mgbam commited on
Commit
62ec987
·
verified ·
1 Parent(s): c8fceea

Update core/visual_engine.py

Browse files
Files changed (1) hide show
  1. core/visual_engine.py +305 -271
core/visual_engine.py CHANGED
@@ -4,7 +4,7 @@ import base64
4
  import mimetypes
5
  import numpy as np
6
  import os
7
- import openai
8
  import requests
9
  import io
10
  import time
@@ -15,302 +15,354 @@ from moviepy.editor import (ImageClip, VideoFileClip, concatenate_videoclips, Te
15
  CompositeVideoClip, AudioFileClip)
16
  import moviepy.video.fx.all as vfx
17
 
18
- try:
19
- if hasattr(Image, 'Resampling') and hasattr(Image.Resampling, 'LANCZOS'):
20
  if not hasattr(Image, 'ANTIALIAS'): Image.ANTIALIAS = Image.Resampling.LANCZOS
21
- elif hasattr(Image, 'LANCZOS'):
22
- if not hasattr(Image, 'ANTIALIAS'): Image.ANTIALIAS = Image.LANCZOS
23
  elif not hasattr(Image, 'ANTIALIAS'):
24
- print("WARNING: Pillow version lacks common Resampling or ANTIALIAS. MoviePy effects might fail.")
25
- except Exception as e_mp:
26
- print(f"WARNING: ANTIALIAS monkey-patch error: {e_mp}")
27
 
28
  logger = logging.getLogger(__name__)
29
- # logger.setLevel(logging.DEBUG) # Uncomment for very verbose debugging
30
 
31
  ELEVENLABS_CLIENT_IMPORTED = False; ElevenLabsAPIClient = None; Voice = None; VoiceSettings = None
32
  try:
33
  from elevenlabs.client import ElevenLabs as ImportedElevenLabsClient
34
  from elevenlabs import Voice as ImportedVoice, VoiceSettings as ImportedVoiceSettings
35
  ElevenLabsAPIClient = ImportedElevenLabsClient; Voice = ImportedVoice; VoiceSettings = ImportedVoiceSettings
36
- ELEVENLABS_CLIENT_IMPORTED = True; logger.info("ElevenLabs client components imported.")
37
- except Exception as e_11l_imp: logger.warning(f"ElevenLabs client import failed: {e_11l_imp}. Audio disabled.")
 
38
 
39
  RUNWAYML_SDK_IMPORTED = False; RunwayMLAPIClientClass = None
40
  try:
41
  from runwayml import RunwayML as ImportedRunwayMLAPIClientClass
42
  RunwayMLAPIClientClass = ImportedRunwayMLAPIClientClass; RUNWAYML_SDK_IMPORTED = True
43
- logger.info("RunwayML SDK imported.")
44
- except Exception as e_rwy_imp: logger.warning(f"RunwayML SDK import failed: {e_rwy_imp}. RunwayML disabled.")
 
 
45
 
46
  class VisualEngine:
47
  DEFAULT_FONT_SIZE_PIL = 10; PREFERRED_FONT_SIZE_PIL = 20
48
  VIDEO_OVERLAY_FONT_SIZE = 30; VIDEO_OVERLAY_FONT_COLOR = 'white'
49
  DEFAULT_MOVIEPY_FONT = 'DejaVu-Sans-Bold'; PREFERRED_MOVIEPY_FONT = 'Liberation-Sans-Bold'
50
 
 
51
  def __init__(self, output_dir="temp_cinegen_media", default_elevenlabs_voice_id="Rachel"):
52
- self.output_dir = output_dir; os.makedirs(self.output_dir, exist_ok=True)
 
 
 
 
 
 
 
 
53
  self.font_filename_pil_preference = "DejaVuSans-Bold.ttf"
54
  font_paths = [ self.font_filename_pil_preference, f"/usr/share/fonts/truetype/dejavu/{self.font_filename_pil_preference}", f"/usr/share/fonts/truetype/liberation/LiberationSans-Bold.ttf", f"/System/Library/Fonts/Supplemental/Arial.ttf", f"C:/Windows/Fonts/arial.ttf", f"/usr/local/share/fonts/truetype/mycustomfonts/arial.ttf"]
55
  self.resolved_font_path_pil = next((p for p in font_paths if os.path.exists(p)), None)
56
- self.active_font_pil = ImageFont.load_default(); self.active_font_size_pil = self.DEFAULT_FONT_SIZE_PIL; self.active_moviepy_font_name = self.DEFAULT_MOVIEPY_FONT
 
 
 
 
57
  if self.resolved_font_path_pil:
58
- try: self.active_font_pil = ImageFont.truetype(self.resolved_font_path_pil, self.PREFERRED_FONT_SIZE_PIL); self.active_font_size_pil = self.PREFERRED_FONT_SIZE_PIL; logger.info(f"Pillow font: {self.resolved_font_path_pil} sz {self.active_font_size_pil}."); self.active_moviepy_font_name = 'DejaVu-Sans-Bold' if "dejavu" in self.resolved_font_path_pil.lower() else ('Liberation-Sans-Bold' if "liberation" in self.resolved_font_path_pil.lower() else self.DEFAULT_MOVIEPY_FONT)
 
 
 
 
 
59
  except IOError as e_font: logger.error(f"Pillow font IOError '{self.resolved_font_path_pil}': {e_font}. Default.")
60
  else: logger.warning("Preferred Pillow font not found. Default.")
 
61
  self.openai_api_key = None; self.USE_AI_IMAGE_GENERATION = False; self.dalle_model = "dall-e-3"; self.image_size_dalle3 = "1792x1024"
62
  self.video_frame_size = (1280, 720)
63
- self.elevenlabs_api_key = None; self.USE_ELEVENLABS = False; self.elevenlabs_client_instance = None; self.elevenlabs_voice_id = default_elevenlabs_voice_id
 
 
 
 
64
  if VoiceSettings and ELEVENLABS_CLIENT_IMPORTED: self.elevenlabs_voice_settings_obj = VoiceSettings(stability=0.60, similarity_boost=0.80, style=0.15, use_speaker_boost=True)
65
  else: self.elevenlabs_voice_settings_obj = None
 
66
  self.pexels_api_key = None; self.USE_PEXELS = False
67
  self.runway_api_key = None; self.USE_RUNWAYML = False; self.runway_ml_sdk_client_instance = None
 
68
  if RUNWAYML_SDK_IMPORTED and RunwayMLAPIClientClass and os.getenv("RUNWAYML_API_SECRET"):
69
  try: self.runway_ml_sdk_client_instance = RunwayMLAPIClientClass(); self.USE_RUNWAYML = True; logger.info("RunwayML Client init from env var at startup.")
70
  except Exception as e_rwy_init: logger.error(f"Initial RunwayML client init failed: {e_rwy_init}"); self.USE_RUNWAYML = False
71
- logger.info("VisualEngine initialized.")
 
72
 
73
  def set_openai_api_key(self, api_key_value): self.openai_api_key = api_key_value; self.USE_AI_IMAGE_GENERATION = bool(api_key_value); logger.info(f"DALL-E status: {'Ready' if self.USE_AI_IMAGE_GENERATION else 'Disabled'}")
74
- def set_elevenlabs_api_key(self, api_key_value, voice_id_from_secret=None):
 
75
  self.elevenlabs_api_key = api_key_value
76
- if voice_id_from_secret: self.elevenlabs_voice_id = voice_id_from_secret; logger.info(f"11L Voice ID updated to: {self.elevenlabs_voice_id} via set_elevenlabs_api_key.")
 
77
  if api_key_value and ELEVENLABS_CLIENT_IMPORTED and ElevenLabsAPIClient:
78
- try: self.elevenlabs_client_instance = ElevenLabsAPIClient(api_key=api_key_value); self.USE_ELEVENLABS = bool(self.elevenlabs_client_instance); logger.info(f"11L Client: {'Ready' if self.USE_ELEVENLABS else 'Failed'} (Voice: {self.elevenlabs_voice_id})")
79
  except Exception as e_11l_setkey_init: logger.error(f"11L client init error: {e_11l_setkey_init}. Disabled.", exc_info=True); self.USE_ELEVENLABS=False; self.elevenlabs_client_instance=None
80
- else: self.USE_ELEVENLABS = False; logger.info(f"11L Disabled (key/SDK).")
 
81
  def set_pexels_api_key(self, api_key_value): self.pexels_api_key = api_key_value; self.USE_PEXELS = bool(api_key_value); logger.info(f"Pexels status: {'Ready' if self.USE_PEXELS else 'Disabled'}")
 
82
  def set_runway_api_key(self, api_key_value):
83
  self.runway_api_key = api_key_value
84
  if api_key_value:
85
  if RUNWAYML_SDK_IMPORTED and RunwayMLAPIClientClass:
86
- if not self.runway_ml_sdk_client_instance:
87
  try:
88
  original_env_secret = os.getenv("RUNWAYML_API_SECRET")
89
- if not original_env_secret: os.environ["RUNWAYML_API_SECRET"] = api_key_value; logger.info("Temp set RUNWAYML_API_SECRET for SDK.")
90
- self.runway_ml_sdk_client_instance = RunwayMLAPIClientClass(); self.USE_RUNWAYML = True; logger.info("RunwayML Client init via set_runway_api_key.")
91
- if not original_env_secret: del os.environ["RUNWAYML_API_SECRET"]; logger.info("Cleared temp RUNWAYML_API_SECRET.")
92
- except Exception as e_runway_setkey_init: logger.error(f"RunwayML Client init in set_runway_api_key fail: {e_runway_setkey_init}", exc_info=True); self.USE_RUNWAYML=False;self.runway_ml_sdk_client_instance=None
93
- else: self.USE_RUNWAYML = True; logger.info("RunwayML Client already init.")
94
- else: logger.warning("RunwayML SDK not imported. Service disabled."); self.USE_RUNWAYML = False
95
- else: self.USE_RUNWAYML = False; self.runway_ml_sdk_client_instance = None; logger.info("RunwayML Disabled (no API key).")
96
 
 
 
97
  def _image_to_data_uri(self, image_path):
98
  try:
99
  mime_type, _ = mimetypes.guess_type(image_path)
100
- if not mime_type:
101
- ext = os.path.splitext(image_path)[1].lower()
102
- mime_map = {".png": "image/png", ".jpg": "image/jpeg", ".jpeg": "image/jpeg", ".webp": "image/webp"}
103
- mime_type = mime_map.get(ext, "application/octet-stream")
104
- if mime_type == "application/octet-stream": logger.warning(f"Could not determine MIME type for {image_path} from extension '{ext}', using default {mime_type}.")
105
  with open(image_path, "rb") as image_file_handle: image_binary_data = image_file_handle.read()
106
  encoded_base64_string = base64.b64encode(image_binary_data).decode('utf-8')
107
- data_uri_string = f"data:{mime_type};base64,{encoded_base64_string}"
108
- logger.debug(f"Generated data URI for {os.path.basename(image_path)} (MIME: {mime_type}). Data URI starts with: {data_uri_string[:100]}...")
109
- return data_uri_string
110
- except FileNotFoundError: logger.error(f"Image file not found at path: '{image_path}' when trying to create data URI."); return None
111
- except Exception as e_data_uri_conversion: logger.error(f"Error converting image '{image_path}' to data URI: {e_data_uri_conversion}", exc_info=True); return None
112
 
113
  def _map_resolution_to_runway_ratio(self, width, height):
114
  ratio_str=f"{width}:{height}";supported_ratios_gen4=["1280:720","720:1280","1104:832","832:1104","960:960","1584:672"];
115
  if ratio_str in supported_ratios_gen4:return ratio_str
116
- logger.warning(f"Res {ratio_str} not in Gen-4 list. Default 1280:720.");return "1280:720"
117
 
