|
54 |
f"/usr/share/fonts/truetype/liberation/LiberationSans-Bold.ttf",
|
55 |
+
f"/System/Library/Fonts/Supplemental/Arial.ttf", f"C:/Windows/Fonts/arial.ttf",
|
|
|
56 |
f"/usr/local/share/fonts/truetype/mycustomfonts/arial.ttf"
|
57 |
]
|
58 |
self.font_path_pil = next((p for p in font_paths_to_try if os.path.exists(p)), None)
|
|
|
62 |
self.video_overlay_font = 'DejaVu-Sans-Bold'
|
63 |
|
64 |
try:
|
65 |
+
self.font = ImageFont.truetype(self.font_path_pil, self.font_size_pil) if self.font_path_pil else ImageFont.load_default()
|
66 |
+
if self.font_path_pil: logger.info(f"Pillow font loaded: {self.font_path_pil}.")
|
67 |
+
else: logger.warning("Using default Pillow font."); self.font_size_pil = 10
|
68 |
+
except IOError as e_font: logger.error(f"Pillow font loading IOError: {e_font}. Using default."); self.font = ImageFont.load_default(); self.font_size_pil = 10
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
69 |
|
70 |
+
self.openai_api_key = None; self.USE_AI_IMAGE_GENERATION = False
|
71 |
+
self.dalle_model = "dall-e-3"; self.image_size_dalle3 = "1792x1024"
|
|
|
|
|
72 |
self.video_frame_size = (1280, 720)
|
73 |
+
self.elevenlabs_api_key = None; self.USE_ELEVENLABS = False; self.elevenlabs_client = None
|
|
|
|
|
74 |
self.elevenlabs_voice_id = default_elevenlabs_voice_id
|
75 |
+
if VoiceSettings and ELEVENLABS_CLIENT_IMPORTED: self.elevenlabs_voice_settings = VoiceSettings(stability=0.60, similarity_boost=0.80, style=0.15, use_speaker_boost=True)
|
76 |
+
else: self.elevenlabs_voice_settings = None
|
77 |
+
self.pexels_api_key = None; self.USE_PEXELS = False
|
78 |
+
self.runway_api_key = None; self.USE_RUNWAYML = False; self.runway_client = None
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
79 |
logger.info("VisualEngine initialized.")
|
80 |
|
81 |
+
def set_openai_api_key(self,k): self.openai_api_key=k; self.USE_AI_IMAGE_GENERATION=bool(k); logger.info(f"DALL-E ({self.dalle_model}) {'Ready.' if k else 'Disabled.'}")
|
82 |
+
def set_elevenlabs_api_key(self,api_key, voice_id_from_secret=None):
|
83 |
+
self.elevenlabs_api_key=api_key
|
84 |
+
if voice_id_from_secret: self.elevenlabs_voice_id = voice_id_from_secret
|
|
|
|
|
|
|
|
|
|
|
85 |
if api_key and ELEVENLABS_CLIENT_IMPORTED and ElevenLabsAPIClient:
|
86 |
+
try: self.elevenlabs_client = ElevenLabsAPIClient(api_key=api_key); self.USE_ELEVENLABS=bool(self.elevenlabs_client); logger.info(f"ElevenLabs Client {'Ready' if self.USE_ELEVENLABS else 'Failed Init'} (Voice ID: {self.elevenlabs_voice_id}).")
|
87 |
+
except Exception as e: logger.error(f"ElevenLabs client init error: {e}. Disabled.", exc_info=True); self.USE_ELEVENLABS=False
|
88 |
+
else: self.USE_ELEVENLABS=False; logger.info("ElevenLabs Disabled (no key or SDK).")
|
89 |
+
def set_pexels_api_key(self,k): self.pexels_api_key=k; self.USE_PEXELS=bool(k); logger.info(f"Pexels Search {'Ready.' if k else 'Disabled.'}")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
90 |
def set_runway_api_key(self, k):
|
91 |
self.runway_api_key = k
|
92 |
+
if k and RUNWAYML_SDK_IMPORTED and RunwayMLClient: # This SDK part is still hypothetical
|
93 |
+
try: self.USE_RUNWAYML = True; logger.info(f"RunwayML Client (Placeholder SDK) {'Ready.' if self.USE_RUNWAYML else 'Failed Init.'}")
|
94 |
+
except Exception as e: logger.error(f"RunwayML client (Placeholder SDK) init error: {e}. Disabled.", exc_info=True); self.USE_RUNWAYML = False
|
95 |
+
elif k: self.USE_RUNWAYML = True; logger.info("RunwayML API Key set (direct API or placeholder).")
