support ffmpeg stream for inference_realesrgan_video (#308)

* support ffmpeg stream for inference_realesrgan_video

* fix code format

Co-authored-by: yanzewu <yanzewu@tencent.com>

docs/anime_video_model.md

@@ -36,7 +36,12 @@ The following are some demos (best view in the full screen mode).
 # download model
 wget https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.5.0/realesr-animevideov3.pth -P realesrgan/weights
 # inference
-python inference_realesrgan_video.py -i inputs/video/onepiece_demo.mp4 -n realesr-animevideov3 -s 2 --suffix outx2
+python inference_realesrgan_video.py -i inputs/video/onepiece_demo.mp4 -n realesr-animevideov3 -s 2 --suffix outx2 --stream
+```
+```console
+Usage:
+--stream                 with this option, the enhanced frames are sent directly to a ffmpeg stream,
+                         avoiding storing large (usually tens of GB) intermediate results.        
 ```
 
 ### NCNN Executable File

inference_realesrgan_video.py

@@ -1,6 +1,8 @@
 import argparse
+import cv2
 import glob
 import mimetypes
+import numpy as np
 import os
 import queue
 import shutil
@@ -13,6 +15,192 @@ from realesrgan import IOConsumer, PrefetchReader, RealESRGANer
 from realesrgan.archs.srvgg_arch import SRVGGNetCompact
 
 
+def get_frames(args, extract_frames=False):
+    # input can be a video file / a folder of frames / an image
+    is_video = False
+    if mimetypes.guess_type(args.input)[0].startswith('video'):  # is a video file
+        is_video = True
+        video_name = os.path.splitext(os.path.basename(args.input))[0]
+        if extract_frames:
+            frame_folder = os.path.join('tmp_frames', video_name)
+            os.makedirs(frame_folder, exist_ok=True)
+            # use ffmpeg to extract frames
+            os.system(f'ffmpeg -i {args.input} -qscale:v 1 -qmin 1 -qmax 1 -vsync 0  {frame_folder}/frame%08d.png')
+            # get image path list
+            paths = sorted(glob.glob(os.path.join(frame_folder, '*')))
+        else:
+            paths = []
+        # get input video fps
+        if args.fps is None:
+            import ffmpeg
+            probe = ffmpeg.probe(args.input)
+            video_streams = [stream for stream in probe['streams'] if stream['codec_type'] == 'video']
+            args.fps = eval(video_streams[0]['avg_frame_rate'])
+    elif mimetypes.guess_type(args.input)[0].startswith('image'):  # is an image file
+        paths = [args.input]
+    else:
+        paths = sorted(glob.glob(os.path.join(args.input, '*')))
+        assert len(paths) > 0, 'the input folder is empty'
+
+    if args.fps is None:
+        args.fps = 24
+
+    return is_video, paths
+
+
+def inference_stream(args, upsampler, face_enhancer):
+    try:
+        import ffmpeg
+    except ImportError:
+        import pip
+        pip.main(['install', '--user', 'ffmpeg-python'])
+        import ffmpeg
+
+    is_video, paths = get_frames(args, extract_frames=False)
+    video_name = os.path.splitext(os.path.basename(args.input))[0]
+    video_save_path = os.path.join(args.output, f'{video_name}_{args.suffix}.mp4')
+
+    # decoder
+    if is_video:
+        # get height and width
+        probe = ffmpeg.probe(args.input)
+        video_streams = [stream for stream in probe['streams'] if stream['codec_type'] == 'video']
+        width = video_streams[0]['width']
+        height = video_streams[0]['height']
+
+        # set up frame decoder
+        decoder = (
+            ffmpeg.input(args.input).output('pipe:', format='rawvideo', pix_fmt='rgb24', loglevel='warning').run_async(
+                pipe_stdin=True, pipe_stdout=True, cmd=args.ffmpeg_bin))
+    else:
+        from PIL import Image
+        tmp_img = Image.open(paths[0])
+        width, height = tmp_img.size
+        idx = 0
+
+    out_width, out_height = int(width * args.outscale), int(height * args.outscale)
+    if out_height > 2160:
+        print('You are generating video that is larger than 4K, which will be very slow due to IO speed.',
+              'We highly recommend to decrease the outscale(aka, -s).')
+    # encoder
+    if is_video:
+        audio = ffmpeg.input(args.input).audio
+        encoder = (
+            ffmpeg.input(
+                'pipe:', format='rawvideo', pix_fmt='rgb24', s=f'{out_width}x{out_height}', framerate=args.fps).output(
+                    audio, video_save_path, pix_fmt='yuv420p', vcodec='libx264', loglevel='info',
+                    acodec='copy').overwrite_output().run_async(pipe_stdin=True, pipe_stdout=True, cmd=args.ffmpeg_bin))
+    else:
+        encoder = (
+            ffmpeg.input(
+                'pipe:', format='rawvideo', pix_fmt='rgb24', s=f'{out_width}x{out_height}',
+                framerate=args.fps).output(video_save_path, pix_fmt='yuv420p', vcodec='libx264',
+                                           loglevel='info').overwrite_output().run_async(