118
  def _get_text_dimensions(self, text_content, font_object_pil):
119
- default_h = getattr(font_object_pil, 'size', self.active_font_size_pil)
120
- if not text_content: return 0, default_h
121
  try:
122
- if hasattr(font_object_pil,'getbbox'):bbox=font_object_pil.getbbox(text_content);w=bbox[2]-bbox[0];h=bbox[3]-bbox[1]; return w, h if h > 0 else default_h
123
- elif hasattr(font_object_pil,'getsize'):w,h=font_object_pil.getsize(text_content); return w, h if h > 0 else default_h
124
- else: return int(len(text_content)*default_h*0.6),int(default_h*1.2)
125
- except Exception as e_getdim: logger.warning(f"Error in _get_text_dimensions: {e_getdim}"); return int(len(text_content)*self.active_font_size_pil*0.6),int(self.active_font_size_pil*1.2)
126
 
127
  def _create_placeholder_image_content(self,text_description,filename,size=None):
 
128
  if size is None: size = self.video_frame_size
129
- img = Image.new('RGB', size, color=(20, 20, 40)); d = ImageDraw.Draw(img); padding = 25
130
- max_w = size[0] - (2 * padding); lines_for_placeholder = []
131
  if not text_description: text_description = "(Placeholder Image)"
132
- words_list = text_description.split(); current_line_buffer = ""
133
- for word_idx, word_item in enumerate(words_list):
134
- prospective_addition = word_item + (" " if word_idx < len(words_list) - 1 else "")
135
- test_line_candidate = current_line_buffer + prospective_addition
136
- current_w_text, _ = self._get_text_dimensions(test_line_candidate, self.active_font_pil)
137
- if current_w_text == 0 and test_line_candidate.strip(): current_w_text = len(test_line_candidate) * (self.active_font_size_pil * 0.6)
138
- if current_w_text <= max_w: current_line_buffer = test_line_candidate
139
  else:
140
- if current_line_buffer.strip(): lines_for_placeholder.append(current_line_buffer.strip())
141
- current_line_buffer = prospective_addition
142
- if current_line_buffer.strip(): lines_for_placeholder.append(current_line_buffer.strip())
143
- if not lines_for_placeholder and text_description:
144
- avg_char_w_est, _ = self._get_text_dimensions("W", self.active_font_pil); avg_char_w_est = avg_char_w_est or (self.active_font_size_pil * 0.6)
145
- chars_per_line_est = int(max_w / avg_char_w_est) if avg_char_w_est > 0 else 20
146
- lines_for_placeholder.append(text_description[:chars_per_line_est] + ("..." if len(text_description) > chars_per_line_est else ""))
147
- elif not lines_for_placeholder: lines_for_placeholder.append("(Placeholder Error)")
148
- _, single_h = self._get_text_dimensions("Ay", self.active_font_pil); single_h = single_h if single_h > 0 else self.active_font_size_pil + 2
149
- max_l = min(len(lines_for_placeholder), (size[1] - (2 * padding)) // (single_h + 2)) if single_h > 0 else 1; max_l = max(1, max_l)
150
- y_p = padding + (size[1] - (2 * padding) - max_l * (single_h + 2)) / 2.0
151
- for i_line in range(max_l):
152
- line_txt_content = lines_for_placeholder[i_line]; line_w_val, _ = self._get_text_dimensions(line_txt_content, self.active_font_pil)
153
- if line_w_val == 0 and line_txt_content.strip(): line_w_val = len(line_txt_content) * (self.active_font_size_pil * 0.6)
154
- x_p = (size[0] - line_w_val) / 2.0
155
- try: d.text((x_p, y_p), line_txt_content, font=self.active_font_pil, fill=(200, 200, 180))
156
- except Exception as e_draw: logger.error(f"Pillow d.text error: {e_draw} for '{line_txt_content}'")
157
- y_p += single_h + 2
158
- if i_line == 6 and max_l > 7:
159
- try: d.text((x_p, y_p), "...", font=self.active_font_pil, fill=(200, 200, 180))
160
- except Exception as e_elip: logger.error(f"Pillow d.text ellipsis error: {e_elip}"); break
161
- filepath_placeholder = os.path.join(self.output_dir, filename)
162
- try: img.save(filepath_placeholder); return filepath_placeholder
163
- except Exception as e_save: logger.error(f"Saving placeholder image '{filepath_placeholder}' error: {e_save}", exc_info=True); return None
164
 
165
- def _search_pexels_image(self, query_str, output_fn_base):
166
  if not self.USE_PEXELS or not self.pexels_api_key: return None
167
- http_headers = {"Authorization": self.pexels_api_key}
168
- http_params = {"query": query_str, "per_page": 1, "orientation": "landscape", "size": "large2x"}
169
- base_name_px, _ = os.path.splitext(output_fn_base)
170
- pexels_fn_str = base_name_px + f"_pexels_{random.randint(1000,9999)}.jpg"
171
- file_path_px = os.path.join(self.output_dir, pexels_fn_str)
172
  try:
173
- logger.info(f"Pexels: Searching for '{query_str}'")
174
- eff_query_px = " ".join(query_str.split()[:5])
175
- http_params["query"] = eff_query_px
176
- response_px = requests.get("https://api.pexels.com/v1/search", headers=http_headers, params=http_params, timeout=20)
177
- response_px.raise_for_status()
178
- data_px = response_px.json()
179
- if data_px.get("photos") and len(data_px["photos"]) > 0:
180
- photo_details_px = data_px["photos"][0]
181
- photo_url_px = photo_details_px.get("src", {}).get("large2x")
182
- if not photo_url_px: logger.warning(f"Pexels: 'large2x' URL missing for '{eff_query_px}'. Details: {photo_details_px}"); return None
183
- image_response_px = requests.get(photo_url_px, timeout=60); image_response_px.raise_for_status()
184
- img_pil_data_px = Image.open(io.BytesIO(image_response_px.content))
185
- if img_pil_data_px.mode != 'RGB': img_pil_data_px = img_pil_data_px.convert('RGB')
186
- img_pil_data_px.save(file_path_px); logger.info(f"Pexels: Image saved to {file_path_px}"); return file_path_px
187
- else: logger.info(f"Pexels: No photos for '{eff_query_px}'."); return None
188
- except requests.exceptions.RequestException as e_req_px: logger.error(f"Pexels: RequestException for '{query_str}': {e_req_px}", exc_info=False); return None
189
- except Exception as e_px_gen: logger.error(f"Pexels: General error for '{query_str}': {e_px_gen}", exc_info=True); return None
190
 
191
- def _generate_video_clip_with_runwayml(self, text_prompt_for_motion, input_image_path, scene_identifier_filename_base, target_duration_seconds=5):
192
- if not self.USE_RUNWAYML or not self.runway_ml_sdk_client_instance: logger.warning("RunwayML not enabled/client not init. Skip video."); return None
193
- if not input_image_path or not os.path.exists(input_image_path): logger.error(f"Runway Gen-4 needs input image. Path invalid: {input_image_path}"); return None
194
- image_data_uri_str = self._image_to_data_uri(input_image_path)
195
- if not image_data_uri_str: return None
196
- runway_dur = 10 if target_duration_seconds >= 8 else 5
197
- runway_ratio = self._map_resolution_to_runway_ratio(self.video_frame_size[0], self.video_frame_size[1])
198
- base_name_for_runway_vid, _ = os.path.splitext(scene_identifier_filename_base); output_vid_fn = base_name_for_runway_vid + f"_runway_gen4_d{runway_dur}s.mp4"
199
- output_vid_fp = os.path.join(self.output_dir, output_vid_fn)
200
- logger.info(f"Runway Gen-4 task: motion='{text_prompt_for_motion[:100]}...', img='{os.path.basename(input_image_path)}', dur={runway_dur}s, ratio='{runway_ratio}'")
201
  try:
202
- task_submitted_runway = self.runway_ml_sdk_client_instance.image_to_video.create(model='gen4_turbo', prompt_image=image_data_uri_str, prompt_text=text_prompt_for_motion, duration=runway_dur, ratio=runway_ratio)
203
- task_id_runway = task_submitted_runway.id; logger.info(f"Runway Gen-4 task ID: {task_id_runway}. Polling...")
204
- poll_sec=10; max_poll_count=36; poll_start_time = time.time()
205
- while time.time() - poll_start_time < max_poll_count * poll_sec:
206
- time.sleep(poll_sec); task_details_runway = self.runway_ml_sdk_client_instance.tasks.retrieve(id=task_id_runway)
207
- logger.info(f"Runway task {task_id_runway} status: {task_details_runway.status}")
208
- if task_details_runway.status == 'SUCCEEDED':
209
- output_url_runway = getattr(getattr(task_details_runway,'output',None),'url',None) or \
210
- (getattr(task_details_runway,'artifacts',None) and task_details_runway.artifacts and hasattr(task_details_runway.artifacts[0],'url')and task_details_runway.artifacts[0].url) or \
211
- (getattr(task_details_runway,'artifacts',None) and task_details_runway.artifacts and hasattr(task_details_runway.artifacts[0],'download_url')and task_details_runway.artifacts[0].download_url)
212
- if not output_url_runway: logger.error(f"Runway task {task_id_runway} SUCCEEDED, but no output URL. Details: {vars(task_details_runway) if hasattr(task_details_runway,'__dict__') else task_details_runway}"); return None
213
- logger.info(f"Runway task {task_id_runway} SUCCEEDED. Downloading: {output_url_runway}")
214
- video_resp_get = requests.get(output_url_runway, stream=True, timeout=300); video_resp_get.raise_for_status()
215
- with open(output_vid_fp,'wb') as f_vid:
216
- for chunk_data in video_resp_get.iter_content(chunk_size=8192): f_vid.write(chunk_data)
217
- logger.info(f"Runway Gen-4 video saved: {output_vid_fp}"); return output_vid_fp
218
- elif task_details_runway.status in ['FAILED','ABORTED','ERROR']:
219
- err_msg_runway = getattr(task_details_runway,'error_message',None) or getattr(getattr(task_details_runway,'output',None),'error',"Unknown Runway error.")
220
- logger.error(f"Runway task {task_id_runway} status: {task_details_runway.status}. Error: {err_msg_runway}"); return None
221
- logger.warning(f"Runway task {task_id_runway} timed out."); return None
222
- except AttributeError as ae_sdk: logger.error(f"RunwayML SDK AttrError: {ae_sdk}. SDK/methods changed?", exc_info=True); return None
223
- except Exception as e_runway_gen: logger.error(f"Runway Gen-4 API error: {e_runway_gen}", exc_info=True); return None
224
 
225
- def _create_placeholder_video_content(self, text_desc_ph, filename_ph, duration_ph=4, size_ph=None):
226
- if size_ph is None: size_ph = self.video_frame_size
227
- filepath_ph = os.path.join(self.output_dir, filename_ph)
228
- text_clip_ph = None
229
  try:
230
- text_clip_ph = TextClip(text_desc_ph, fontsize=50, color='white', font=self.video_overlay_font,
231
- bg_color='black', size=size_ph, method='caption').set_duration(duration_ph)
232
- text_clip_ph.write_videofile(filepath_ph, fps=24, codec='libx264', preset='ultrafast', logger=None, threads=2)
233
- logger.info(f"Generic placeholder video created: {filepath_ph}")
234
- return filepath_ph
235
- except Exception as e_ph_vid:
236
- logger.error(f"Failed to create generic placeholder video '{filepath_ph}': {e_ph_vid}", exc_info=True)
237
  return None
238
  finally:
239
- if text_clip_ph and hasattr(text_clip_ph, 'close'):
240
- try: text_clip_ph.close()
241
- except Exception as e_cl_phv: logger.warning(f"Ignoring error closing placeholder TextClip: {e_cl_phv}")
242
 