|
96 |
+
else: self.USE_RUNWAYML = False; logger.info("RunwayML Disabled (no API key).")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
97 |
|
98 |
def _get_text_dimensions(self, text_content, font_obj):
|
99 |
default_line_height = getattr(font_obj, 'size', self.font_size_pil)
|
100 |
+
if not text_content: return 0, default_line_height
|
|
|
101 |
try:
|
102 |
if hasattr(font_obj, 'getbbox'):
|
103 |
+
bbox = font_obj.getbbox(text_content); width = bbox[2] - bbox[0]; height = bbox[3] - bbox[1]
|
|
|
|
|
104 |
return width, height if height > 0 else default_line_height
|
105 |
elif hasattr(font_obj, 'getsize'):
|
106 |
width, height = font_obj.getsize(text_content)
|
107 |
return width, height if height > 0 else default_line_height
|
108 |
+
else: return int(len(text_content) * default_line_height * 0.6), int(default_line_height * 1.2)
|
109 |
+
except Exception as e: logger.warning(f"Error in _get_text_dimensions for '{text_content[:20]}...': {e}"); return int(len(text_content) * self.font_size_pil * 0.6),int(self.font_size_pil * 1.2)
|
|
|
|
|
|
|
110 |
|
111 |
def _create_placeholder_image_content(self, text_description, filename, size=None):
|
112 |
+
if size is None: size = self.video_frame_size
|
113 |
+
img = Image.new('RGB', size, color=(20, 20, 40)); draw = ImageDraw.Draw(img)
|
114 |
+
padding = 25; max_text_width = size[0] - (2 * padding); lines = []
|
115 |
+
if not text_description: text_description = "(Placeholder: No text description provided)"
|
116 |
+
words = text_description.split(); current_line = ""
|
|
|
|
|
|
|
|
|
|
|
|
|
117 |
for word in words:
|
118 |
+
test_line = current_line + word + " "; line_width_test, _ = self._get_text_dimensions(test_line.strip(), self.font)
|
119 |
+
if line_width_test <= max_text_width: current_line = test_line
|
|
|
|
|
120 |
else:
|
121 |
+
if current_line.strip(): lines.append(current_line.strip())
|
|
|
122 |
word_width, _ = self._get_text_dimensions(word, self.font)
|
123 |
if word_width > max_text_width:
|
124 |
avg_char_w = self._get_text_dimensions("A", self.font)[0] or 10
|
125 |
chars_that_fit = int(max_text_width / avg_char_w) if avg_char_w > 0 else 10
|
126 |
+
lines.append(word[:chars_that_fit-3] + "..." if len(word) > chars_that_fit else word)
|
|
|
|
|
|
|
127 |
current_line = ""
|
128 |
+
else: current_line = word + " "
|
129 |
+
if current_line.strip(): lines.append(current_line.strip())
|
|
|
|
|
130 |
if not lines and text_description:
|
131 |
+
avg_char_w = self._get_text_dimensions("A", self.font)[0] or 10; chars_that_fit = int(max_text_width / avg_char_w) if avg_char_w > 0 else 10
|
132 |
+
lines.append(text_description[:chars_that_fit-3] + "..." if len(text_description) > chars_that_fit else text_description)
|
133 |
+
elif not lines: lines.append("(Placeholder Text Error)")
|
134 |
+
_, single_line_height = self._get_text_dimensions("Ay", self.font); single_line_height = single_line_height if single_line_height > 0 else (self.font_size_pil + 2)
|
135 |
+
line_spacing = 2; max_lines_to_display = min(len(lines), (size[1]-(2*padding))//(single_line_height+line_spacing)) if single_line_height > 0 else 1
|
136 |
+
