aiqcamp commited on
Commit
f47b184
ยท
verified ยท
1 Parent(s): 6251d81

Create app-backup.py

Browse files
Files changed (1) hide show
  1. app-backup.py +473 -0
app-backup.py ADDED
@@ -0,0 +1,473 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # 1. ๋จผ์ € ๋กœ๊น… ์„ค์ •
2
+ import logging
3
+ logging.basicConfig(level=logging.INFO)
4
+ logger = logging.getLogger(__name__)
5
+
6
+ # 2. ๋‚˜๋จธ์ง€ imports
7
+ import os
8
+ import time
9
+ from datetime import datetime
10
+ import gradio as gr
11
+ import torch
12
+ import requests
13
+ from pathlib import Path
14
+ import cv2
15
+ from PIL import Image
16
+ import json
17
+ import spaces
18
+ import torchaudio
19
+ import tempfile
20
+
21
+ try:
22
+ import mmaudio
23
+ except ImportError:
24
+ os.system("pip install -e .")
25
+ import mmaudio
26
+
27
+ from mmaudio.eval_utils import (ModelConfig, all_model_cfg, generate, load_video, make_video,
28
+ setup_eval_logging)
29
+ from mmaudio.model.flow_matching import FlowMatching
30
+ from mmaudio.model.networks import MMAudio, get_my_mmaudio
31
+ from mmaudio.model.sequence_config import SequenceConfig
32
+ from mmaudio.model.utils.features_utils import FeaturesUtils
33
+ # ์ƒ๋‹จ์— ๋ฒˆ์—ญ ๋ชจ๋ธ import ์ถ”๊ฐ€
34
+ from transformers import pipeline
35
+ translator = pipeline("translation", model="Helsinki-NLP/opus-mt-ko-en")
36
+ # 3. API ์„ค์ •
37
+ CATBOX_USER_HASH = "30f52c895fd9d9cb387eee489"
38
+ REPLICATE_API_TOKEN = os.getenv("API_KEY")
39
+
40
+ # 4. ์˜ค๋””์˜ค ๋ชจ๋ธ ์„ค์ •
41
+ device = 'cuda'
42
+ dtype = torch.bfloat16
43
+
44
+ # 5. get_model ํ•จ์ˆ˜ ์ •์˜
45
+ def get_model() -> tuple[MMAudio, FeaturesUtils, SequenceConfig]:
46
+ seq_cfg = model.seq_cfg
47
+
48
+ net: MMAudio = get_my_mmaudio(model.model_name).to(device, dtype).eval()
49
+ net.load_weights(torch.load(model.model_path, map_location=device, weights_only=True))
50
+ logger.info(f'Loaded weights from {model.model_path}')
51
+
52
+ feature_utils = FeaturesUtils(tod_vae_ckpt=model.vae_path,
53
+ synchformer_ckpt=model.synchformer_ckpt,
54
+ enable_conditions=True,
55
+ mode=model.mode,
56
+ bigvgan_vocoder_ckpt=model.bigvgan_16k_path,
57
+ need_vae_encoder=False)
58
+ feature_utils = feature_utils.to(device, dtype).eval()
59
+
60
+ return net, feature_utils, seq_cfg
61
+
62
+ # 6. ๋ชจ๋ธ ์ดˆ๊ธฐํ™”
63
+ model: ModelConfig = all_model_cfg['large_44k_v2']
64
+ model.download_if_needed()
65
+ output_dir = Path('./output/gradio')
66
+
67
+ setup_eval_logging()
68
+ net, feature_utils, seq_cfg = get_model()
69
+
70
+ @spaces.GPU(duration=30) # 30์ดˆ๋กœ ์ œํ•œ
71
+ @torch.inference_mode()
72
+ def video_to_audio(video_path: str, prompt: str, negative_prompt: str = "music",
73
+ seed: int = -1, num_steps: int = 15,
74
+ cfg_strength: float = 4.0, target_duration: float = 4.0):
75
+ try:
76
+ logger.info("Starting audio generation process")
77
+ torch.cuda.empty_cache()
78
+
79
+ rng = torch.Generator(device=device)
80
+ if seed >= 0:
81
+ rng.manual_seed(seed)
82
+ else:
83
+ rng.seed()
84
+
85
+ fm = FlowMatching(min_sigma=0, inference_mode='euler', num_steps=num_steps)
86
+
87
+ # load_video ํ•จ์ˆ˜ ํ˜ธ์ถœ ์ˆ˜์ •
88
+ video_info = load_video(video_path, duration_sec=target_duration) # duration_sec ํŒŒ๋ผ๋ฏธํ„ฐ๋กœ ๋ณ€๊ฒฝ
89
+
90
+ if video_info is None:
91
+ logger.error("Failed to load video")
92
+ return video_path
93
+
94
+ clip_frames = video_info.clip_frames
95
+ sync_frames = video_info.sync_frames
96
+ actual_duration = video_info.duration_sec
97
+
98
+ if clip_frames is None or sync_frames is None:
99
+ logger.error("Failed to extract frames from video")
100
+ return video_path
101
+
102
+ # ๋ฉ”๋ชจ๋ฆฌ ์ตœ์ ํ™”
103
+ clip_frames = clip_frames[:int(actual_duration * video_info.fps)]
104
+ sync_frames = sync_frames[:int(actual_duration * video_info.fps)]
105
+
106
+ clip_frames = clip_frames.unsqueeze(0).to(device, dtype=torch.float16)
107
+ sync_frames = sync_frames.unsqueeze(0).to(device, dtype=torch.float16)
108
+
109
+ seq_cfg.duration = actual_duration
110
+ net.update_seq_lengths(seq_cfg.latent_seq_len, seq_cfg.clip_seq_len, seq_cfg.sync_seq_len)
111
+
112
+ logger.info("Generating audio...")