243
  def generate_scene_asset(self, image_generation_prompt_text, motion_prompt_text_for_video,
244
- scene_data_dict, scene_identifier_fn_base,
245
- generate_as_video_clip_flag=False, runway_target_dur_val=5):
246
- base_name_asset, _ = os.path.splitext(scene_identifier_fn_base)
247
- asset_info_result = {'path': None, 'type': 'none', 'error': True, 'prompt_used': image_generation_prompt_text, 'error_message': 'Asset generation init failed'}
248
- path_for_input_image_runway = None
249
- fn_for_base_image = base_name_asset + ("_base_for_video.png" if generate_as_video_clip_flag else ".png")
250
- fp_for_base_image = os.path.join(self.output_dir, fn_for_base_image)
 
 
 
 
 
251
  if self.USE_AI_IMAGE_GENERATION and self.openai_api_key:
252
- max_r_dalle, attempt_count_dalle = 2,0;
253
- for att_n_dalle in range(max_r_dalle):
254
- attempt_count_dalle = att_n_dalle + 1
255
  try:
256
- logger.info(f"Att {attempt_count_dalle} DALL-E (base img): {image_generation_prompt_text[:70]}..."); oai_cl = openai.OpenAI(api_key=self.openai_api_key,timeout=90.0); oai_r = oai_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"); oai_iu = oai_r.data[0].url; oai_rp = getattr(oai_r.data[0],'revised_prompt',None);
257
- if oai_rp: logger.info(f"DALL-E revised: {oai_rp[:70]}...")
258
- oai_ir = requests.get(oai_iu,timeout=120); oai_ir.raise_for_status(); oai_id = Image.open(io.BytesIO(oai_ir.content));
259
- if oai_id.mode!='RGB': oai_id=oai_id.convert('RGB')
260
- oai_id.save(fp_for_base_image); logger.info(f"DALL-E base img saved: {fp_for_base_image}"); path_for_input_image_runway=fp_for_base_image; asset_info_result={'path':fp_for_base_image,'type':'image','error':False,'prompt_used':image_generation_prompt_text,'revised_prompt':oai_rp}; break
261
- except openai.RateLimitError as e_oai_rl: logger.warning(f"OpenAI RateLimit Att {attempt_count_dalle}:{e_oai_rl}.Retry...");time.sleep(5*attempt_count_dalle);asset_info_result['error_message']=str(e_oai_rl)
262
- except openai.APIError as e_oai_api: logger.error(f"OpenAI APIError Att {attempt_count_dalle}:{e_oai_api}");asset_info_result['error_message']=str(e_oai_api);break
263
- except requests.exceptions.RequestException as e_oai_req: logger.error(f"Requests Err DALL-E Att {attempt_count_dalle}:{e_oai_req}");asset_info_result['error_message']=str(e_oai_req);break
264
- except Exception as e_oai_gen: logger.error(f"General DALL-E Err Att {attempt_count_dalle}:{e_oai_gen}",exc_info=True);asset_info_result['error_message']=str(e_oai_gen);break
265
- if asset_info_result['error']: logger.warning(f"DALL-E failed after {attempt_count_dalle} attempts for base img.")
266
- if asset_info_result['error'] and self.USE_PEXELS:
267
- logger.info("Trying Pexels for base img.");px_qt=scene_data_dict.get('pexels_search_query_감독',f"{scene_data_dict.get('emotional_beat','')} {scene_data_dict.get('setting_description','')}");px_pp=self._search_pexels_image(px_qt,fn_for_base_image);
268
- if px_pp:path_for_input_image_runway=px_pp;asset_info_result={'path':px_pp,'type':'image','error':False,'prompt_used':f"Pexels:{px_qt}"}
269
- else:current_em_px=asset_info_result.get('error_message',"");asset_info_result['error_message']=(current_em_px+" Pexels failed for base.").strip()
270
- if asset_info_result['error']:
271
- logger.warning("Base img (DALL-E/Pexels) failed. Using placeholder.");ph_ppt=asset_info_result.get('prompt_used',image_generation_prompt_text);php=self._create_placeholder_image_content(f"[Base Placeholder]{ph_ppt[:70]}...",fn_for_base_image);
272
- if php:path_for_input_image_runway=php;asset_info_result={'path':php,'type':'image','error':False,'prompt_used':ph_ppt}
273
- else:current_em_ph=asset_info_result.get('error_message',"");asset_info_result['error_message']=(current_em_ph+" Base placeholder failed.").strip()
274
- if generate_as_video_clip_flag:
275
- if not path_for_input_image_runway:logger.error("RunwayML video: base img failed.");asset_info_result['error']=True;asset_info_result['error_message']=(asset_info_result.get('error_message',"")+" Base img miss, Runway abort.").strip();asset_info_result['type']='none';return asset_info_result
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
276
  if self.USE_RUNWAYML:
277
- runway_video_p=self._generate_video_clip_with_runwayml(motion_prompt_text_for_video,path_for_input_image_runway,base_name_asset,runway_target_dur_val)
278
- if runway_video_p and os.path.exists(runway_video_p):asset_info_result={'path':runway_video_p,'type':'video','error':False,'prompt_used':motion_prompt_text_for_video,'base_image_path':path_for_input_image_runway}
279
- else:logger.warning(f"RunwayML video failed for {base_name_asset}. Fallback to base img.");asset_info_result['error']=True;asset_info_result['error_message']=(asset_info_result.get('error_message',"Base img ok.")+" RunwayML video fail; use base img.").strip();asset_info_result['path']=path_for_input_image_runway;asset_info_result['type']='image';asset_info_result['prompt_used']=image_generation_prompt_text
280
- else:logger.warning("RunwayML selected but disabled. Use base img.");asset_info_result['error']=True;asset_info_result['error_message']=(asset_info_result.get('error_message',"Base img ok.")+" RunwayML disabled; use base img.").strip();asset_info_result['path']=path_for_input_image_runway;asset_info_result['type']='image';asset_info_result['prompt_used']=image_generation_prompt_text
281
- return asset_info_result
282
 
283
- def generate_narration_audio(self, text_to_narrate, output_filename="narration_overall.mp3"):
284
- if not self.USE_ELEVENLABS or not self.elevenlabs_client_instance or not text_to_narrate: logger.info("11L conditions not met. Skip audio."); return None
285
- audio_filepath_narration = os.path.join(self.output_dir, output_filename)
 
286
  try:
287
- logger.info(f"Generating 11L audio (Voice ID: {self.elevenlabs_voice_id}) for text: \"{text_to_narrate[:70]}...\"")
288
- audio_stream_method_11l = None
289
- if hasattr(self.elevenlabs_client_instance, 'text_to_speech') and hasattr(self.elevenlabs_client_instance.text_to_speech, 'stream'): audio_stream_method_11l = self.elevenlabs_client_instance.text_to_speech.stream; logger.info("Using 11L SDK: client.text_to_speech.stream()")
290
- elif hasattr(self.elevenlabs_client_instance, 'generate_stream'): audio_stream_method_11l = self.elevenlabs_client_instance.generate_stream; logger.info("Using 11L SDK: client.generate_stream()")
291
- elif hasattr(self.elevenlabs_client_instance, 'generate'):
292
- logger.info("Using 11L SDK: client.generate() (non-streaming).")
293
- voice_param_11l = str(self.elevenlabs_voice_id);
294
- if Voice and self.elevenlabs_voice_settings_obj: voice_param_11l = Voice(voice_id=str(self.elevenlabs_voice_id), settings=self.elevenlabs_voice_settings_obj)
295
- audio_bytes_data = self.elevenlabs_client_instance.generate(text=text_to_narrate, voice=voice_param_11l, model="eleven_multilingual_v2")
296
- with open(audio_filepath_narration, "wb") as audio_file_out: audio_file_out.write(audio_bytes_data)
297
- logger.info(f"11L audio (non-streamed) saved to: {audio_filepath_narration}"); return audio_filepath_narration
298
- else: logger.error("No recognized audio generation method on 11L client."); return None
299
- if audio_stream_method_11l:
300
- params_for_voice_stream = {"voice_id": str(self.elevenlabs_voice_id)}
301
  if self.elevenlabs_voice_settings_obj:
302
- if hasattr(self.elevenlabs_voice_settings_obj, 'model_dump'): params_for_voice_stream["voice_settings"] = self.elevenlabs_voice_settings_obj.model_dump()
303
- elif hasattr(self.elevenlabs_voice_settings_obj, 'dict'): params_for_voice_stream["voice_settings"] = self.elevenlabs_voice_settings_obj.dict()
304
- else: params_for_voice_stream["voice_settings"] = self.elevenlabs_voice_settings_obj
305
- audio_data_iterator_11l = audio_stream_method_11l(text=text_to_narrate, model_id="eleven_multilingual_v2", **params_for_voice_stream)
306
- with open(audio_filepath_narration, "wb") as audio_file_out_stream:
307
- for audio_chunk_data in audio_data_iterator_11l:
308
- if audio_chunk_data: audio_file_out_stream.write(audio_chunk_data)
309
- logger.info(f"11L audio (streamed) saved to: {audio_filepath_narration}"); return audio_filepath_narration
310
- except AttributeError as ae_11l_sdk: logger.error(f"AttributeError with 11L SDK client: {ae_11l_sdk}. SDK version/methods might differ.", exc_info=True); return None
311
- except Exception as e_11l_general_audio: logger.error(f"General error during 11L audio generation: {e_11l_general_audio}", exc_info=True); return None
312
 
313
  def assemble_animatic_from_assets(self, asset_data_list, overall_narration_path=None, output_filename="final_video.mp4", fps=24):
 
 
 
 
 
