如何使用 ffmpeg-python 在视频上叠加帧序列？

我在下面尝试过，但它只显示背景视频。

background_video = ffmpeg.input( "input.mp4")
overlay_video = ffmpeg.input(f'{frames_folder}*.png', pattern_type='glob', framerate=25)
subprocess = ffmpeg.overlay(
                          background_video,
                          overlay_video,
                        ).filter("setsar", sar=1)

我还尝试将帧序列组装成 .webm/.mov 视频，但透明度丢失。视频以黑色为背景。

P.s - 帧大小与背景视频大小相同。所以不需要缩放。

Edit

我尝试了@Rotem的建议

首先尝试使用单个 PNG 图像

overlay_video =  ffmpeg.input('test-frame.png')

它不适用于 OpenCV 生成的帧，但适用于任何其他 png 图像。这很奇怪，当我手动查看这些帧文件夹时，它显示空白图像（链接到我的框架文件夹 https://drive.google.com/drive/folders/1i3Ktoo0C97OPaqnsH2DDoWECfKQJ2z_x?usp=sharing）。 但是，如果我将这些帧转换为视频（见下文），它会正确显示我在每个帧上绘制的内容。

output_options = {
                    'crf': 20,
                    'preset': 'slower',
                    'movflags': 'faststart',
                    'pix_fmt': 'yuv420p'
                }
ffmpeg.input(f'{frames_folder}*.png', pattern_type='glob', framerate=25 , reinit_filter=0).output(
                    'movie.avi',
                    **output_options
                ).global_args('-report').run()

尝试从所有 PNG 图像创建视频而不叠加

它按预期工作，唯一的问题是透明度。有没有办法制作透明背景视频？我尝试了 .webm/.mov/.avi 但没有运气。

添加 .global_args('-report') 并检查日志文件

Report written to "ffmpeg-20221119-110731.log"
Log level: 48
ffmpeg version 5.1 Copyright (c) 2000-2022 the FFmpeg developers
  built with Apple clang version 13.1.6 (clang-1316.0.21.2.5)
  configuration: --prefix=/opt/homebrew/Cellar/ffmpeg/5.1 --enable-shared --enable-pthreads --enable-version3 --cc=clang --host-cflags= --host-ldflags= --enable-ffplay --enable-gnutls --enable-gpl --enable-libaom --enable-libbluray --enable-libdav1d --enable-libmp3lame --enable-libopus --enable-librav1e --enable-librist --enable-librubberband --enable-libsnappy --enable-libsrt --enable-libtesseract --enable-libtheora --enable-libvidstab --enable-libvmaf --enable-libvorbis --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxml2 --enable-libxvid --enable-lzma --enable-libfontconfig --enable-libfreetype --enable-frei0r --enable-libass --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-libspeex --enable-libsoxr --enable-libzmq --enable-libzimg --disable-libjack --disable-indev=jack --enable-videotoolbox --enable-neon
  libavutil      57. 28.100 / 57. 28.100
  libavcodec     59. 37.100 / 59. 37.100
  libavformat    59. 27.100 / 59. 27.100
  libavdevice    59.  7.100 / 59.  7.100
  libavfilter     8. 44.100 /  8. 44.100
  libswscale      6.  7.100 /  6.  7.100
  libswresample   4.  7.100 /  4.  7.100
  libpostproc    56.  6.100 / 56.  6.100
Input #0, image2, from './frames/*.png':
  Duration: 00:00:05.00, start: 0.000000, bitrate: N/A
  Stream #0:0: Video: png, rgba(pc), 1920x1080, 25 fps, 25 tbr, 25 tbn
Codec AVOption crf (Select the quality for constant quality mode) specified for output file #0 (movie.avi) has not been used for any stream. The most likely reason is either wrong type (e.g. a video option with no video streams) or that it is a private option of some encoder which was not actually used for any stream.
Codec AVOption preset (Configuration preset) specified for output file #0 (movie.avi) has not been used for any stream. The most likely reason is either wrong type (e.g. a video option with no video streams) or that it is a private option of some encoder which was not actually used for any stream.
Stream mapping:
  Stream #0:0 -> #0:0 (png (native) -> mpeg4 (native))
Press [q] to stop, [?] for help
Output #0, avi, to 'movie.avi':
  Metadata:
    ISFT            : Lavf59.27.100
  Stream #0:0: Video: mpeg4 (FMP4 / 0x34504D46), yuv420p(tv, progressive), 1920x1080, q=2-31, 200 kb/s, 25 fps, 25 tbn
    Metadata:
      encoder         : Lavc59.37.100 mpeg4
    Side data:
      cpb: bitrate max/min/avg: 0/0/200000 buffer size: 0 vbv_delay: N/A
frame=  125 fps= 85 q=31.0 Lsize=     491kB time=00:00:05.00 bitrate= 804.3kbits/s speed=3.39x    
video:482kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 1.772174%