if max_lines_to_display <= 0: max_lines_to_display = 1
|
137 |
+
total_text_block_height = max_lines_to_display * single_line_height + (max_lines_to_display-1)*line_spacing
|
138 |
+
y_text_start = padding + (size[1]-(2*padding)-total_text_block_height)/2.0; current_y = y_text_start
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
139 |
for i in range(max_lines_to_display):
|
140 |
+
line_content = lines[i]; line_width_actual, _ = self._get_text_dimensions(line_content, self.font)
|
141 |
+
x_text = max(padding, (size[0]-line_width_actual)/2.0)
|
142 |
+
draw.text((x_text, current_y), line_content, font=self.font, fill=(200,200,180)); current_y += single_line_height + line_spacing
|
143 |
+
if i==6 and max_lines_to_display > 7 and len(lines) > max_lines_to_display:
|
144 |
+
ellipsis_width, _ = self._get_text_dimensions("...",self.font); x_ellipsis = max(padding, (size[0]-ellipsis_width)/2.0)
|
145 |
+
draw.text((x_ellipsis, current_y), "...", font=self.font, fill=(200,200,180)); break
|
|
|
|
|
|
|
|
|
146 |
filepath = os.path.join(self.output_dir, filename)
|
147 |
+
try: img.save(filepath); return filepath
|
148 |
+
except Exception as e: logger.error(f"Error saving placeholder image {filepath}: {e}", exc_info=True); return None
|
|
|
|
|
|
|
|
|
149 |
|
150 |
def _search_pexels_image(self, query, output_filename_base):
|
151 |
+
if not self.USE_PEXELS or not self.pexels_api_key: return None
|
|
|
152 |
headers = {"Authorization": self.pexels_api_key}
|
153 |
params = {"query": query, "per_page": 1, "orientation": "landscape", "size": "large2x"}
|
154 |
base_name, _ = os.path.splitext(output_filename_base)
|
|
|
177 |
else:
|
178 |
logger.info(f"No photos found on Pexels for query: '{effective_query}'")
|
179 |
return None
|
180 |
+
except requests.exceptions.RequestException as e_req: logger.error(f"Pexels request error for query '{query}': {e_req}", exc_info=True)
|
181 |
+
except json.JSONDecodeError as e_json: logger.error(f"Pexels JSON decode error for query '{query}': {e_json}", exc_info=True)
|
182 |
+
except Exception as e: logger.error(f"General Pexels error for query '{query}': {e}", exc_info=True)
|
|
|
|
|
|
|
183 |
return None
|
184 |
|
185 |
def _generate_video_clip_with_runwayml(self, pt, iip, sifnb, tds=5):
|
186 |
+
if not self.USE_RUNWAYML or not self.runway_api_key: logger.warning("RunwayML disabled."); return None
|
187 |
+
if not iip or not os.path.exists(iip): logger.error(f"Runway Gen-4 needs input image. Path invalid: {iip}"); return None
|
|
|
|
|
|
|
|
|
188 |
runway_dur = 10 if tds > 7 else 5
|
189 |
+
ovfn = sifnb.replace(".png", f"_runway_gen4_d{runway_dur}s.mp4") # sifnb should be base name
|
190 |
ovfp = os.path.join(self.output_dir, ovfn)
|
191 |
logger.info(f"Runway Gen-4 (Placeholder) img: {os.path.basename(iip)}, motion: '{pt[:100]}...', dur: {runway_dur}s")
|
192 |
logger.warning("Using PLACEHOLDER video for Runway Gen-4.")
|
193 |
+
img_clip=None; txt_c=None; final_ph_clip=None
|
|
|
|
|
194 |
try:
|
195 |
img_clip = ImageClip(iip).set_duration(runway_dur)
|
196 |
txt = f"Runway Gen-4 Placeholder\nInput: {os.path.basename(iip)}\nMotion: {pt[:50]}..."