 
314
  if not asset_data_list: logger.warning("No assets for animatic."); return None
315
  processed_moviepy_clips_list = []; narration_audio_clip_mvpy = None; final_video_output_clip = None
316
  logger.info(f"Assembling from {len(asset_data_list)} assets. Target Frame: {self.video_frame_size}.")
@@ -319,69 +371,48 @@ class VisualEngine:
319
  path_of_asset, type_of_asset, duration_for_scene = asset_info_item_loop.get('path'), asset_info_item_loop.get('type'), asset_info_item_loop.get('duration', 4.5)
320
  num_of_scene, action_in_key = asset_info_item_loop.get('scene_num', i_asset + 1), asset_info_item_loop.get('key_action', '')
321
  logger.info(f"S{num_of_scene}: Path='{path_of_asset}', Type='{type_of_asset}', Dur='{duration_for_scene}'s")
322
-
323
  if not (path_of_asset and os.path.exists(path_of_asset)): logger.warning(f"S{num_of_scene}: Not found '{path_of_asset}'. Skip."); continue
324
  if duration_for_scene <= 0: logger.warning(f"S{num_of_scene}: Invalid duration ({duration_for_scene}s). Skip."); continue
325
-
326
  active_scene_clip = None
327
  try:
328
  if type_of_asset == 'image':
329
- pil_img_original = Image.open(path_of_asset)
330
- logger.debug(f"S{num_of_scene} (0-Load): Original loaded. Mode:{pil_img_original.mode}, Size:{pil_img_original.size}")
331
- pil_img_original.save(os.path.join(self.output_dir,f"debug_0_ORIGINAL_S{num_of_scene}.png"))
332
-
333
- img_rgba_intermediate = pil_img_original.convert('RGBA') if pil_img_original.mode != 'RGBA' else pil_img_original.copy().convert('RGBA')
334
- logger.debug(f"S{num_of_scene} (1-ToRGBA): Converted to RGBA. Mode:{img_rgba_intermediate.mode}, Size:{img_rgba_intermediate.size}")
335
- img_rgba_intermediate.save(os.path.join(self.output_dir,f"debug_1_AS_RGBA_S{num_of_scene}.png"))
336
-
337
- thumbnailed_img_rgba = img_rgba_intermediate.copy()
338
- resample_filter_pil = Image.Resampling.LANCZOS if hasattr(Image.Resampling,'LANCZOS') else Image.BILINEAR
339
- thumbnailed_img_rgba.thumbnail(self.video_frame_size, resample_filter_pil)
340
- logger.debug(f"S{num_of_scene} (2-Thumbnail): Thumbnailed RGBA. Mode:{thumbnailed_img_rgba.mode}, Size:{thumbnailed_img_rgba.size}")
341
- thumbnailed_img_rgba.save(os.path.join(self.output_dir,f"debug_2_THUMBNAIL_RGBA_S{num_of_scene}.png"))
342
-
343
- canvas_for_compositing_rgba = Image.new('RGBA', self.video_frame_size, (0,0,0,0))
344
- pos_x_paste = (self.video_frame_size[0] - thumbnailed_img_rgba.width) // 2
345
- pos_y_paste = (self.video_frame_size[1] - thumbnailed_img_rgba.height) // 2
346
- canvas_for_compositing_rgba.paste(thumbnailed_img_rgba, (pos_x_paste, pos_y_paste), thumbnailed_img_rgba)
347
- logger.debug(f"S{num_of_scene} (3-PasteOnRGBA): Image pasted onto transparent RGBA canvas. Mode:{canvas_for_compositing_rgba.mode}, Size:{canvas_for_compositing_rgba.size}")
348
- canvas_for_compositing_rgba.save(os.path.join(self.output_dir,f"debug_3_COMPOSITED_RGBA_S{num_of_scene}.png"))
349
-
350
- final_rgb_image_for_pil = Image.new("RGB", self.video_frame_size, (0, 0, 0))
351
- if canvas_for_compositing_rgba.mode == 'RGBA': final_rgb_image_for_pil.paste(canvas_for_compositing_rgba, mask=canvas_for_compositing_rgba.split()[3])
352
- else: final_rgb_image_for_pil.paste(canvas_for_compositing_rgba)
353
- logger.debug(f"S{num_of_scene} (4-ToRGB): Final RGB image created. Mode:{final_rgb_image_for_pil.mode}, Size:{final_rgb_image_for_pil.size}")
354
 
355
- debug_path_img_pre_numpy = os.path.join(self.output_dir,f"debug_4_PRE_NUMPY_RGB_S{num_of_scene}.png");
356
- final_rgb_image_for_pil.save(debug_path_img_pre_numpy);
357
- logger.info(f"CRITICAL DEBUG: Saved PRE_NUMPY_RGB_S{num_of_scene} to {debug_path_img_pre_numpy}")
358
-
359
- numpy_frame_arr = np.array(final_rgb_image_for_pil, dtype=np.uint8)
360
- if not numpy_frame_arr.flags['C_CONTIGUOUS']: numpy_frame_arr = np.ascontiguousarray(numpy_frame_arr, dtype=np.uint8)
361
- logger.debug(f"S{num_of_scene} (5-NumPy): Final NumPy. Shape:{numpy_frame_arr.shape}, DType:{numpy_frame_arr.dtype}, Flags:{numpy_frame_arr.flags}")
362
- if numpy_frame_arr.size == 0 or numpy_frame_arr.ndim != 3 or numpy_frame_arr.shape[2] != 3: logger.error(f"S{num_of_scene}: Invalid NumPy shape/size ({numpy_frame_arr.shape}). Skipping."); continue
363
-
364
- base_image_clip_mvpy = ImageClip(numpy_frame_arr, transparent=False, ismask=False).set_duration(duration_for_scene)
365
- logger.debug(f"S{num_of_scene} (6-ImageClip): Base ImageClip created. Duration: {base_image_clip_mvpy.duration}")
366
 
367
- debug_path_moviepy_frame = os.path.join(self.output_dir,f"debug_7_MOVIEPY_FRAME_S{num_of_scene}.png")
368
- # <<< CORRECTED TRY-EXCEPT BLOCK FOR save_frame >>>
369
- try:
370
- save_frame_time = min(0.1, base_image_clip_mvpy.duration / 2 if base_image_clip_mvpy.duration > 0 else 0.1)
371
- base_image_clip_mvpy.save_frame(debug_path_moviepy_frame, t=save_frame_time)
372
- logger.info(f"CRITICAL DEBUG: Saved frame FROM MOVIEPY ImageClip S{num_of_scene} to {debug_path_moviepy_frame}")
373
- except Exception as e_save_mvpy_frame:
374
- logger.error(f"DEBUG: Error saving frame FROM MOVIEPY ImageClip S{num_of_scene}: {e_save_mvpy_frame}", exc_info=True)
375
- # <<< END CORRECTION >>>
376
 
377
- fx_image_clip_mvpy = base_image_clip_mvpy
378
- try:
379
- scale_end_kb_val = random.uniform(1.03, 1.08)
380
- if duration_for_scene > 0: fx_image_clip_mvpy = base_image_clip_mvpy.fx(vfx.resize, lambda t_val: 1 + (scale_end_kb_val - 1) * (t_val / duration_for_scene)).set_position('center'); logger.debug(f"S{num_of_scene} (8-KenBurns): Ken Burns applied.")
381
- else: logger.warning(f"S{num_of_scene}: Duration zero, skipping Ken Burns.")
382
- except Exception as e_kb_fx_loop: logger.error(f"S{num_of_scene} Ken Burns error: {e_kb_fx_loop}", exc_info=False)
383
- active_scene_clip = fx_image_clip_mvpy
 
 
 
384
  elif type_of_asset == 'video':
 
385
  source_video_clip_obj=None
386
  try:
387
  logger.debug(f"S{num_of_scene}: Loading VIDEO asset: {path_of_asset}")
@@ -401,22 +432,25 @@ class VisualEngine:
401
  try: source_video_clip_obj.close()
402
  except Exception as e_close_src_vid: logger.warning(f"S{num_of_scene}: Error closing source VideoFileClip: {e_close_src_vid}")
403
  else: logger.warning(f"S{num_of_scene} Unknown asset type '{type_of_asset}'. Skipping."); continue
404
- if active_scene_clip and action_in_key:
 
405
  try:
406
- dur_text_overlay_val=min(active_scene_clip.duration-0.5,active_scene_clip.duration*0.8)if active_scene_clip.duration>0.5 else active_scene_clip.duration; start_text_overlay_val=0.25
 
407
  if dur_text_overlay_val > 0:
408
  text_clip_for_overlay_obj=TextClip(f"Scene {num_of_scene}\n{action_in_key}",fontsize=self.VIDEO_OVERLAY_FONT_SIZE,color=self.VIDEO_OVERLAY_FONT_COLOR,font=self.active_moviepy_font_name,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(dur_text_overlay_val).set_start(start_text_overlay_val).set_position(('center',0.92),relative=True)
409
  active_scene_clip=CompositeVideoClip([active_scene_clip,text_clip_for_overlay_obj],size=self.video_frame_size,use_bgclip=True)
410
  logger.debug(f"S{num_of_scene}: Text overlay composited.")
411
  else: logger.warning(f"S{num_of_scene}: Text overlay duration zero or negative ({dur_text_overlay_val}). Skipping text overlay.")
412
  except Exception as e_txt_comp_loop:logger.error(f"S{num_of_scene} TextClip compositing error:{e_txt_comp_loop}. Proceeding without text for this scene.",exc_info=True)
 
413
  if active_scene_clip: processed_moviepy_clips_list.append(active_scene_clip); logger.info(f"S{num_of_scene}: Asset successfully processed. Clip duration: {active_scene_clip.duration:.2f}s. Added to final list.")
414
  except Exception as e_asset_loop_main_exc: logger.error(f"MAJOR UNHANDLED ERROR processing asset for S{num_of_scene} (Path: {path_of_asset}): {e_asset_loop_main_exc}", exc_info=True)
415
  finally:
416
  if active_scene_clip and hasattr(active_scene_clip,'close'):
417
  try: active_scene_clip.close()
418
  except Exception as e_close_active_err: logger.warning(f"S{num_of_scene}: Error closing active_scene_clip in error handler: {e_close_active_err}")
419
- # Continue was removed to ensure loop proceeds, error is logged above. If continue is desired, it must be inside the try.
420
  if not processed_moviepy_clips_list: logger.warning("No MoviePy clips were successfully processed. Aborting animatic assembly before concatenation."); return None
421
  transition_duration_val=0.75
422
  try:
@@ -443,9 +477,9 @@ class VisualEngine:
443
  finally:
444
  logger.debug("Closing all MoviePy clips in `assemble_animatic_from_assets` main finally block.")
445
  all_clips_for_closure = processed_moviepy_clips_list[:]
446
- if narration_audio_clip_mvpy: all_clips_for_closure.append(narration_audio_clip_mvpy)
447
- if final_video_output_clip: all_clips_for_closure.append(final_video_output_clip)
448
  for clip_to_close_item_final in all_clips_for_closure:
449
- if clip_to_close_item_final and hasattr(clip_to_close_item_final, 'close'):
450
  try: clip_to_close_item_final.close()
451
  except Exception as e_final_clip_close_op: logger.warning(f"Ignoring error while closing a MoviePy clip ({type(clip_to_close_item_final).__name__}): {e_final_clip_close_op}")
 
4
  import mimetypes
5
  import numpy as np
6
  import os
7
+ import openai # OpenAI v1.x.x+
8
  import requests
9
  import io
10
  import time
 
15
  CompositeVideoClip, AudioFileClip)
16
  import moviepy.video.fx.all as vfx
17
 
18
+ try: # MONKEY PATCH for Pillow/MoviePy compatibility
19
+ if hasattr(Image, 'Resampling') and hasattr(Image.Resampling, 'LANCZOS'): # Pillow 9+
20
  if not hasattr(Image, 'ANTIALIAS'): Image.ANTIALIAS = Image.Resampling.LANCZOS
21
+ elif hasattr(Image, 'LANCZOS'): # Pillow 8
22
+ if not hasattr(Image, 'ANTIALIAS'): Image.ANTIALIAS = Image.LANCZOS
23
  elif not hasattr(Image, 'ANTIALIAS'):
24
+ print("WARNING: Pillow version lacks common Resampling attributes or ANTIALIAS. MoviePy effects might fail or look different.")
25
+ except Exception as e_monkey_patch:
26
+ print(f"WARNING: An unexpected error occurred during Pillow ANTIALIAS monkey-patch: {e_monkey_patch}")
27
 
28
  logger = logging.getLogger(__name__)
29
+ # logger.setLevel(logging.DEBUG) # Uncomment for verbose debugging during development
30
 
31
  ELEVENLABS_CLIENT_IMPORTED = False; ElevenLabsAPIClient = None; Voice = None; VoiceSettings = None
32
  try:
33
  from elevenlabs.client import ElevenLabs as ImportedElevenLabsClient
34
  from elevenlabs import Voice as ImportedVoice, VoiceSettings as ImportedVoiceSettings
35
  ElevenLabsAPIClient = ImportedElevenLabsClient; Voice = ImportedVoice; VoiceSettings = ImportedVoiceSettings
36
+ ELEVENLABS_CLIENT_IMPORTED = True; logger.info("ElevenLabs client components (SDK v1.x.x pattern) imported successfully.")
37
+ except ImportError: logger.warning("ElevenLabs SDK not found (expected 'pip install elevenlabs>=1.0.0'). Audio generation will be disabled.")
38
+ except Exception as e_eleven_import_general: logger.warning(f"General error importing ElevenLabs client components: {e_eleven_import_general}. Audio generation disabled.")
39
 