我在下面使用了绘制框架。

for i in range(num_frames):
            transparent_img = np.zeros((height, width, 4),  dtype=np.uint8)
            cv2.line(transparent_img, (x1,y1), (x2,y2) ,(255, 255, 255), thickness=1, lineType=cv2.LINE_AA)
            self.frames.append(transparent_img)


## To Save each frame of the video in the given folder
for i, f in enumerate(frames):
    cv2.imwrite("{}/{:0{n}d}.png".format(path_to_frames, i, n=num_digits), f)

以下是您的两个问题的答案：

• 要在 BGRA 图像上绘制白线，请使用(255, 255, 255, 255)颜色而不是(255, 255, 255).
  最后255应用 alpha（透明度）通道值，并且255使线条完全不透明。

• 要创建具有透明背景的视频，请尝试：.webm文件类型、用途libvpx-vp9视频编解码器，并使用-pix_fmt yuva420p - the a of yuva应用 alpha（透明度）通道。

这是一个“自包含”代码示例（请阅读注释）：

import cv2
import numpy as np
import ffmpeg

# Create synthetic MP4 video file from testing
ffmpeg.input('testsrc=size=192x108:rate=1:duration=10', f='lavfi').output('tmp.mp4').overwrite_output().run()

transparent_img = np.zeros((108, 192, 4), np.uint8)

width, height, fps = 192, 108, 1

def make_sample_image(i):
    p = width//60
    img = np.zeros((height, width, 4), np.uint8)  # Fully transparent
    cv2.putText(img, str(i), (width//2-p*10*len(str(i)), height//2+p*10), cv2.FONT_HERSHEY_DUPLEX, p, (255, 255, 255, 255), p*2)  # White number
    return img

# Create 10 PNG files with transparent background an white number (counter).
for i in range(1, 11):
    transparent_img = make_sample_image(i)
    cv2.imwrite(f'{i:03d}.png', transparent_img)


output_options = {  'vcodec' : 'libvpx-vp9',  # libvpx-vp9 supports transparency.
                    'crf': 20,
                    #'preset': 'slower',  # Not supported by libvpx-vp9
                    #'movflags': 'faststart',  # Not supported by WebM
                    'pix_fmt': 'yuva420p'  # yuva420p includes transparency.
                }

frames_folder = './'

# Create video with transparency:
# reinit_filter=0 is required, only if the PNG images have different characteristics (example: some are RGB and some RGBA).
# Use %03d.png instead of glob pattern, becuase my Windows version of FFmpeg doesn't support glob pattern.
ffmpeg.input(f'{frames_folder}%03d.png', framerate=fps, reinit_filter=0).output(
                    'movie.webm',  # WebM container supports transparency
                    **output_options
                ).global_args('-report').overwrite_output().run()


# Overlay the PNG on top of tmp.mp4
background_video = ffmpeg.input( "tmp.mp4")
overlay_video = ffmpeg.input(f'{frames_folder}%03d.png', framerate=fps)
subprocess = ffmpeg.overlay(
                          background_video,
                          overlay_video,
                        ).filter("setsar", sar=1)

subprocess.output('overlay_video.webm', **output_options).global_args('-report').overwrite_output().run()