File size: 15,552 Bytes
09b19a1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
#!/usr/bin/env python3

import os
import sys
import shutil
# single thread doubles cuda performance - needs to be set before torch import
if any(arg.startswith('--execution-provider') for arg in sys.argv):
    os.environ['OMP_NUM_THREADS'] = '1'

import warnings
from typing import List
import platform
import signal
import argparse
import torch
import onnxruntime

import roop.globals
import roop.metadata
import roop.utilities as util
import roop.ui as ui
from settings import Settings
from roop.face_helper import extract_face_images
from chain_img_processor import ChainImgProcessor, ChainVideoProcessor, ChainBatchImageProcessor

clip_text = None


if 'ROCMExecutionProvider' in roop.globals.execution_providers:
    del torch

warnings.filterwarnings('ignore', category=FutureWarning, module='insightface')
warnings.filterwarnings('ignore', category=UserWarning, module='torchvision')


def parse_args() -> None:
    signal.signal(signal.SIGINT, lambda signal_number, frame: destroy())
    program = argparse.ArgumentParser(formatter_class=lambda prog: argparse.HelpFormatter(prog, max_help_position=100))
    program.add_argument('-s', '--source', help='select a source image', dest='source_path')
    program.add_argument('-t', '--target', help='select a target image or video', dest='target_path')
    program.add_argument('-o', '--output', help='select output file or directory', dest='output_path')
    program.add_argument('-f', '--folder', help='select a target folder with images or videos to batch process', dest='target_folder_path')
    program.add_argument('--frame-processor', help='frame processors (choices: face_swapper, face_enhancer, ...)', dest='frame_processor', default=['face_swapper'], nargs='+')
    program.add_argument('--keep-fps', help='keep target fps', dest='keep_fps', action='store_true')
    program.add_argument('--keep-frames', help='keep temporary frames', dest='keep_frames', action='store_true')
    program.add_argument('--skip-audio', help='skip target audio', dest='skip_audio', action='store_true')
    program.add_argument('--many-faces', help='process every face', dest='many_faces', action='store_true')
    program.add_argument('--source-face_index', help='index position of source face in image', dest='source_face_index', type=int, default=0)
    program.add_argument('--target-face_index', help='index position of target face in image', dest='target_face_index', type=int, default=0)
    program.add_argument('--video-encoder', help='adjust output video encoder', dest='video_encoder', default='libx264', choices=['libx264', 'libx265', 'libvpx-vp9'])
    program.add_argument('--video-quality', help='adjust output video quality', dest='video_quality', type=int, default=18, choices=range(52), metavar='[0-51]')
    program.add_argument('--max-memory', help='maximum amount of RAM in GB', dest='max_memory', type=int, default=suggest_max_memory())
    program.add_argument('--execution-provider', help='available execution provider (choices: cpu, ...)', dest='execution_provider', default=['cpu'], choices=suggest_execution_providers(), nargs='+')
    program.add_argument('--execution-threads', help='number of execution threads', dest='execution_threads', type=int, default=suggest_execution_threads())
    program.add_argument('-v', '--version', action='version', version=f'{roop.metadata.name} {roop.metadata.version}')

    args = program.parse_args()

    roop.globals.source_path = args.source_path
    roop.globals.target_path = args.target_path
    roop.globals.output_path = util.normalize_output_path(roop.globals.source_path, roop.globals.target_path, args.output_path)
    roop.globals.target_folder_path = args.target_folder_path
    roop.globals.headless = args.source_path or args.target_path or args.output_path
    # Always enable all processors when using GUI
    if not roop.globals.headless:
        roop.globals.frame_processors = ['face_swapper', 'face_enhancer']
    else:
        roop.globals.frame_processors = args.frame_processor

    roop.globals.keep_fps = args.keep_fps
    roop.globals.keep_frames = args.keep_frames
    roop.globals.skip_audio = args.skip_audio
    roop.globals.many_faces = args.many_faces
    roop.globals.source_face_index = args.source_face_index
    roop.globals.target_face_index = args.target_face_index
    roop.globals.video_encoder = args.video_encoder
    roop.globals.video_quality = args.video_quality
    roop.globals.max_memory = args.max_memory
    roop.globals.execution_providers = decode_execution_providers(args.execution_provider)
    roop.globals.execution_threads = args.execution_threads


def encode_execution_providers(execution_providers: List[str]) -> List[str]:
    return [execution_provider.replace('ExecutionProvider', '').lower() for execution_provider in execution_providers]


def decode_execution_providers(execution_providers: List[str]) -> List[str]:
    return [provider for provider, encoded_execution_provider in zip(onnxruntime.get_available_providers(), encode_execution_providers(onnxruntime.get_available_providers()))
            if any(execution_provider in encoded_execution_provider for execution_provider in execution_providers)]


def suggest_max_memory() -> int:
    if platform.system().lower() == 'darwin':
        return 4
    return 16


def suggest_execution_providers() -> List[str]:
    return encode_execution_providers(onnxruntime.get_available_providers())