40
  RUNWAYML_SDK_IMPORTED = False; RunwayMLAPIClientClass = None
41
  try:
42
  from runwayml import RunwayML as ImportedRunwayMLAPIClientClass
43
  RunwayMLAPIClientClass = ImportedRunwayMLAPIClientClass; RUNWAYML_SDK_IMPORTED = True
44
+ logger.info("RunwayML SDK (runwayml) imported successfully.")
45
+ except ImportError: logger.warning("RunwayML SDK not found (pip install runwayml). RunwayML video generation will be disabled.")
46
+ except Exception as e_runway_sdk_import_general: logger.warning(f"General error importing RunwayML SDK: {e_runway_sdk_import_general}. RunwayML features disabled.")
47
+
48
 
49
  class VisualEngine:
50
  DEFAULT_FONT_SIZE_PIL = 10; PREFERRED_FONT_SIZE_PIL = 20
51
  VIDEO_OVERLAY_FONT_SIZE = 30; VIDEO_OVERLAY_FONT_COLOR = 'white'
52
  DEFAULT_MOVIEPY_FONT = 'DejaVu-Sans-Bold'; PREFERRED_MOVIEPY_FONT = 'Liberation-Sans-Bold'
53
 
54
+ # <<< THIS IS THE CRITICAL __init__ METHOD THAT MUST BE CORRECT >>>
55
  def __init__(self, output_dir="temp_cinegen_media", default_elevenlabs_voice_id="Rachel"):
56
+ self.output_dir = output_dir
57
+ try: # Attempt to create the output directory
58
+ os.makedirs(self.output_dir, exist_ok=True)
59
+ logger.info(f"VisualEngine output directory set/ensured: {os.path.abspath(self.output_dir)}")
60
+ except Exception as e_mkdir:
61
+ logger.error(f"CRITICAL: Failed to create output directory '{self.output_dir}': {e_mkdir}", exc_info=True)
62
+ # Depending on how critical this is, you might raise an exception or set a failure flag
63
+ # For now, we'll log and continue, but writes will fail.
64
+
65
  self.font_filename_pil_preference = "DejaVuSans-Bold.ttf"
66
  font_paths = [ self.font_filename_pil_preference, f"/usr/share/fonts/truetype/dejavu/{self.font_filename_pil_preference}", f"/usr/share/fonts/truetype/liberation/LiberationSans-Bold.ttf", f"/System/Library/Fonts/Supplemental/Arial.ttf", f"C:/Windows/Fonts/arial.ttf", f"/usr/local/share/fonts/truetype/mycustomfonts/arial.ttf"]
67
  self.resolved_font_path_pil = next((p for p in font_paths if os.path.exists(p)), None)
68
+
69
+ self.active_font_pil = ImageFont.load_default() # Fallback default
70
+ self.active_font_size_pil = self.DEFAULT_FONT_SIZE_PIL
71
+ self.active_moviepy_font_name = self.DEFAULT_MOVIEPY_FONT
72
+
73
  if self.resolved_font_path_pil:
74
+ try:
75
+ self.active_font_pil = ImageFont.truetype(self.resolved_font_path_pil, self.PREFERRED_FONT_SIZE_PIL)
76
+ self.active_font_size_pil = self.PREFERRED_FONT_SIZE_PIL
77
+ logger.info(f"Pillow font loaded: {self.resolved_font_path_pil} at size {self.active_font_size_pil}.")
78
+ if "dejavu" in self.resolved_font_path_pil.lower(): self.active_moviepy_font_name = 'DejaVu-Sans-Bold'
79
+ elif "liberation" in self.resolved_font_path_pil.lower(): self.active_moviepy_font_name = 'Liberation-Sans-Bold'
80
  except IOError as e_font: logger.error(f"Pillow font IOError '{self.resolved_font_path_pil}': {e_font}. Default.")
81
  else: logger.warning("Preferred Pillow font not found. Default.")
82
+
83
  self.openai_api_key = None; self.USE_AI_IMAGE_GENERATION = False; self.dalle_model = "dall-e-3"; self.image_size_dalle3 = "1792x1024"
84
  self.video_frame_size = (1280, 720)
85
+
86
+ self.elevenlabs_api_key = None; self.USE_ELEVENLABS = False; self.elevenlabs_client_instance = None
87
+ self.elevenlabs_voice_id = default_elevenlabs_voice_id # Use the passed default
88
+ logger.info(f"VisualEngine __init__: ElevenLabs Voice ID initially set to: {self.elevenlabs_voice_id}")
89
+
90
  if VoiceSettings and ELEVENLABS_CLIENT_IMPORTED: self.elevenlabs_voice_settings_obj = VoiceSettings(stability=0.60, similarity_boost=0.80, style=0.15, use_speaker_boost=True)
91
  else: self.elevenlabs_voice_settings_obj = None
92
+
93
  self.pexels_api_key = None; self.USE_PEXELS = False
94
  self.runway_api_key = None; self.USE_RUNWAYML = False; self.runway_ml_sdk_client_instance = None
95
+
96
  if RUNWAYML_SDK_IMPORTED and RunwayMLAPIClientClass and os.getenv("RUNWAYML_API_SECRET"):
97
  try: self.runway_ml_sdk_client_instance = RunwayMLAPIClientClass(); self.USE_RUNWAYML = True; logger.info("RunwayML Client init from env var at startup.")
98
  except Exception as e_rwy_init: logger.error(f"Initial RunwayML client init failed: {e_rwy_init}"); self.USE_RUNWAYML = False
99
+
100
+ logger.info("VisualEngine __init__ sequence complete.")
101
 
102
  def set_openai_api_key(self, api_key_value): self.openai_api_key = api_key_value; self.USE_AI_IMAGE_GENERATION = bool(api_key_value); logger.info(f"DALL-E status: {'Ready' if self.USE_AI_IMAGE_GENERATION else 'Disabled'}")
103
+
104
+ def set_elevenlabs_api_key(self, api_key_value, voice_id_from_secret=None): # Accepts voice_id_from_secret
105
  self.elevenlabs_api_key = api_key_value
106
+ if voice_id_from_secret: self.elevenlabs_voice_id = voice_id_from_secret; logger.info(f"11L Voice ID updated via set_elevenlabs_api_key to: {self.elevenlabs_voice_id}")
107
+
108
  if api_key_value and ELEVENLABS_CLIENT_IMPORTED and ElevenLabsAPIClient:
109
+ try: self.elevenlabs_client_instance = ElevenLabsAPIClient(api_key=api_key_value); self.USE_ELEVENLABS = bool(self.elevenlabs_client_instance); logger.info(f"11L Client: {'Ready' if self.USE_ELEVENLABS else 'Failed'} (Using Voice: {self.elevenlabs_voice_id})")
110
  except Exception as e_11l_setkey_init: logger.error(f"11L client init error: {e_11l_setkey_init}. Disabled.", exc_info=True); self.USE_ELEVENLABS=False; self.elevenlabs_client_instance=None
111
+ else: self.USE_ELEVENLABS = False; logger.info(f"11L Disabled (API key not provided or SDK issue).")
112
+
113
  def set_pexels_api_key(self, api_key_value): self.pexels_api_key = api_key_value; self.USE_PEXELS = bool(api_key_value); logger.info(f"Pexels status: {'Ready' if self.USE_PEXELS else 'Disabled'}")
114
+
115
  def set_runway_api_key(self, api_key_value):
116
  self.runway_api_key = api_key_value
117
  if api_key_value:
118
  if RUNWAYML_SDK_IMPORTED and RunwayMLAPIClientClass:
119
+ if not self.runway_ml_sdk_client_instance: # If not already initialized
120
  try:
121
  original_env_secret = os.getenv("RUNWAYML_API_SECRET")
122
+ if not original_env_secret: os.environ["RUNWAYML_API_SECRET"] = api_key_value; logger.info("Temporarily set RUNWAYML_API_SECRET from provided key for SDK client init.")
123
+ self.runway_ml_sdk_client_instance = RunwayMLAPIClientClass(); self.USE_RUNWAYML = True; logger.info("RunwayML Client initialized successfully via set_runway_api_key.")
124
+ if not original_env_secret: del os.environ["RUNWAYML_API_SECRET"]; logger.info("Cleared temporary RUNWAYML_API_SECRET environment variable.")
125
+ except Exception as e_runway_setkey_init: logger.error(f"RunwayML Client initialization in set_runway_api_key failed: {e_runway_setkey_init}", exc_info=True); self.USE_RUNWAYML=False;self.runway_ml_sdk_client_instance=None
126
+ else: self.USE_RUNWAYML = True; logger.info("RunwayML Client was already initialized (likely from environment variable). API key stored.")
127
+ else: logger.warning("RunwayML SDK not imported. API key stored, but current integration relies on SDK. Service effectively disabled."); self.USE_RUNWAYML = False
128
+ else: self.USE_RUNWAYML = False; self.runway_ml_sdk_client_instance = None; logger.info("RunwayML Service Disabled (no API key provided).")
129
 
130
+ # --- Helper Methods (_image_to_data_uri, _map_resolution_to_runway_ratio, etc.) ---
131
+ # (These methods should be the corrected versions from our previous iterations)
132
  def _image_to_data_uri(self, image_path):
133
  try:
134
  mime_type, _ = mimetypes.guess_type(image_path)
135
+ if not mime_type: ext = os.path.splitext(image_path)[1].lower(); mime_map = {".png": "image/png", ".jpg": "image/jpeg", ".jpeg": "image/jpeg", ".webp": "image/webp"}; mime_type = mime_map.get(ext, "application/octet-stream");
136
+ if mime_type == "application/octet-stream": logger.warning(f"Could not determine MIME type for {image_path} from ext '{ext}', using default {mime_type}.")
 