+                                               pipe_stdin=True, pipe_stdout=True, cmd=args.ffmpeg_bin))
+
+    while True:
+        if is_video:
+            img_bytes = decoder.stdout.read(width * height * 3)  # 3 bytes for one pixel
+            if not img_bytes:
+                break
+            img = np.frombuffer(img_bytes, np.uint8).reshape([height, width, 3])
+        else:
+            if idx >= len(paths):
+                break
+            img = cv2.imread(paths[idx])
+            idx += 1
+
+        try:
+            if args.face_enhance:
+                _, _, output = face_enhancer.enhance(img, has_aligned=False, only_center_face=False, paste_back=True)
+            else:
+                output, _ = upsampler.enhance(img, outscale=args.outscale)
+        except RuntimeError as error:
+            print('Error', error)
+            print('If you encounter CUDA out of memory, try to set --tile with a smaller number.')
+        else:
+            output = output.astype(np.uint8).tobytes()
+            encoder.stdin.write(output)
+
+        torch.cuda.synchronize()
+
+    if is_video:
+        decoder.stdin.close()
+        decoder.wait()
+    encoder.stdin.close()
+    encoder.wait()
+
+
+def inference_frames(args, upsampler, face_enhancer):
+    is_video, paths = get_frames(args, extract_frames=True)
+    video_name = os.path.splitext(os.path.basename(args.input))[0]
+
+    # for saving restored frames
+    save_frame_folder = os.path.join(args.output, video_name, 'frames_tmpout')
+    os.makedirs(save_frame_folder, exist_ok=True)
+
+    timer = AvgTimer()
+    timer.start()
+    pbar = tqdm(total=len(paths), unit='frame', desc='inference')
+    # set up prefetch reader
+    reader = PrefetchReader(paths, num_prefetch_queue=4)
+    reader.start()
+
+    que = queue.Queue()
+    consumers = [IOConsumer(args, que, f'IO_{i}') for i in range(args.consumer)]
+    for consumer in consumers:
+        consumer.start()
+
+    for idx, (path, img) in enumerate(zip(paths, reader)):
+        imgname, extension = os.path.splitext(os.path.basename(path))
+        if len(img.shape) == 3 and img.shape[2] == 4:
+            img_mode = 'RGBA'
+        else:
+            img_mode = None
+
+        try:
+            if args.face_enhance:
+                _, _, output = face_enhancer.enhance(img, has_aligned=False, only_center_face=False, paste_back=True)
+            else:
+                output, _ = upsampler.enhance(img, outscale=args.outscale)
+        except RuntimeError as error:
+            print('Error', error)
+            print('If you encounter CUDA out of memory, try to set --tile with a smaller number.')
+
+        else:
+            if args.ext == 'auto':
+                extension = extension[1:]
+            else:
+                extension = args.ext
+            if img_mode == 'RGBA':  # RGBA images should be saved in png format
+                extension = 'png'
+            save_path = os.path.join(save_frame_folder, f'{imgname}_out.{extension}')
+
+            que.put({'output': output, 'save_path': save_path})
+
+        pbar.update(1)
+        torch.cuda.synchronize()
+        timer.record()
+        avg_fps = 1. / (timer.get_avg_time() + 1e-7)
+        pbar.set_description(f'idx {idx}, fps {avg_fps:.2f}')
+
+    for _ in range(args.consumer):
+        que.put('quit')
+    for consumer in consumers:
+        consumer.join()
+    pbar.close()
+
+    # merge frames to video
+    video_save_path = os.path.join(args.output, f'{video_name}_{args.suffix}.mp4')
+    os.system(f'ffmpeg -r {args.fps} -i {save_frame_folder}/frame%08d_out.{extension} -i {args.input}'
+              f' -map 0:v:0 -map 1:a:0 -c:a copy -c:v libx264 -r {args.fps} -pix_fmt yuv420p  {video_save_path}')
+    # delete tmp file
+    shutil.rmtree(save_frame_folder)
+    frame_folder = os.path.join('tmp_frames', video_name)
+    if os.path.isdir(frame_folder):
+        shutil.rmtree(frame_folder)
+
+
 def main():
     """Inference demo for Real-ESRGAN.
     It mainly for restoring anime videos.
@@ -39,6 +227,8 @@ def main():
         '--fp32', action='store_true', help='Use fp32 precision during inference. Default: fp16 (half precision).')
     parser.add_argument('--fps', type=float, default=None, help='FPS of the output video')
     parser.add_argument('--consumer', type=int, default=4, help='Number of IO consumers')
+    parser.add_argument('--stream', action='store_true')
+    parser.add_argument('--ffmpeg_bin', type=str, default='ffmpeg', help='The path to ffmpeg')
 
     parser.add_argument(
         '--alpha_upsampler',
@@ -52,6 +242,8 @@ def main():
         help='Image extension. Options: auto | jpg | png, auto means using the same extension as inputs')
     args = parser.parse_args()
 
+    args.input = args.input.rstrip('/').rstrip('\\')