def suggest_execution_threads() -> int:
    if 'DmlExecutionProvider' in roop.globals.execution_providers:
        return 1
    if 'ROCMExecutionProvider' in roop.globals.execution_providers:
        return 1
    return 8


def limit_resources() -> None:
    # prevent tensorflow memory leak
    # gpus = tensorflow.config.experimental.list_physical_devices('GPU')
    # for gpu in gpus:
        # tensorflow.config.experimental.set_virtual_device_configuration(gpu, [
            # tensorflow.config.experimental.VirtualDeviceConfiguration(memory_limit=1024)
        # ])
    # limit memory usage
    if roop.globals.max_memory:
        memory = roop.globals.max_memory * 1024 ** 3
        if platform.system().lower() == 'darwin':
            memory = roop.globals.max_memory * 1024 ** 6
        if platform.system().lower() == 'windows':
            import ctypes
            kernel32 = ctypes.windll.kernel32
            kernel32.SetProcessWorkingSetSize(-1, ctypes.c_size_t(memory), ctypes.c_size_t(memory))
        else:
            import resource
            resource.setrlimit(resource.RLIMIT_DATA, (memory, memory))


def release_resources() -> None:
    if 'CUDAExecutionProvider' in roop.globals.execution_providers:
        torch.cuda.empty_cache()


def pre_check() -> bool:
    if sys.version_info < (3, 9):
        update_status('Python version is not supported - please upgrade to 3.9 or higher.')
        return False
    
    download_directory_path = util.resolve_relative_path('../models')
    util.conditional_download(download_directory_path, ['https://huggingface.co/countfloyd/deepfake/resolve/main/inswapper_128.onnx'])
    util.conditional_download(download_directory_path, ['https://huggingface.co/countfloyd/deepfake/resolve/main/GFPGANv1.4.pth'])
    util.conditional_download(download_directory_path, ['https://github.com/csxmli2016/DMDNet/releases/download/v1/DMDNet.pth'])
    download_directory_path = util.resolve_relative_path('../models/CLIP')
    util.conditional_download(download_directory_path, ['https://huggingface.co/countfloyd/deepfake/resolve/main/rd64-uni-refined.pth'])
    download_directory_path = util.resolve_relative_path('../models/CodeFormer')
    util.conditional_download(download_directory_path, ['https://github.com/sczhou/CodeFormer/releases/download/v0.1.0/codeformer.pth'])
    download_directory_path = util.resolve_relative_path('../models/CodeFormer/facelib')
    util.conditional_download(download_directory_path, ['https://github.com/sczhou/CodeFormer/releases/download/v0.1.0/detection_Resnet50_Final.pth'])
    util.conditional_download(download_directory_path, ['https://github.com/sczhou/CodeFormer/releases/download/v0.1.0/parsing_parsenet.pth'])
    download_directory_path = util.resolve_relative_path('../models/CodeFormer/realesrgan')
    util.conditional_download(download_directory_path, ['https://github.com/sczhou/CodeFormer/releases/download/v0.1.0/RealESRGAN_x2plus.pth'])

    if not shutil.which('ffmpeg'):
       update_status('ffmpeg is not installed.')
    return True


def update_status(message: str, scope: str = 'ROOP.CORE') -> None:
    print(f'[{scope}] {message}')
    # if not roop.globals.headless:
        # ui.update_status(message)



def start() -> None:
    if roop.globals.headless:
        faces = extract_face_images(roop.globals.source_path,  (False, 0))
        roop.globals.SELECTED_FACE_DATA_INPUT = faces[roop.globals.source_face_index]
        faces = extract_face_images(roop.globals.target_path,  (False, util.has_image_extension(roop.globals.target_path)))
        roop.globals.SELECTED_FACE_DATA_OUTPUT = faces[roop.globals.target_face_index]
        if 'face_enhancer' in roop.globals.frame_processors:
            roop.globals.selected_enhancer = 'GFPGAN'
       
    batch_process(None, False, None)


def InitPlugins():
    if not roop.globals.IMAGE_CHAIN_PROCESSOR:
        roop.globals.IMAGE_CHAIN_PROCESSOR = ChainImgProcessor()
        roop.globals.BATCH_IMAGE_CHAIN_PROCESSOR = ChainBatchImageProcessor()
        roop.globals.VIDEO_CHAIN_PROCESSOR = ChainVideoProcessor()
        roop.globals.IMAGE_CHAIN_PROCESSOR.init_with_plugins()
        roop.globals.BATCH_IMAGE_CHAIN_PROCESSOR.init_with_plugins()
        roop.globals.VIDEO_CHAIN_PROCESSOR.init_with_plugins()


def get_processing_plugins(use_clip):
    processors = "faceswap"
    if use_clip:
        processors += ",txt2clip"
    
    if roop.globals.selected_enhancer == 'GFPGAN':
        processors += ",gfpgan"
    elif roop.globals.selected_enhancer == 'Codeformer':
        processors += ",codeformer"
    elif roop.globals.selected_enhancer == 'DMDNet':
        processors += ",dmdnet"
    
    return processors


def live_swap(frame, swap_mode, use_clip, clip_text):
    if frame is None:
        return frame