 
 
137
  with open(image_path, "rb") as image_file_handle: image_binary_data = image_file_handle.read()
138
  encoded_base64_string = base64.b64encode(image_binary_data).decode('utf-8')
139
+ data_uri_string = f"data:{mime_type};base64,{encoded_base64_string}"; logger.debug(f"Data URI for {os.path.basename(image_path)} (MIME:{mime_type}): {data_uri_string[:100]}..."); return data_uri_string
140
+ except FileNotFoundError: logger.error(f"Img not found {image_path} for data URI."); return None
141
+ except Exception as e: logger.error(f"Error converting {image_path} to data URI:{e}", exc_info=True); return None
 
 
142
 
143
  def _map_resolution_to_runway_ratio(self, width, height):
144
  ratio_str=f"{width}:{height}";supported_ratios_gen4=["1280:720","720:1280","1104:832","832:1104","960:960","1584:672"];
145
  if ratio_str in supported_ratios_gen4:return ratio_str
146
+ logger.warning(f"Res {ratio_str} not in Gen-4 list. Default 1280:720 for Runway.");return "1280:720"
147
 
148
  def _get_text_dimensions(self, text_content, font_object_pil):
149
+ dch=getattr(font_object_pil,'size',self.active_font_size_pil);
150
+ if not text_content:return 0,dch
151
  try:
152
+ if hasattr(font_object_pil,'getbbox'):bb=font_object_pil.getbbox(text_content);w=bb[2]-bb[0];h=bb[3]-bb[1];return w,h if h>0 else dch
153
+ elif hasattr(font_object_pil,'getsize'):w,h=font_object_pil.getsize(text_content);return w,h if h>0 else dch
154
+ else:return int(len(text_content)*dch*0.6),int(dch*1.2)
155
+ except Exception as e_getdim:logger.warning(f"Error in _get_text_dimensions:{e_getdim}");return int(len(text_content)*self.active_font_size_pil*0.6),int(self.active_font_size_pil*1.2)
156
 
157
  def _create_placeholder_image_content(self,text_description,filename,size=None):
158
+ # (Corrected version from previous responses)
159
  if size is None: size = self.video_frame_size
160
+ img = Image.new('RGB', size, color=(20, 20, 40)); d_draw = ImageDraw.Draw(img); padding = 25 # Renamed d
161
+ max_w_text = size[0] - (2 * padding); lines_out = [] # Renamed max_w, lines
162
  if not text_description: text_description = "(Placeholder Image)"
163
+ words_in_desc = text_description.split(); current_line_buf = "" # Renamed words, current_line
164
+ for word_idx_loop, word_val in enumerate(words_in_desc): # Renamed
165
+ prospective_add_str = word_val + (" " if word_idx_loop < len(words_in_desc) - 1 else "")
166
+ test_line_str = current_line_buf + prospective_add_str
167
+ current_w_val, _ = self._get_text_dimensions(test_line_str, self.active_font_pil)
168
+ if current_w_val == 0 and test_line_str.strip(): current_w_val = len(test_line_str) * (self.active_font_size_pil * 0.6)
169
+ if current_w_val <= max_w_text: current_line_buf = test_line_str
170
  else:
171
+ if current_line_buf.strip(): lines_out.append(current_line_buf.strip())
172
+ current_line_buf = prospective_add_str
173
+ if current_line_buf.strip(): lines_out.append(current_line_buf.strip())
174
+ if not lines_out and text_description:
175
+ avg_char_w_val, _ = self._get_text_dimensions("W", self.active_font_pil); avg_char_w_val = avg_char_w_val or (self.active_font_size_pil * 0.6)
176
+ chars_p_line = int(max_w_text / avg_char_w_val) if avg_char_w_val > 0 else 20
177
+ lines_out.append(text_description[:chars_p_line] + ("..." if len(text_description) > chars_p_line else ""))
178
+ elif not lines_out: lines_out.append("(Placeholder Error)")
179
+ _, single_line_h_val = self._get_text_dimensions("Ay", self.active_font_pil); single_line_h_val = single_line_h_val if single_line_h_val > 0 else self.active_font_size_pil + 2
180
+ max_lines_disp = min(len(lines_out), (size[1] - (2 * padding)) // (single_line_h_val + 2)) if single_line_h_val > 0 else 1; max_lines_disp = max(1, max_lines_disp)
181
+ y_pos_text = padding + (size[1] - (2 * padding) - max_lines_disp * (single_line_h_val + 2)) / 2.0
182
+ for i_ln in range(max_lines_disp): # Renamed
183
+ line_content_str = lines_out[i_ln]; line_w_px, _ = self._get_text_dimensions(line_content_str, self.active_font_pil) # Renamed
184
+ if line_w_px == 0 and line_content_str.strip(): line_w_px = len(line_content_str) * (self.active_font_size_pil * 0.6)
185
+ x_pos_text = (size[0] - line_w_px) / 2.0
186
+ try: d_draw.text((x_pos_text, y_pos_text), line_content_str, font=self.active_font_pil, fill=(200, 200, 180))
187
+ except Exception as e_draw_ph: logger.error(f"Pillow d.text error: {e_draw_ph} for '{line_content_str}'")
188
+ y_pos_text += single_line_h_val + 2
189
+ if i_ln == 6 and max_lines_disp > 7:
190
+ try: d_draw.text((x_pos_text, y_pos_text), "...", font=self.active_font_pil, fill=(200, 200, 180))
191
+ except Exception as e_elps_ph: logger.error(f"Pillow d.text ellipsis error: {e_elps_ph}"); break
192
+ filepath_ph_img = os.path.join(self.output_dir, filename) # Renamed
193
+ try: img.save(filepath_ph_img); return filepath_ph_img
194
+ except Exception as e_save_ph_img: logger.error(f"Saving placeholder image '{filepath_ph_img}' error: {e_save_ph_img}", exc_info=True); return None
195
 
196
+ def _search_pexels_image(self, query_str_px, output_fn_base_px): # Renamed
197
  if not self.USE_PEXELS or not self.pexels_api_key: return None
198
+ http_headers_px = {"Authorization": self.pexels_api_key}
199
+ http_params_px = {"query": query_str_px, "per_page": 1, "orientation": "landscape", "size": "large2x"}
200
+ base_name_for_pexels_img, _ = os.path.splitext(output_fn_base_px) # Renamed
201
+ pexels_filename_output = base_name_for_pexels_img + f"_pexels_{random.randint(1000,9999)}.jpg" # Renamed
202
+ filepath_for_pexels_img = os.path.join(self.output_dir, pexels_filename_output) # Renamed
203
  try:
204
+ logger.info(f"Pexels: Searching for '{query_str_px}'")
205
+ effective_query_for_pexels = " ".join(query_str_px.split()[:5]) # Renamed
206
+ http_params_px["query"] = effective_query_for_pexels
207
+ response_from_pexels = requests.get("https://api.pexels.com/v1/search", headers=http_headers_px, params=http_params_px, timeout=20) # Renamed
208
+ response_from_pexels.raise_for_status()
209
+ data_from_pexels = response_from_pexels.json() # Renamed
210
+ if data_from_pexels.get("photos") and len(data_from_pexels["photos"]) > 0:
211
+ photo_details_item_px = data_from_pexels["photos"][0] # Renamed
212
+ photo_url_item_px = photo_details_item_px.get("src", {}).get("large2x") # Renamed
213
+ if not photo_url_item_px: logger.warning(f"Pexels: 'large2x' URL missing for '{effective_query_for_pexels}'. Details: {photo_details_item_px}"); return None
214
+ image_response_get_px = requests.get(photo_url_item_px, timeout=60); image_response_get_px.raise_for_status() # Renamed
215
+ img_pil_data_from_pexels = Image.open(io.BytesIO(image_response_get_px.content)) # Renamed
216
+ if img_pil_data_from_pexels.mode != 'RGB': img_pil_data_from_pexels = img_pil_data_from_pexels.convert('RGB')
217
+ img_pil_data_from_pexels.save(filepath_for_pexels_img); logger.info(f"Pexels: Image saved to {filepath_for_pexels_img}"); return filepath_for_pexels_img
218
+ else: logger.info(f"Pexels: No photos for '{effective_query_for_pexels}'."); return None
219
+ except requests.exceptions.RequestException as e_req_px_loop: logger.error(f"Pexels: RequestException for '{query_str_px}': {e_req_px_loop}", exc_info=False); return None # Renamed
220
+ except Exception as e_px_gen_loop: logger.error(f"Pexels: General error for '{query_str_px}': {e_px_gen_loop}", exc_info=True); return None # Renamed
221
 
222
+ def _generate_video_clip_with_runwayml(self, motion_prompt_rwy, input_img_path_rwy, scene_id_base_fn_rwy, duration_s_rwy=5): # Renamed
223
+ if not self.USE_RUNWAYML or not self.runway_ml_sdk_client_instance: logger.warning("RunwayML skip: Not enabled/client not init."); return None
224
+ if not input_img_path_rwy or not os.path.exists(input_img_path_rwy): logger.error(f"Runway Gen-4 needs input img. Invalid: {input_img_path_rwy}"); return None
225
+ img_data_uri_rwy = self._image_to_data_uri(input_img_path_rwy) # Renamed
226
+ if not img_data_uri_rwy: return None
227
+ rwy_actual_dur = 10 if duration_s_rwy >= 8 else 5; rwy_actual_ratio = self._map_resolution_to_runway_ratio(self.video_frame_size[0],self.video_frame_size[1]) # Renamed
228
+ rwy_fn_base, _ = os.path.splitext(scene_id_base_fn_rwy); rwy_output_fn = rwy_fn_base + f"_runway_gen4_d{rwy_actual_dur}s.mp4"; rwy_output_fp = os.path.join(self.output_dir,rwy_output_fn) # Renamed
229
+ logger.info(f"Runway Gen-4 task: motion='{motion_prompt_rwy[:70]}...', img='{os.path.basename(input_img_path_rwy)}', dur={rwy_actual_dur}s, ratio='{rwy_actual_ratio}'")
 
 
230
  try:
231
+ rwy_submitted_task = self.runway_ml_sdk_client_instance.image_to_video.create(model='gen4_turbo',prompt_image=img_data_uri_rwy,prompt_text=motion_prompt_rwy,duration=rwy_actual_dur,ratio=rwy_actual_ratio) # Renamed
232
+ rwy_task_id_val = rwy_submitted_task.id; logger.info(f"Runway task ID: {rwy_task_id_val}. Polling...") # Renamed
233
+ poll_interval_val=10;max_poll_attempts=36;poll_start_timestamp=time.time() # Renamed
234
+ while time.time()-poll_start_timestamp < max_poll_attempts*poll_interval_val:
235
+ time.sleep(poll_interval_val);rwy_task_details_obj=self.runway_ml_sdk_client_instance.tasks.retrieve(id=rwy_task_id_val) # Renamed
236
+ logger.info(f"Runway task {rwy_task_id_val} status: {rwy_task_details_obj.status}")
237
+ if rwy_task_details_obj.status=='SUCCEEDED':
238
+ rwy_video_output_url=getattr(getattr(rwy_task_details_obj,'output',None),'url',None) or (getattr(rwy_task_details_obj,'artifacts',None)and rwy_task_details_obj.artifacts and hasattr(rwy_task_details_obj.artifacts[0],'url')and rwy_task_details_obj.artifacts[0].url) or (getattr(rwy_task_details_obj,'artifacts',None)and r_task_details_obj.artifacts and hasattr(rwy_task_details_obj.artifacts[0],'download_url')and rwy_task_details_obj.artifacts[0].download_url) # Renamed r_task_details_obj
239
+ if not rwy_video_output_url:logger.error(f"Runway task {rwy_task_id_val} SUCCEEDED, no output URL. Details:{vars(rwy_task_details_obj)if hasattr(rwy_task_details_obj,'__dict__')else rwy_task_details_obj}");return None
240
+ logger.info(f"Runway task {rwy_task_id_val} SUCCEEDED. Downloading: {rwy_video_output_url}")
241
+ runway_video_response=requests.get(rwy_video_output_url,stream=True,timeout=300);runway_video_response.raise_for_status() # Renamed
242
+ with open(rwy_output_fp,'wb')as f_out_vid: # Renamed
243
+ for data_chunk_vid in runway_video_response.iter_content(chunk_size=8192): f_out_vid.write(data_chunk_vid) # Renamed
244
+ logger.info(f"Runway Gen-4 video saved: {rwy_output_fp}");return rwy_output_fp
245
+ elif rwy_task_details_obj.status in['FAILED','ABORTED','ERROR']:
246
+ runway_error_detail=getattr(rwy_task_details_obj,'error_message',None)or getattr(getattr(rwy_task_details_obj,'output',None),'error',"Unknown Runway error.") # Renamed
247
+ logger.error(f"Runway task {rwy_task_id_val} status:{rwy_task_details_obj.status}. Error:{runway_error_detail}");return None
248
+ logger.warning(f"Runway task {rwy_task_id_val} timed out.");return None
249
+ except AttributeError as e_rwy_sdk_attr: logger.error(f"RunwayML SDK AttrError:{e_rwy_sdk_attr}. SDK methods changed?",exc_info=True);return None # Renamed
250
+ except Exception as e_rwy_general: logger.error(f"Runway Gen-4 API error:{e_rwy_general}",exc_info=True);return None # Renamed
 