+
     # ---------------------- determine models according to model names ---------------------- #
     args.model_name = args.model_name.split('.pth')[0]
     if args.model_name in ['RealESRGAN_x4plus', 'RealESRNet_x4plus']:  # x4 RRDBNet model
@@ -84,6 +276,11 @@ def main():
         pre_pad=args.pre_pad,
         half=not args.fp32)
 
+    if 'anime' in args.model_name and args.face_enhance:
+        print('face_enhance is not supported in anime models, we turned this option off for you. '
+              'if you insist on turning it on, please manually comment the relevant lines of code.')
+        args.face_enhance = False
+
     if args.face_enhance:  # Use GFPGAN for face enhancement
         from gfpgan import GFPGANer
         face_enhancer = GFPGANer(
@@ -92,93 +289,15 @@ def main():
             arch='clean',
             channel_multiplier=2,
             bg_upsampler=upsampler)
-    os.makedirs(args.output, exist_ok=True)
-    # for saving restored frames
-    save_frame_folder = os.path.join(args.output, 'frames_tmpout')
-    os.makedirs(save_frame_folder, exist_ok=True)
-
-    # input can be a video file / a folder of frames / an image
-    if mimetypes.guess_type(args.input)[0].startswith('video'):  # is a video file
-        video_name = os.path.splitext(os.path.basename(args.input))[0]
-        frame_folder = os.path.join('tmp_frames', video_name)
-        os.makedirs(frame_folder, exist_ok=True)
-        # use ffmpeg to extract frames
-        os.system(f'ffmpeg -i {args.input} -qscale:v 1 -qmin 1 -qmax 1 -vsync 0  {frame_folder}/frame%08d.png')
-        # get image path list
-        paths = sorted(glob.glob(os.path.join(frame_folder, '*')))
-        # get input video fps
-        if args.fps is None:
-
-            import ffmpeg
-            probe = ffmpeg.probe(args.input)
-            video_streams = [stream for stream in probe['streams'] if stream['codec_type'] == 'video']
-            args.fps = eval(video_streams[0]['avg_frame_rate'])
-    elif mimetypes.guess_type(args.input)[0].startswith('image'):  # is an image file
-        paths = [args.input]
-        video_name = 'video'
     else:
-        paths = sorted(glob.glob(os.path.join(args.input, '*')))
-        video_name = 'video'
+        face_enhancer = None
 
-    timer = AvgTimer()
-    timer.start()
-    pbar = tqdm(total=len(paths), unit='frame', desc='inference')
-    # set up prefetch reader
-    reader = PrefetchReader(paths, num_prefetch_queue=4)
-    reader.start()
-
-    que = queue.Queue()
-    consumers = [IOConsumer(args, que, f'IO_{i}') for i in range(args.consumer)]
-    for consumer in consumers:
-        consumer.start()
-
-    for idx, (path, img) in enumerate(zip(paths, reader)):
-        imgname, extension = os.path.splitext(os.path.basename(path))
-        if len(img.shape) == 3 and img.shape[2] == 4:
-            img_mode = 'RGBA'
-        else:
-            img_mode = None
-
-        try:
-            if args.face_enhance:
-                _, _, output = face_enhancer.enhance(img, has_aligned=False, only_center_face=False, paste_back=True)
-            else:
-                output, _ = upsampler.enhance(img, outscale=args.outscale)
-        except RuntimeError as error:
-            print('Error', error)
-            print('If you encounter CUDA out of memory, try to set --tile with a smaller number.')
-
-        else:
-            if args.ext == 'auto':
-                extension = extension[1:]
-            else:
-                extension = args.ext
-            if img_mode == 'RGBA':  # RGBA images should be saved in png format
-                extension = 'png'
-            save_path = os.path.join(save_frame_folder, f'{imgname}_out.{extension}')
-
-            que.put({'output': output, 'save_path': save_path})
-
-        pbar.update(1)
-        torch.cuda.synchronize()
-        timer.record()
-        avg_fps = 1. / (timer.get_avg_time() + 1e-7)
-        pbar.set_description(f'idx {idx}, fps {avg_fps:.2f}')
-
-    for _ in range(args.consumer):
-        que.put('quit')
-    for consumer in consumers:
-        consumer.join()
-    pbar.close()
+    os.makedirs(args.output, exist_ok=True)
 
-    # merge frames to video
-    video_save_path = os.path.join(args.output, f'{video_name}_{args.suffix}.mp4')
-    os.system(f'ffmpeg -r {args.fps} -i {save_frame_folder}/frame%08d_out.{extension} -i {args.input}'
-              f' -map 0:v:0 -map 1:a:0 -c:a copy -c:v libx264 -r {args.fps} -pix_fmt yuv420p  {video_save_path}')
-    # delete tmp file
-    shutil.rmtree(save_frame_folder)
-    if os.path.isdir(frame_folder):
-        shutil.rmtree(frame_folder)
+    if args.stream:
+        inference_stream(args, upsampler, face_enhancer)
+    else:
+        inference_frames(args, upsampler, face_enhancer)
 
 
 if __name__ == '__main__':