|
197 |
+
txt_c = TextClip(txt, fontsize=24,color='white',font=self.video_overlay_font,bg_color='rgba(0,0,0,0.5)',size=(self.video_frame_size[0]*0.8,None),method='caption').set_duration(runway_dur).set_position('center')
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
198 |
final_ph_clip = CompositeVideoClip([img_clip, txt_c], size=img_clip.size)
|
199 |
+
final_ph_clip.write_videofile(ovfp,fps=24,codec='libx264',preset='ultrafast',logger=None,threads=2)
|
200 |
+
logger.info(f"Runway Gen-4 placeholder video: {ovfp}"); return ovfp
|
201 |
+
except Exception as e: logger.error(f"Runway Gen-4 placeholder error: {e}",exc_info=True); return None
|
|
|
|
|
|
|
202 |
finally:
|
203 |
+
if img_clip and hasattr(img_clip,'close'): img_clip.close()
|
204 |
+
if txt_c and hasattr(txt_c,'close'): txt_c.close()
|
205 |
+
if final_ph_clip and hasattr(final_ph_clip,'close'): final_ph_clip.close()
|
|
|
|
|
|
|
206 |
|
207 |
+
def _create_placeholder_video_content(self, text_description, filename, duration=4, size=None): # Generic placeholder
|
208 |
if size is None:
|
209 |
size = self.video_frame_size
|
210 |
filepath = os.path.join(self.output_dir, filename)
|
211 |
+
txt_clip = None # Initialize for finally block
|
212 |
try:
|
213 |
+
txt_clip = TextClip(text_description,
|
214 |
+
fontsize=50,
|
215 |
+
color='white',
|
216 |
+
font=self.video_overlay_font,
|
217 |
+
bg_color='black',
|
218 |
+
size=size,
|
219 |
+
method='caption').set_duration(duration)
|
220 |
+
|
221 |
+
txt_clip.write_videofile(filepath,
|
222 |
+
fps=24,
|
223 |
+
codec='libx264',
|
224 |
+
preset='ultrafast',
|
225 |
+
logger=None,
|
226 |
+
threads=2)
|
|
|
|
|
|
|
227 |
logger.info(f"Generic placeholder video created successfully: {filepath}")
|
228 |
return filepath
|
229 |
except Exception as e:
|
|
|
240 |
scene_data, scene_identifier_filename_base,
|
241 |
generate_as_video_clip=False, runway_target_duration=5):
|
242 |
base_name = scene_identifier_filename_base
|
243 |
+
asset_info = {'path': None, 'type': 'none', 'error': True, 'prompt_used': image_generation_prompt_text, 'error_message': 'Generation not attempted'}
|
|
|
|
|
|
|
|
|
|
|
|
|
244 |
input_image_for_runway_path = None
|
245 |
image_filename_for_base = base_name + "_base_image.png"
|
246 |
+
temp_image_asset_info = {'error': True, 'prompt_used': image_generation_prompt_text, 'error_message': 'Base image generation not attempted'}
|
|
|
|
|
|
|
|
|
247 |
|
248 |
if self.USE_AI_IMAGE_GENERATION and self.openai_api_key:
|
249 |
max_r, att_n = 2, 0
|
|
|
252 |
img_fp_dalle = os.path.join(self.output_dir, image_filename_for_base)
|
253 |
logger.info(f"Attempt {att_n+1} DALL-E (base img): {image_generation_prompt_text[:100]}...")
|
254 |
cl = openai.OpenAI(api_key=self.openai_api_key, timeout=90.0)
|
255 |
+
r = cl.images.generate(model=self.dalle_model, prompt=image_generation_prompt_text, n=1, size=self.image_size_dalle3, quality="hd", response_format="url", style="vivid")
|
256 |
+
iu = r.data[0].url; rp = getattr(r.data[0], 'revised_prompt', None)
|
257 |
+
if rp: logger.info(f"DALL-E revised: {rp[:100]}...")
|
258 |
+
ir = requests.get(iu, timeout=120); ir.raise_for_status()
|
259 |
+
id_img = Image.open(io.BytesIO(ir.content));
|
260 |
+
if id_img.mode != 'RGB': id_img = id_img.convert('RGB')
|
261 |
+
id_img.save(img_fp_dalle); logger.info(f"DALL-E base image: {img_fp_dalle}");
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
262 |
input_image_for_runway_path = img_fp_dalle
|
263 |
+
temp_image_asset_info = {'path': img_fp_dalle, 'type': 'image', 'error': False, 'prompt_used': image_generation_prompt_text, 'revised_prompt': rp}
|
|
|
|
|
|
|
|
|
|
|
|
|
264 |
break
|
265 |
+
except openai.RateLimitError as e: logger.warning(f"OpenAI Rate Limit {att_n+1}: {e}. Retry..."); time.sleep(5*(att_n+1)); temp_image_asset_info['error_message']=str(e)
|
266 |
+
except Exception as e: logger.error(f"DALL-E error: {e}", exc_info=True); temp_image_asset_info['error_message']=str(e); break
|
267 |
+
if temp_image_asset_info['error']: logger.warning(f"DALL-E failed after {att_n+1} attempts for base image.")