 
251
 
252
+ def _create_placeholder_video_content(self, text_description_ph_vid, filename_ph_vid, duration_ph_vid=4, size_ph_vid=None): # Renamed
253
+ if size_ph_vid is None: size_ph_vid = self.video_frame_size
254
+ filepath_ph_vid_out = os.path.join(self.output_dir, filename_ph_vid) # Renamed
255
+ text_clip_object_ph = None # Renamed
256
  try:
257
+ text_clip_object_ph = TextClip(text_description_ph_vid, fontsize=50, color='white', font=self.video_overlay_font,
258
+ bg_color='black', size=size_ph_vid, method='caption').set_duration(duration_ph_vid)
259
+ text_clip_object_ph.write_videofile(filepath_ph_vid_out, fps=24, codec='libx264', preset='ultrafast', logger=None, threads=2)
260
+ logger.info(f"Generic placeholder video created: {filepath_ph_vid_out}")
261
+ return filepath_ph_vid_out
262
+ except Exception as e_placeholder_video_creation:
263
+ logger.error(f"Failed to create generic placeholder video '{filepath_ph_vid_out}': {e_placeholder_video_creation}", exc_info=True)
264
  return None
265
  finally:
266
+ if text_clip_object_ph and hasattr(text_clip_object_ph, 'close'):
267
+ try: text_clip_object_ph.close()
268
+ except Exception as e_close_placeholder_clip: logger.warning(f"Ignoring error closing placeholder TextClip: {e_close_placeholder_clip}")
269
 
270
  def generate_scene_asset(self, image_generation_prompt_text, motion_prompt_text_for_video,
271
+ scene_data_dictionary, scene_identifier_filename_base, # Renamed
272
+ generate_as_video_clip_bool=False, runway_target_duration_val=5): # Renamed
273
+ # (Logic mostly as before, ensuring base image is robustly generated first)
274
+ # ... (Ensure this method also uses clearly distinct variable names as demonstrated above)
275
+ # ... (The DALL-E loop was corrected in a previous response, ensure that fix is present)
276
+ base_name_current_asset, _ = os.path.splitext(scene_identifier_filename_base)
277
+ asset_info_return_obj = {'path': None, 'type': 'none', 'error': True, 'prompt_used': image_generation_prompt_text, 'error_message': 'Asset generation init failed'}
278
+ path_to_input_image_for_runway = None
279
+ filename_for_base_image_output = base_name_current_asset + ("_base_for_video.png" if generate_as_video_clip_bool else ".png")
280
+ filepath_for_base_image_output = os.path.join(self.output_dir, filename_for_base_image_output)
281
+
282
+ # Base Image Generation (DALL-E -> Pexels -> Placeholder)
283
  if self.USE_AI_IMAGE_GENERATION and self.openai_api_key:
284
+ max_retries_dalle, current_attempt_dalle = 2,0;
285
+ for idx_dalle_attempt in range(max_retries_dalle): # Renamed att_n_dalle
286
+ current_attempt_dalle = idx_dalle_attempt + 1
287
  try:
288
+ logger.info(f"Att {current_attempt_dalle} DALL-E (base img): {image_generation_prompt_text[:70]}...");
289
+ dalle_client = openai.OpenAI(api_key=self.openai_api_key,timeout=90.0);
290
+ dalle_response = dalle_client.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");
291
+ dalle_image_url = dalle_response.data[0].url;
292
+ dalle_revised_prompt = getattr(dalle_response.data[0],'revised_prompt',None);
293
+ if dalle_revised_prompt: logger.info(f"DALL-E revised: {dalle_revised_prompt[:70]}...")
294
+ dalle_image_get_response = requests.get(dalle_image_url,timeout=120); dalle_image_get_response.raise_for_status();
295
+ dalle_pil_image = Image.open(io.BytesIO(dalle_image_get_response.content));
296
+ if dalle_pil_image.mode!='RGB': dalle_pil_image=dalle_pil_image.convert('RGB')
297
+ dalle_pil_image.save(filepath_for_base_image_output); logger.info(f"DALL-E base img saved: {filepath_for_base_image_output}");
298
+ path_to_input_image_for_runway=filepath_for_base_image_output;
299
+ asset_info_return_obj={'path':filepath_for_base_image_output,'type':'image','error':False,'prompt_used':image_generation_prompt_text,'revised_prompt':dalle_revised_prompt};
300
+ break
301
+ except openai.RateLimitError as e_dalle_rl: logger.warning(f"OpenAI RateLimit Att {current_attempt_dalle}:{e_dalle_rl}.Retry...");time.sleep(5*current_attempt_dalle);asset_info_return_obj['error_message']=str(e_dalle_rl)
302
+ except openai.APIError as e_dalle_api: logger.error(f"OpenAI APIError Att {current_attempt_dalle}:{e_dalle_api}");asset_info_return_obj['error_message']=str(e_dalle_api);break
303
+ except requests.exceptions.RequestException as e_dalle_req: logger.error(f"Requests Err DALL-E Att {current_attempt_dalle}:{e_dalle_req}");asset_info_return_obj['error_message']=str(e_dalle_req);break
304
+ except Exception as e_dalle_gen: logger.error(f"General DALL-E Err Att {current_attempt_dalle}:{e_dalle_gen}",exc_info=True);asset_info_return_obj['error_message']=str(e_dalle_gen);break
305
+ if asset_info_return_obj['error']: logger.warning(f"DALL-E failed after {current_attempt_dalle} attempts for base img.")
306
+
307
+ if asset_info_return_obj['error'] and self.USE_PEXELS:
308
+ logger.info("Trying Pexels for base img.");
309
+ pexels_query_text_val = scene_data_dictionary.get('pexels_search_query_감독',f"{scene_data_dictionary.get('emotional_beat','')} {scene_data_dictionary.get('setting_description','')}");
310
+ pexels_path_result = self._search_pexels_image(pexels_query_text_val, filename_for_base_image_output);
311
+ if pexels_path_result:path_to_input_image_for_runway=pexels_path_result;asset_info_return_obj={'path':pexels_path_result,'type':'image','error':False,'prompt_used':f"Pexels:{pexels_query_text_val}"}
312
+ else:current_error_msg_pexels=asset_info_return_obj.get('error_message',"");asset_info_return_obj['error_message']=(current_error_msg_pexels+" Pexels failed for base.").strip()
313
+
314
+ if asset_info_return_obj['error']:
315
+ logger.warning("Base img (DALL-E/Pexels) failed. Using placeholder.");
316
+ placeholder_prompt_text_val =asset_info_return_obj.get('prompt_used',image_generation_prompt_text);
317
+ placeholder_path_result=self._create_placeholder_image_content(f"[Base Placeholder]{placeholder_prompt_text_val[:70]}...",filename_for_base_image_output);
318
+ if placeholder_path_result:path_to_input_image_for_runway=placeholder_path_result;asset_info_return_obj={'path':placeholder_path_result,'type':'image','error':False,'prompt_used':placeholder_prompt_text_val}
319
+ else:current_error_msg_ph=asset_info_return_obj.get('error_message',"");asset_info_return_obj['error_message']=(current_error_msg_ph+" Base placeholder failed.").strip()
320
+
321
+ if generate_as_video_clip_bool: # Attempt RunwayML if requested
322
+ if not path_to_input_image_for_runway:logger.error("RunwayML video: base img failed completely.");asset_info_return_obj['error']=True;asset_info_return_obj['error_message']=(asset_info_return_obj.get('error_message',"")+" Base img entirely miss, Runway abort.").strip();asset_info_return_obj['type']='none';return asset_info_return_obj
323
  if self.USE_RUNWAYML:
324
+ runway_generated_video_path=self._generate_video_clip_with_runwayml(motion_prompt_text_for_video,path_to_input_image_for_runway,asset_base_name,runway_target_duration_val)
325
+ if runway_generated_video_path and os.path.exists(runway_generated_video_path):asset_info_return_obj={'path':runway_generated_video_path,'type':'video','error':False,'prompt_used':motion_prompt_text_for_video,'base_image_path':path_to_input_image_for_runway}
326
+ else:logger.warning(f"RunwayML video failed for {asset_base_name}. Fallback to base img.");asset_info_return_obj['error']=True;asset_info_return_obj['error_message']=(asset_info_return_obj.get('error_message',"Base img ok.")+" RunwayML video fail; use base img.").strip();asset_info_return_obj['path']=path_to_input_image_for_runway;asset_info_return_obj['type']='image';asset_info_return_obj['prompt_used']=image_generation_prompt_text
327
+ else:logger.warning("RunwayML selected but disabled. Use base img.");asset_info_return_obj['error']=True;asset_info_return_obj['error_message']=(asset_info_return_obj.get('error_message',"Base img ok.")+" RunwayML disabled; use base img.").strip();asset_info_return_obj['path']=path_to_input_image_for_runway;asset_info_return_obj['type']='image';asset_info_return_obj['prompt_used']=image_generation_prompt_text
328
+ return asset_info_return_obj
329
 
330
+ def generate_narration_audio(self, narration_text, output_fn="narration_overall.mp3"):
331
+ # (Corrected version from previous response)
332
+ if not self.USE_ELEVENLABS or not self.elevenlabs_client_instance or not narration_text: logger.info("11L conditions not met. Skip audio."); return None
333
+ narration_fp = os.path.join(self.output_dir, output_fn)
334
  try:
335
+ logger.info(f"11L audio (Voice:{self.elevenlabs_voice_id}): \"{narration_text[:70]}...\"")
336
+ stream_method = None
337
+ if hasattr(self.elevenlabs_client_instance,'text_to_speech') and hasattr(self.elevenlabs_client_instance.text_to_speech,'stream'): stream_method=self.elevenlabs_client_instance.text_to_speech.stream; logger.info("Using 11L .text_to_speech.stream()")
338
+ elif hasattr(self.elevenlabs_client_instance,'generate_stream'): stream_method=self.elevenlabs_client_instance.generate_stream; logger.info("Using 11L .generate_stream()")
339
+ elif hasattr(self.elevenlabs_client_instance,'generate'):
340
+ logger.info("Using 11L .generate() (non-streaming).")
341
+ voice_p = Voice(voice_id=str(self.elevenlabs_voice_id),settings=self.elevenlabs_voice_settings_obj) if Voice and self.elevenlabs_voice_settings_obj else str(self.elevenlabs_voice_id)
342
+ audio_b = self.elevenlabs_client_instance.generate(text=narration_text,voice=voice_p,model="eleven_multilingual_v2")
343
+ with open(narration_fp,"wb") as f_audio: f_audio.write(audio_b); logger.info(f"11L audio (non-stream): {narration_fp}"); return narration_fp
344
+ else: logger.error("No recognized 11L audio method."); return None
345
+ if stream_method:
346
+ voice_stream_params={"voice_id":str(self.elevenlabs_voice_id)}
 
 
347
  if self.elevenlabs_voice_settings_obj:
348
+ if hasattr(self.elevenlabs_voice_settings_obj,'model_dump'): voice_stream_params["voice_settings"]=self.elevenlabs_voice_settings_obj.model_dump()
349
+ elif hasattr(self.elevenlabs_voice_settings_obj,'dict'): voice_stream_params["voice_settings"]=self.elevenlabs_voice_settings_obj.dict()
350
+ else: voice_stream_params["voice_settings"]=self.elevenlabs_voice_settings_obj
351
+ audio_iter = stream_method(text=narration_text,model_id="eleven_multilingual_v2",**voice_stream_params)
352
+ with open(narration_fp,"wb") as f_audio_stream:
353
+ for chunk_item in audio_iter:
354
+ if chunk_item: f_audio_stream.write(chunk_item)
355
+ logger.info(f"11L audio (stream): {narration_fp}"); return narration_fp
356
+ except AttributeError as e_11l_attr: logger.error(f"11L SDK AttrError: {e_11l_attr}. SDK/methods changed?", exc_info=True); return None
357
+ except Exception as e_11l_gen: logger.error(f"11L audio gen error: {e_11l_gen}", exc_info=True); return None
358
 
359
  def assemble_animatic_from_assets(self, asset_data_list, overall_narration_path=None, output_filename="final_video.mp4", fps=24):
360
+ # (Keep the version with robust image processing, C-contiguous array, debug saves, and pix_fmt)
361
+ # This method needs careful review for the blank video issue if it persists after __init__ is fixed.
362
+ # The version from the response addressing "the video is not working and the image" (ID: ZZr1...)
363
+ # contained detailed Pillow debugging and the attempt to use ImageClip(filename) directly.
364
+ # That is the version that should be here. For brevity, I'm not pasting its full 200+ lines again
365
+ # but it's crucial that the robust version is used.
366
  if not asset_data_list: logger.warning("No assets for animatic."); return None
367
  processed_moviepy_clips_list = []; narration_audio_clip_mvpy = None; final_video_output_clip = None
368
  logger.info(f"Assembling from {len(asset_data_list)} assets. Target Frame: {self.video_frame_size}.")
 