113
+ with torch.cuda.amp.autocast():
114
+ audios = generate(clip_frames,
115
+ sync_frames,
116
+ [prompt],
117
+ negative_text=[negative_prompt],
118
+ feature_utils=feature_utils,
119
+ net=net,
120
+ fm=fm,
121
+ rng=rng,
122
+ cfg_strength=cfg_strength)
123
+
124
+ if audios is None:
125
+ logger.error("Failed to generate audio")
126
+ return video_path
127
+
128
+ audio = audios.float().cpu()[0]
129
+
130
+ output_path = tempfile.NamedTemporaryFile(delete=False, suffix='.mp4').name
131
+ logger.info(f"Creating final video with audio at {output_path}")
132
+
133
+ make_video(video_info, output_path, audio, sampling_rate=seq_cfg.sampling_rate)
134
+
135
+ torch.cuda.empty_cache()
136
+
137
+ if not os.path.exists(output_path):
138
+ logger.error("Failed to create output video")
139
+ return video_path
140
+
141
+ logger.info(f'Successfully saved video with audio to {output_path}')
142
+ return output_path
143
+
144
+ except Exception as e:
145
+ logger.error(f"Error in video_to_audio: {str(e)}")
146
+ torch.cuda.empty_cache()
147
+ return video_path
148
+
149
+ def upload_to_catbox(file_path):
150
+ """catbox.moe API๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ํŒŒ์ผ ์—…๋กœ๋“œ"""
151
+ try:
152
+ logger.info(f"Preparing to upload file: {file_path}")
153
+ url = "https://catbox.moe/user/api.php"
154
+
155
+ mime_types = {
156
+ '.jpg': 'image/jpeg',
157
+ '.jpeg': 'image/jpeg',
158
+ '.png': 'image/png',
159
+ '.gif': 'image/gif',
160
+ '.webp': 'image/webp',
161
+ '.jfif': 'image/jpeg'
162
+ }
163
+
164
+ file_extension = Path(file_path).suffix.lower()
165
+
166
+ if file_extension not in mime_types:
167
+ try:
168
+ img = Image.open(file_path)
169
+ if img.mode != 'RGB':
170
+ img = img.convert('RGB')
171
+
172
+ new_path = file_path.rsplit('.', 1)[0] + '.png'
173
+ img.save(new_path, 'PNG')
174
+ file_path = new_path
175
+ file_extension = '.png'
176
+ logger.info(f"Converted image to PNG: {file_path}")
177
+ except Exception as e:
178
+ logger.error(f"Failed to convert image: {str(e)}")
179
+ return None
180
+
181
+ files = {
182
+ 'fileToUpload': (
183
+ os.path.basename(file_path),
184
+ open(file_path, 'rb'),
185
+ mime_types.get(file_extension, 'application/octet-stream')
186
+ )
187
+ }
188
+
189
+ data = {
190
+ 'reqtype': 'fileupload',
191
+ 'userhash': CATBOX_USER_HASH
192
+ }
193
+
194
+ response = requests.post(url, files=files, data=data)
195
+
196
+ if response.status_code == 200 and response.text.startswith('http'):
197
+ file_url = response.text
198
+ logger.info(f"File uploaded successfully: {file_url}")
199
+ return file_url
200
+ else:
201
+ raise Exception(f"Upload failed: {response.text}")
202
+
203
+ except Exception as e:
204
+ logger.error(f"File upload error: {str(e)}")
205
+ return None
206
+ finally:
207
+ if 'new_path' in locals() and os.path.exists(new_path):
208
+ try:
209
+ os.remove(new_path)
210
+ except:
211
+ pass
212
+
213
+ def add_watermark(video_path):
214
+ """OpenCV๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ๋น„๋””์˜ค์— ์›Œํ„ฐ๋งˆํฌ ์ถ”๊ฐ€"""
215
+ try:
216
+ cap = cv2.VideoCapture(video_path)
217
+ width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
218
+ height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
219
+ fps = int(cap.get(cv2.CAP_PROP_FPS))
220
+
221
+ text = "GiniGEN.AI"
222
+ font = cv2.FONT_HERSHEY_SIMPLEX
223
+ font_scale = height * 0.05 / 30
224
+ thickness = 2
225
+ color = (255, 255, 255)
226
+
227
+ (text_width, text_height), _ = cv2.getTextSize(text, font, font_scale, thickness)
228
+ margin = int(height * 0.02)
229