|
268 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
269 |
if temp_image_asset_info['error'] and self.USE_PEXELS:
|
270 |
+
pqt = scene_data.get('pexels_search_query_감독', f"{scene_data.get('emotional_beat','')} {scene_data.get('setting_description','')}")
|
|
|
271 |
pp = self._search_pexels_image(pqt, image_filename_for_base)
|
272 |
+
if pp: input_image_for_runway_path = pp; temp_image_asset_info = {'path': pp, 'type': 'image', 'error': False, 'prompt_used': f"Pexels: {pqt}"}
|
273 |
+
else: current_em = temp_image_asset_info.get('error_message',""); temp_image_asset_info['error_message']=(current_em + " Pexels failed.").strip()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
274 |
|
275 |
if temp_image_asset_info['error']:
|
276 |
logger.warning("Base image (DALL-E/Pexels) failed. Placeholder base image.")
|
277 |
ppt = temp_image_asset_info.get('prompt_used', image_generation_prompt_text)
|
278 |
+
php = self._create_placeholder_image_content(f"[Base Img Placeholder] {ppt[:100]}...", image_filename_for_base) # Use image_filename_for_base
|
279 |
+
if php: input_image_for_runway_path = php; temp_image_asset_info = {'path': php, 'type': 'image', 'error': False, 'prompt_used': ppt}
|
280 |
+
else: current_em=temp_image_asset_info.get('error_message',"");temp_image_asset_info['error_message']=(current_em + " Base placeholder failed.").strip()
|
281 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
282 |
if generate_as_video_clip:
|
283 |
if self.USE_RUNWAYML and input_image_for_runway_path:
|
284 |
+
video_path = self._generate_video_clip_with_runwayml(motion_prompt_text_for_video, input_image_for_runway_path, base_name, runway_target_duration)
|
|
|
|
|
|
|
|
|
|
|
285 |
if video_path and os.path.exists(video_path):
|
286 |
+
return {'path': video_path, 'type': 'video', 'error': False, 'prompt_used': motion_prompt_text_for_video, 'base_image_path': input_image_for_runway_path}
|
287 |
+
else: asset_info = temp_image_asset_info; asset_info['error'] = True; asset_info['error_message'] = "RunwayML video gen failed; using base image."; asset_info['type'] = 'image'; return asset_info
|
288 |
+
elif not self.USE_RUNWAYML: asset_info = temp_image_asset_info; asset_info['error_message'] = "RunwayML disabled; using base image."; asset_info['type'] = 'image'; return asset_info
|
289 |
+
else: asset_info = temp_image_asset_info; asset_info['error_message'] = (asset_info.get('error_message',"") + " Base image failed, Runway video not attempted.").strip(); asset_info['type'] = 'image'; return asset_info
|
290 |
+
else: return temp_image_asset_info
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
291 |
|
292 |
def generate_narration_audio(self, ttn, ofn="narration_overall.mp3"):
|
293 |
+
if not self.USE_ELEVENLABS or not self.elevenlabs_client or not ttn: logger.info("11L skip."); return None; afp=os.path.join(self.output_dir,ofn)
|
294 |
+
try: logger.info(f"11L audio (Voice:{self.elevenlabs_voice_id}): {ttn[:70]}..."); asm=None
|
295 |
+
if hasattr(self.elevenlabs_client,'text_to_speech')and hasattr(self.elevenlabs_client.text_to_speech,'stream'):asm=self.elevenlabs_client.text_to_speech.stream;logger.info("Using 11L .text_to_speech.stream()")
|
296 |
+
elif hasattr(self.elevenlabs_client,'generate_stream'):asm=self.elevenlabs_client.generate_stream;logger.info("Using 11L .generate_stream()")
|
297 |
+