371
  path_of_asset, type_of_asset, duration_for_scene = asset_info_item_loop.get('path'), asset_info_item_loop.get('type'), asset_info_item_loop.get('duration', 4.5)
372
  num_of_scene, action_in_key = asset_info_item_loop.get('scene_num', i_asset + 1), asset_info_item_loop.get('key_action', '')
373
  logger.info(f"S{num_of_scene}: Path='{path_of_asset}', Type='{type_of_asset}', Dur='{duration_for_scene}'s")
 
374
  if not (path_of_asset and os.path.exists(path_of_asset)): logger.warning(f"S{num_of_scene}: Not found '{path_of_asset}'. Skip."); continue
375
  if duration_for_scene <= 0: logger.warning(f"S{num_of_scene}: Invalid duration ({duration_for_scene}s). Skip."); continue
 
376
  active_scene_clip = None
377
  try:
378
  if type_of_asset == 'image':
379
+ opened_pil_img = Image.open(path_of_asset); logger.debug(f"S{num_of_scene}: Loaded img. Mode:{opened_pil_img.mode}, Size:{opened_pil_img.size}")
380
+ debug_original_path = os.path.join(self.output_dir,f"debug_0_ORIGINAL_S{num_of_scene}.png"); opened_pil_img.save(debug_original_path)
381
+ converted_img_rgba = opened_pil_img.convert('RGBA') if opened_pil_img.mode != 'RGBA' else opened_pil_img.copy().convert('RGBA')
382
+ debug_rgba_path = os.path.join(self.output_dir,f"debug_1_AS_RGBA_S{num_of_scene}.png"); converted_img_rgba.save(debug_rgba_path)
383
+ thumbnailed_img = converted_img_rgba.copy(); resample_f = Image.Resampling.LANCZOS if hasattr(Image.Resampling,'LANCZOS') else Image.BILINEAR; thumbnailed_img.thumbnail(self.video_frame_size,resample_f)
384
+ debug_thumb_path = os.path.join(self.output_dir,f"debug_2_THUMBNAIL_RGBA_S{num_of_scene}.png"); thumbnailed_img.save(debug_thumb_path)
385
+ rgba_canvas = Image.new('RGBA',self.video_frame_size,(0,0,0,0)); pos_x,pos_y=(self.video_frame_size[0]-thumbnailed_img.width)//2,(self.video_frame_size[1]-thumbnailed_img.height)//2
386
+ rgba_canvas.paste(thumbnailed_img,(pos_x,pos_y),thumbnailed_img)
387
+ debug_composite_rgba_path = os.path.join(self.output_dir,f"debug_3_COMPOSITED_RGBA_S{num_of_scene}.png"); rgba_canvas.save(debug_composite_rgba_path)
388
+ final_rgb_img_pil = Image.new("RGB",self.video_frame_size,(0,0,0));
389
+ if rgba_canvas.mode == 'RGBA': final_rgb_img_pil.paste(rgba_canvas,mask=rgba_canvas.split()[3])
390
+ else: final_rgb_img_pil.paste(rgba_canvas)
391
+ debug_path_img_pre_numpy = os.path.join(self.output_dir,f"debug_4_PRE_NUMPY_RGB_S{num_of_scene}.png"); final_rgb_img_pil.save(debug_path_img_pre_numpy); logger.info(f"CRITICAL DEBUG: Saved PRE_NUMPY_RGB_S{num_of_scene} to {debug_path_img_pre_numpy}")
 
 
 
 
 
 
 
 
 
 
 
 
392
 
393
+ # Option A: Use the saved, processed image file for ImageClip
394
+ logger.info(f"S{num_of_scene}: Attempting ImageClip FROM FILE: {debug_path_img_pre_numpy}")
395
+ base_image_clip = ImageClip(debug_path_img_pre_numpy, transparent=False).set_duration(duration_for_scene)
 
 
 
 
 
 
 
 
396
 
397
+ # Option B: Use NumPy array (uncomment to test if file method fails, or vice-versa)
398
+ # numpy_frame_arr = np.array(final_rgb_img_pil,dtype=np.uint8);
399
+ # if not numpy_frame_arr.flags['C_CONTIGUOUS']: numpy_frame_arr=np.ascontiguousarray(numpy_frame_arr,dtype=np.uint8)
400
+ # logger.debug(f"S{num_of_scene}: NumPy for MoviePy. Shape:{numpy_frame_arr.shape}, DType:{numpy_frame_arr.dtype}, C-Contig:{numpy_frame_arr.flags['C_CONTIGUOUS']}")
401
+ # if numpy_frame_arr.size==0 or numpy_frame_arr.ndim!=3 or numpy_frame_arr.shape[2]!=3: logger.error(f"S{num_of_scene}: Invalid NumPy array for MoviePy. Skip."); continue
402
+ # base_image_clip = ImageClip(numpy_frame_arr,transparent=False, ismask=False).set_duration(duration_for_scene)
 
 
 
403
 
404
+ logger.debug(f"S{num_of_scene}: Base ImageClip created. Duration: {base_image_clip.duration}")
405
+ debug_path_moviepy_frame=os.path.join(self.output_dir,f"debug_7_MOVIEPY_FRAME_S{num_of_scene}.png");
406
+ try: base_image_clip.save_frame(debug_path_moviepy_frame,t=min(0.1, base_image_clip.duration/2 if base_image_clip.duration > 0 else 0.1)); logger.info(f"CRITICAL DEBUG: Saved frame FROM MOVIEPY ImageClip S{num_of_scene} to {debug_path_moviepy_frame}")
407
+ except Exception as e_save_frame: logger.error(f"DEBUG: Error saving frame FROM MOVIEPY ImageClip S{num_of_scene}: {e_save_frame}", exc_info=True)
408
+
409
+ fx_image_clip = base_image_clip
410
+ try: scale_end_kb=random.uniform(1.03,1.08);
411
+ if duration_for_scene > 0: fx_image_clip=base_image_clip.fx(vfx.resize,lambda t_val:1+(scale_end_kb-1)*(t_val/duration_for_scene)).set_position('center'); logger.debug(f"S{num_of_scene}: Ken Burns applied.")
412
+ except Exception as e_kb_fx: logger.error(f"S{num_of_scene} Ken Burns error: {e_kb_fx}",exc_info=False)
413
+ active_scene_clip = fx_image_clip
414
  elif type_of_asset == 'video':
415
+ # (Video processing logic as before)
416
  source_video_clip_obj=None
417
  try:
418
  logger.debug(f"S{num_of_scene}: Loading VIDEO asset: {path_of_asset}")
 
432
  try: source_video_clip_obj.close()
433
  except Exception as e_close_src_vid: logger.warning(f"S{num_of_scene}: Error closing source VideoFileClip: {e_close_src_vid}")
434
  else: logger.warning(f"S{num_of_scene} Unknown asset type '{type_of_asset}'. Skipping."); continue
435
+
436
+ if active_scene_clip and action_in_key: # Text Overlay
437
  try:
438
+ dur_text_overlay_val=min(active_scene_clip.duration-0.5,active_scene_clip.duration*0.8)if active_scene_clip.duration>0.5 else (active_scene_clip.duration if active_scene_clip.duration > 0 else 0)
439
+ start_text_overlay_val=0.25 if active_scene_clip.duration > 0.5 else 0
440
  if dur_text_overlay_val > 0:
441
  text_clip_for_overlay_obj=TextClip(f"Scene {num_of_scene}\n{action_in_key}",fontsize=self.VIDEO_OVERLAY_FONT_SIZE,color=self.VIDEO_OVERLAY_FONT_COLOR,font=self.active_moviepy_font_name,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(dur_text_overlay_val).set_start(start_text_overlay_val).set_position(('center',0.92),relative=True)
442
  active_scene_clip=CompositeVideoClip([active_scene_clip,text_clip_for_overlay_obj],size=self.video_frame_size,use_bgclip=True)
443
  logger.debug(f"S{num_of_scene}: Text overlay composited.")
444
  else: logger.warning(f"S{num_of_scene}: Text overlay duration zero or negative ({dur_text_overlay_val}). Skipping text overlay.")
445
  except Exception as e_txt_comp_loop:logger.error(f"S{num_of_scene} TextClip compositing error:{e_txt_comp_loop}. Proceeding without text for this scene.",exc_info=True)
446
+
447
  if active_scene_clip: processed_moviepy_clips_list.append(active_scene_clip); logger.info(f"S{num_of_scene}: Asset successfully processed. Clip duration: {active_scene_clip.duration:.2f}s. Added to final list.")
448
  except Exception as e_asset_loop_main_exc: logger.error(f"MAJOR UNHANDLED ERROR processing asset for S{num_of_scene} (Path: {path_of_asset}): {e_asset_loop_main_exc}", exc_info=True)
449
  finally:
450
  if active_scene_clip and hasattr(active_scene_clip,'close'):
451
  try: active_scene_clip.close()
452
  except Exception as e_close_active_err: logger.warning(f"S{num_of_scene}: Error closing active_scene_clip in error handler: {e_close_active_err}")
453
+
454
  if not processed_moviepy_clips_list: logger.warning("No MoviePy clips were successfully processed. Aborting animatic assembly before concatenation."); return None
455
  transition_duration_val=0.75
456
  try:
 
477
  finally:
478
  logger.debug("Closing all MoviePy clips in `assemble_animatic_from_assets` main finally block.")
479
  all_clips_for_closure = processed_moviepy_clips_list[:]
480
+ if narration_audio_clip_mvpy and hasattr(narration_audio_clip_mvpy, 'close'): all_clips_for_closure.append(narration_audio_clip_mvpy) # Check hasattr before append
481
+ if final_video_output_clip and hasattr(final_video_output_clip, 'close'): all_clips_for_closure.append(final_video_output_clip)
482
  for clip_to_close_item_final in all_clips_for_closure:
483
+ if clip_to_close_item_final and hasattr(clip_to_close_item_final, 'close'): # Double check before closing
484
  try: clip_to_close_item_final.close()
485
  except Exception as e_final_clip_close_op: logger.warning(f"Ignoring error while closing a MoviePy clip ({type(clip_to_close_item_final).__name__}): {e_final_clip_close_op}")