    InitPlugins()
    processors = get_processing_plugins(use_clip)


    temp_frame, _ = roop.globals.IMAGE_CHAIN_PROCESSOR.run_chain(frame,  
                                                    {"swap_mode": swap_mode,
                                                        "original_frame": frame,
                                                        "blend_ratio": roop.globals.blend_ratio,
                                                        "face_distance_threshold": roop.globals.distance_threshold,
                                                        "input_face_datas": [roop.globals.SELECTED_FACE_DATA_INPUT], "target_face_datas": [roop.globals.SELECTED_FACE_DATA_OUTPUT],
                                                        "clip_prompt": clip_text},
                                                        processors)
    return temp_frame



def params_gen_func(proc, frame):
    global clip_text

    return {"original_frame": frame, "blend_ratio": roop.globals.blend_ratio,
             "swap_mode": roop.globals.face_swap_mode, "face_distance_threshold": roop.globals.distance_threshold, 
             "input_face_datas": [roop.globals.SELECTED_FACE_DATA_INPUT], "target_face_datas": [roop.globals.SELECTED_FACE_DATA_OUTPUT],
             "clip_prompt": clip_text}

def batch_process(files, use_clip, new_clip_text) -> None:
    global clip_text

    InitPlugins()
    processors = get_processing_plugins(use_clip)

    clip_text = new_clip_text

    imagefiles = []
    imagefinalnames = []
    videofiles = []
    videofinalnames = []
    need_join = False

    if files is None:
        need_join = True
        if roop.globals.target_folder_path is None:
            roop.globals.target_folder_path = os.path.dirname(roop.globals.target_path)
            files = [os.path.basename(roop.globals.target_path)]
            roop.globals.output_path = os.path.dirname(roop.globals.output_path)
        else:
            files = [f for f in os.listdir(roop.globals.target_folder_path) if os.path.isfile(os.path.join(roop.globals.target_folder_path, f))]
            
        update_status('Sorting videos/images')


    for f in files:
        if need_join:
            fullname = os.path.join(roop.globals.target_folder_path, f)
        else:
            fullname = f
        if util.has_image_extension(fullname):
            imagefiles.append(fullname)
            imagefinalnames.append(util.get_destfilename_from_path(fullname, roop.globals.output_path, f'_fake.{roop.globals.CFG.output_image_format}'))
        elif util.is_video(fullname) or util.has_extension(fullname, ['gif']):
            videofiles.append(fullname)
            videofinalnames.append(util.get_destfilename_from_path(fullname, roop.globals.output_path, f'_fake.{roop.globals.CFG.output_video_format}'))


    if(len(imagefiles) > 0):
        update_status('Processing image(s)')
        roop.globals.BATCH_IMAGE_CHAIN_PROCESSOR.run_batch_chain(imagefiles, imagefinalnames, roop.globals.execution_threads, processors, params_gen_func)
    if(len(videofiles) > 0):
        for index,v in enumerate(videofiles):
            update_status(f'Processing video {v}')
            fps = util.detect_fps(v)
            if roop.globals.keep_frames:
                update_status('Creating temp resources...')
                util.create_temp(v)
                update_status('Extracting frames...')
                util.extract_frames(v)
                temp_frame_paths = util.get_temp_frame_paths(v)
                roop.globals.BATCH_IMAGE_CHAIN_PROCESSOR.run_batch_chain(temp_frame_paths, temp_frame_paths, roop.globals.execution_threads, processors, params_gen_func)
                update_status(f'Creating video with {fps} FPS...')
                util.create_video(v, videofinalnames[index], fps)
            else:
                update_status(f'Creating video with {fps} FPS...')
                roop.globals.VIDEO_CHAIN_PROCESSOR.run_video_chain(v,videofinalnames[index], fps, roop.globals.execution_threads, processors, params_gen_func, roop.globals.target_path)
            if os.path.isfile(videofinalnames[index]):
                if util.has_extension(v, ['gif']):
                    gifname = roop.utilities.get_destfilename_from_path(v, './output', '_fake.gif')
                    update_status('Creating final GIF')
                    util.create_gif_from_video(videofinalnames[index], gifname)
                elif not roop.globals.skip_audio:
                    finalname = roop.utilities.get_destfilename_from_path(videofinalnames[index], roop.globals.output_path, f'_final.{roop.globals.CFG.output_video_format}')
                    util.restore_audio(videofinalnames[index], v, finalname)
                    if os.path.isfile(videofinalnames[index]):
                        os.remove(videofinalnames[index])
            else:
                update_status('Failed!')

            
    update_status('Finished')
    roop.globals.target_folder_path = None


def destroy() -> None:
    if roop.globals.target_path:
        util.clean_temp(roop.globals.target_path)
    sys.exit()


def run() -> None:
    parse_args()
    if not pre_check():
        return
    limit_resources()
    roop.globals.CFG = Settings('config.yaml')
    if roop.globals.headless:
        start()
    else:
        ui.run()