elif hasattr(self.elevenlabs_client,'generate'):logger.info("Using 11L .generate()");vp=Voice(voice_id=str(self.elevenlabs_voice_id),settings=self.elevenlabs_voice_settings)if Voice and self.elevenlabs_voice_settings else str(self.elevenlabs_voice_id);ab=self.elevenlabs_client.generate(text=ttn,voice=vp,model="eleven_multilingual_v2");
|
298 |
+
with open(afp,"wb")as f:f.write(ab);logger.info(f"11L audio (non-stream): {afp}");return afp
|
299 |
+
else:logger.error("No 11L audio method.");return None
|
300 |
+
if asm:vps={"voice_id":str(self.elevenlabs_voice_id)}
|
301 |
+
if self.elevenlabs_voice_settings:
|
302 |
+
if hasattr(self.elevenlabs_voice_settings,'model_dump'):vps["voice_settings"]=self.elevenlabs_voice_settings.model_dump()
|
303 |
+
elif hasattr(self.elevenlabs_voice_settings,'dict'):vps["voice_settings"]=self.elevenlabs_voice_settings.dict()
|
304 |
+
else:vps["voice_settings"]=self.elevenlabs_voice_settings
|
305 |
+
adi=asm(text=ttn,model_id="eleven_multilingual_v2",**vps)
|
306 |
+
with open(afp,"wb")as f:
|
307 |
+
for c in adi:
|
308 |
+
if c:f.write(c)
|
309 |
+
logger.info(f"11L audio (stream): {afp}");return afp
|
310 |
+
except Exception as e:logger.error(f"11L audio error: {e}",exc_info=True);return None
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
311 |
|
312 |
def assemble_animatic_from_assets(self, asset_data_list, overall_narration_path=None, output_filename="final_video.mp4", fps=24):
|
313 |
+
if not asset_data_list: logger.warning("No assets for animatic."); return None
|
314 |
+
processed_clips = []; narration_clip = None; final_clip = None
|
|
|
|
|
|
|
|
|
315 |
logger.info(f"Assembling from {len(asset_data_list)} assets. Frame: {self.video_frame_size}.")
|
316 |
|
317 |
for i, asset_info in enumerate(asset_data_list):
|
318 |
+
asset_path, asset_type, scene_dur = asset_info.get('path'), asset_info.get('type'), asset_info.get('duration', 4.5)
|
319 |
+
scene_num, key_action = asset_info.get('scene_num', i + 1), asset_info.get('key_action', '')
|
|
|
|
|
|
|
320 |
logger.info(f"S{scene_num}: Path='{asset_path}', Type='{asset_type}', Dur='{scene_dur}'s")
|
321 |
|
322 |
+
if not (asset_path and os.path.exists(asset_path)): logger.warning(f"S{scene_num}: Not found '{asset_path}'. Skip."); continue
|
323 |
+
if scene_dur <= 0: logger.warning(f"S{scene_num}: Invalid duration ({scene_dur}s). Skip."); continue
|
|
|
|
|
|
|
|
|
324 |
|
325 |
current_scene_mvpy_clip = None
|
326 |
try:
|
327 |
if asset_type == 'image':
|
328 |
+
pil_img = Image.open(asset_path); logger.debug(f"S{scene_num}: Loaded img. Mode:{pil_img.mode}, Size:{pil_img.size}")
|
|
|
329 |
img_rgba = pil_img.convert('RGBA') if pil_img.mode != 'RGBA' else pil_img.copy()
|
330 |
+
thumb = img_rgba.copy(); rf = Image.Resampling.LANCZOS if hasattr(Image.Resampling,'LANCZOS') else Image.BILINEAR; thumb.thumbnail(self.video_frame_size,rf)
|
331 |
+
cv_rgba = Image.new('RGBA',self.video_frame_size,(0,0,0,0)); xo,yo=(self.video_frame_size[0]-thumb.width)//2,(self.video_frame_size[1]-thumb.height)//2
|
332 |
+
cv_rgba.paste(thumb,(xo,yo),thumb)
|
333 |
+
final_rgb_pil = Image.new("RGB",self.video_frame_size,(0,0,0)); final_rgb_pil.paste(cv_rgba,mask=cv_rgba.split()[3])
|
334 |
+
dbg_path = os.path.join(self.output_dir,f"debug_PRE_NUMPY_S{scene_num}.png"); final_rgb_pil.save(dbg_path); logger.info(f"DEBUG: Saved PRE_NUMPY_S{scene_num} to {dbg_path}")
|
335 |
+
frame_np = np.array(final_rgb_pil,dtype=np.uint8);
|
336 |
+