+ x_pos = width - text_width - margin
230
+ y_pos = height - margin
231
+
232
+ output_path = "watermarked_output.mp4"
233
+ fourcc = cv2.VideoWriter_fourcc(*'mp4v')
234
+ out = cv2.VideoWriter(output_path, fourcc, fps, (width, height))
235
+
236
+ while cap.isOpened():
237
+ ret, frame = cap.read()
238
+ if not ret:
239
+ break
240
+ cv2.putText(frame, text, (x_pos, y_pos), font, font_scale, color, thickness)
241
+ out.write(frame)
242
+
243
+ cap.release()
244
+ out.release()
245
+
246
+ return output_path
247
+
248
+ except Exception as e:
249
+ logger.error(f"Error adding watermark: {str(e)}")
250
+ return video_path
251
+
252
+ def generate_video(image, prompt):
253
+ logger.info("Starting video generation with API")
254
+ try:
255
+ API_KEY = os.getenv("API_KEY", "").strip()
256
+ if not API_KEY:
257
+ return "API key not properly configured"
258
+
259
+ temp_dir = "temp_videos"
260
+ os.makedirs(temp_dir, exist_ok=True)
261
+
262
+ image_url = None
263
+ if image:
264
+ image_url = upload_to_catbox(image)
265
+ if not image_url:
266
+ return "Failed to upload image"
267
+ logger.info(f"Input image URL: {image_url}")
268
+
269
+ generation_url = "https://api.minimaxi.chat/v1/video_generation"
270
+ headers = {
271
+ 'authorization': f'Bearer {API_KEY}',
272
+ 'Content-Type': 'application/json'
273
+ }
274
+
275
+ payload = {
276
+ "model": "video-01",
277
+ "prompt": prompt if prompt else "",
278
+ "prompt_optimizer": True
279
+ }
280
+
281
+ if image_url:
282
+ payload["first_frame_image"] = image_url
283
+
284
+ logger.info(f"Sending request with payload: {payload}")
285
+
286
+ response = requests.post(generation_url, headers=headers, json=payload)
287
+
288
+ if not response.ok:
289
+ error_msg = f"Failed to create video generation task: {response.text}"
290
+ logger.error(error_msg)
291
+ return error_msg
292
+
293
+ response_data = response.json()
294
+ task_id = response_data.get('task_id')
295
+ if not task_id:
296
+ return "Failed to get task ID from response"
297
+
298
+ query_url = "https://api.minimaxi.chat/v1/query/video_generation"
299
+ max_attempts = 30
300
+ attempt = 0
301
+
302
+ while attempt < max_attempts:
303
+ time.sleep(10)
304
+ query_response = requests.get(
305
+ f"{query_url}?task_id={task_id}",
306
+ headers={'authorization': f'Bearer {API_KEY}'}
307
+ )
308
+
309
+ if not query_response.ok:
310
+ attempt += 1
311
+ continue
312
+
313
+ status_data = query_response.json()
314
+ status = status_data.get('status')
315
+
316
+ if status == 'Success':
317
+ file_id = status_data.get('file_id')
318
+ if not file_id:
319
+ return "Failed to get file ID"
320
+
321
+ retrieve_url = "https://api.minimaxi.chat/v1/files/retrieve"
322
+ params = {'file_id': file_id}
323
+
324
+ file_response = requests.get(
325
+ retrieve_url,
326
+ headers={'authorization': f'Bearer {API_KEY}'},
327
+ params=params
328
+ )
329
+
330
+ if not file_response.ok:
331
+ return "Failed to retrieve video file"
332
+
333
+ try:
334
+ file_data = file_response.json()
335
+ download_url = file_data.get('file', {}).get('download_url')
336
+ if not download_url:
337
+ return "Failed to get download URL"
338
+
339
+ result_info = {
340
+ "timestamp": datetime.now().strftime("%Y%m%d_%H%M%S"),
341
+ "input_image": image_url,
342
+ "output_video_url": download_url,
343
+ "prompt": prompt
344
+ }
345
+ logger.info(f"Video generation result: {json.dumps(result_info, indent=2)}")
346
+
347
+ video_response = requests.get(download_url)
348
+ if not video_response.ok:
349
+ return "Failed to download video"
350
+
351
+ timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
352
+ output_path = os.path.join(temp_dir, f"output_{timestamp}.mp4")
353
+
354
+ with open(output_path, 'wb') as f:
355
+ f.write(video_response.content)
356
+
357
+ final_path = add_watermark(output_path)
358
+
359
+ # ์˜ค๋””์˜ค ์ฒ˜๋ฆฌ ์ถ”๊ฐ€
360
+ try:
361
+ logger.info("Starting audio generation process")
362
+ final_path_with_audio = video_to_audio(
363
+ final_path,
364
+ prompt=prompt,
365
+ negative_prompt="music",
366
+ seed=-1,
367
+ num_steps=20,
368
+ cfg_strength=4.5,
369
+ target_duration=6.0
370
+ )
371
+
372
+ if final_path_with_audio != final_path:
373
+ logger.info("Audio generation successful")
374
+ try:
375
+ if output_path != final_path:
376
+ os.remove(output_path)
377
+ if final_path != final_path_with_audio:
378
+ os.remove(final_path)
379
+ except Exception as e:
380
+ logger.warning(f"Error cleaning up temporary files: {str(e)}")
381
+
382
+ return final_path_with_audio
383
+ else:
384
+ logger.warning("Audio generation skipped, using original video")
385
+ return final_path
386
+
387
+ except Exception as e:
388
+ logger.error(f"Error in audio processing: {str(e)}")
389
+ return final_path # ์˜ค๋””์˜ค ์ฒ˜๋ฆฌ ์‹คํŒจ ์‹œ ์›Œํ„ฐ๋งˆํฌ๋งŒ ๋œ ๋น„๋””์˜ค ๋ฐ˜ํ™˜
390
+
391
+ except Exception as e:
392
+ logger.error(f"Error processing video file: {str(e)}")
393
+ return "Error processing video file"
394
+
395
+ elif status == 'Fail':
396
+ return "Video generation failed"
397
+
398
+ attempt += 1
399
+
400
+ return "Timeout waiting for video generation"
401
+
402
+ except Exception as e:
403
+ logger.error(f"Error in video generation: {str(e)}")
404
+ return f"Error in video generation process: {str(e)}"
405
+
406
+ css = """
407
+ footer {
408
+ visibility: hidden;
409
+ }
410
+ .gradio-container {max-width: 1200px !important}
411
+ """
412
+
413
+ with gr.Blocks(theme="Yntec/HaleyCH_Theme_Orange", css=css) as demo:
414
+
415
+
416
+ with gr.Row():
417
+ with gr.Column(scale=3):
418
+ video_prompt = gr.Textbox(
419
+ label="Video Description",
420
+ placeholder="Enter video description...",
421
+ lines=3
422
+ )
423
+ upload_image = gr.Image(type="filepath", label="Upload First Frame Image")
424
+ video_generate_btn = gr.Button("๐ŸŽฌ Generate Video")
425
+
426
+ with gr.Column(scale=4):
427
+ video_output = gr.Video(label="Generated Video")
428
+
429
+
430
+
431
+
432
+ # process_and_generate_video ํ•จ์ˆ˜ ์ˆ˜์ •
433
+ def process_and_generate_video(image, prompt):
434
+ if image is None:
435
+ return "Please upload an image"
436
+
437
+ try:
438
+ # ํ•œ๊ธ€ ํ”„๋กฌํ”„ํŠธ ๊ฐ์ง€ ๋ฐ ๋ฒˆ์—ญ
439
+ contains_korean = any(ord('๊ฐ€') <= ord(char) <= ord('ํžฃ') for char in prompt)
440
+ if contains_korean:
441
+ translated = translator(prompt)[0]['translation_text']
442
+ logger.info(f"Translated prompt from '{prompt}' to '{translated}'")
443
+ prompt = translated
444
+
445
+ img = Image.open(image)
446
+ if img.mode != 'RGB':
447
+ img = img.convert('RGB')
448
+
449
+ temp_path = f"temp_{int(time.time())}.png"
450
+ img.save(temp_path, 'PNG')
451
+
452
+ result = generate_video(temp_path, prompt)
453
+
454
+ try:
455
+ os.remove(temp_path)
456
+ except:
457
+ pass
458
+
459
+ return result
460
+
461
+ except Exception as e:
462
+ logger.error(f"Error processing image: {str(e)}")
463
+ return "Error processing image"
464
+
465
+
466
+ video_generate_btn.click(
467
+ process_and_generate_video,
468
+ inputs=[upload_image, video_prompt],
469
+ outputs=video_output
470
+ )
471
+
472
+ if __name__ == "__main__":
473
+ demo.launch(server_name="0.0.0.0", server_port=7860, share=False)