if not frame_np.flags['C_CONTIGUOUS']: frame_np=np.ascontiguousarray(frame_np,dtype=np.uint8)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
337 |
logger.debug(f"S{scene_num}: NumPy for MoviePy. Shape:{frame_np.shape}, DType:{frame_np.dtype}, C-Contig:{frame_np.flags['C_CONTIGUOUS']}")
|
338 |
+
if frame_np.size==0 or frame_np.ndim!=3 or frame_np.shape[2]!=3: logger.error(f"S{scene_num}: Invalid NumPy. Skip."); continue
|
339 |
+
clip_base = ImageClip(frame_np,transparent=False).set_duration(scene_dur)
|
340 |
+
mvpy_dbg_path=os.path.join(self.output_dir,f"debug_MOVIEPY_FRAME_S{scene_num}.png"); clip_base.save_frame(mvpy_dbg_path,t=0.1); logger.info(f"DEBUG: Saved MOVIEPY_FRAME_S{scene_num} to {mvpy_dbg_path}")
|
|
|
|
|
|
|
|
|
341 |
clip_fx = clip_base
|
342 |
+
try: es=random.uniform(1.03,1.08); clip_fx=clip_base.fx(vfx.resize,lambda t:1+(es-1)*(t/scene_dur) if scene_dur>0 else 1).set_position('center')
|
343 |
+
except Exception as e: logger.error(f"S{scene_num} Ken Burns error: {e}",exc_info=False)
|
|
|
|
|
|
|
|
|
|
|
|
|
344 |
current_scene_mvpy_clip = clip_fx
|
345 |
elif asset_type == 'video':
|
346 |
+
src_clip=None
|
347 |
try:
|
348 |
+
src_clip=VideoFileClip(asset_path,target_resolution=(self.video_frame_size[1],self.video_frame_size[0])if self.video_frame_size else None, audio=False)
|
349 |
+
tmp_clip=src_clip
|
350 |
+
if src_clip.duration!=scene_dur:
|
351 |
+
if src_clip.duration>scene_dur:tmp_clip=src_clip.subclip(0,scene_dur)
|
|
|
|
|
|
|
|
|
|
|
352 |
else:
|
353 |
+
if scene_dur/src_clip.duration > 1.5 and src_clip.duration>0.1:tmp_clip=src_clip.loop(duration=scene_dur)
|
354 |
+
else:tmp_clip=src_clip.set_duration(src_clip.duration);logger.info(f"S{scene_num} Video clip ({src_clip.duration:.2f}s) shorter than target ({scene_dur:.2f}s).")
|
355 |
+
current_scene_mvpy_clip=tmp_clip.set_duration(scene_dur)
|
356 |
+
if current_scene_mvpy_clip.size!=list(self.video_frame_size):current_scene_mvpy_clip=current_scene_mvpy_clip.resize(self.video_frame_size)
|
357 |
+
except Exception as e:logger.error(f"S{scene_num} Video load error '{asset_path}':{e}",exc_info=True);continue
|
|
|
|
|
|
|
|
|
|
|
|
|
358 |
finally:
|
359 |
+
if src_clip and src_clip is not current_scene_mvpy_clip and hasattr(src_clip,'close'):src_clip.close()
|
360 |
+
else: logger.warning(f"S{scene_num} Unknown asset type '{asset_type}'. Skip."); continue
|
|
|
|
|
|
|
|
|
361 |
if current_scene_mvpy_clip and key_action:
|
362 |
try:
|
363 |
+
to_dur=min(current_scene_mvpy_clip.duration-0.5,current_scene_mvpy_clip.duration*0.8)if current_scene_mvpy_clip.duration>0.5 else current_scene_mvpy_clip.duration
|
364 |
+
to_start=0.25
|
365 |
+
txt_c=TextClip(f"Scene {scene_num}\n{key_action}",fontsize=self.video_overlay_font_size,color=self.video_overlay_font_color,font=self.video_overlay_font,bg_color='rgba(10,10,20,0.7)',method='caption',align='West',size=(self.video_frame_size[0]*0.9,None),kerning=-1,stroke_color='black',stroke_width=1.5).set_duration(to_dur).set_start(to_start).set_position(('center',0.92),relative=True)
|
366 |
+
current_scene_mvpy_clip=CompositeVideoClip([current_scene_mvpy_clip,txt_c],size=self.video_frame_size,use_bgclip=True)
|
367 |
+
except Exception as e:logger.error(f"S{scene_num} TextClip error:{e}. No text.",exc_info=True)
|
368 |
+
if current_scene_mvpy_clip:processed_clips.append(current_scene_mvpy_clip);logger.info(f"S{scene_num} Processed. Dur:{current_scene_mvpy_clip.duration:.2f}s.")
|
369 |
+
except Exception as e:logger.error(f"MAJOR Error S{scene_num} ({asset_path}):{e}",exc_info=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
370 |
finally:
|
371 |
+
if current_scene_mvpy_clip and hasattr(current_scene_mvpy_clip,'close'):
|
372 |
+
try: current_scene_mvpy_clip.close()
|
373 |
+
except: pass
|
|
|
|
|
374 |
|
375 |
+
if not processed_clips:logger.warning("No clips processed. Abort.");return None
|
376 |
+
td=0.75
|
|
|
|
|
377 |
try:
|
378 |
+
logger.info(f"Concatenating {len(processed_clips)} clips.");
|
379 |
+
if len(processed_clips)>1:final_clip=concatenate_videoclips(processed_clips,padding=-td if td>0 else 0,method="compose")
|
380 |
+
elif processed_clips:final_clip=processed_clips[0]
|
381 |
+
if not final_clip:logger.error("Concatenation failed.");return None
|
|
|
|
|
|
|
|
|
382 |
logger.info(f"Concatenated dur:{final_clip.duration:.2f}s")
|
383 |
+
if td>0 and final_clip.duration>0:
|
384 |
+
if final_clip.duration>td*2:final_clip=final_clip.fx(vfx.fadein,td).fx(vfx.fadeout,td)
|
385 |
+
else:final_clip=final_clip.fx(vfx.fadein,min(td,final_clip.duration/2.0))
|
386 |
+
if overall_narration_path and os.path.exists(overall_narration_path) and final_clip.duration>0:
|
387 |
+
try:narration_clip=AudioFileClip(overall_narration_path);final_clip=final_clip.set_audio(narration_clip);logger.info("Narration added.")
|
388 |
+
except Exception as e:logger.error(f"Narration add error:{e}",exc_info=True)
|
389 |
+
elif final_clip.duration<=0:logger.warning("Video no duration. No audio.")
|
390 |
+
if final_clip and final_clip.duration>0:
|
391 |
+
op=os.path.join(self.output_dir,output_filename);logger.info(f"Writing video:{op} (Dur:{final_clip.duration:.2f}s)")
|
392 |
+
final_clip.write_videofile(op,fps=fps,codec='libx264',preset='medium',audio_codec='aac',temp_audiofile=os.path.join(self.output_dir,f'temp-audio-{os.urandom(4).hex()}.m4a'),remove_temp=True,threads=os.cpu_count()or 2,logger='bar',bitrate="5000k",ffmpeg_params=["-pix_fmt", "yuv420p"])
|
393 |
+
logger.info(f"Video created:{op}");return op
|
394 |
+
else:logger.error("Final clip invalid. No write.");return None
|
395 |
+
except Exception as e:logger.error(f"Video write error:{e}",exc_info=True);return None
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
396 |
finally:
|
397 |
logger.debug("Closing all MoviePy clips in `assemble_animatic_from_assets` finally block.")
|
398 |
clips_to_close = processed_clips + ([narration_clip] if narration_clip else []) + ([final_clip] if final_clip else [])
|
399 |
for clip_obj in clips_to_close:
|
400 |
if clip_obj and hasattr(clip_obj, 'close'):
|
401 |
+
try: clip_obj.close()
|
402 |
+
except Exception as e_close: logger.warning(f"Ignoring error while closing a clip: {e_close}")
|
|
|
|