Newest 'ffmpeg' Questions - Stack Overflow

Les articles publiés sur le site

• OpenCV compilation : How to specify the location of FFmpeg library with cmake

  6 juin 2017, par Beanocean

  I want to compile OpenCV-2.13.2 with gcc-4.8.2, but the version installed in system path is gcc-4.4.6. So I installed gcc-4.8.2 in /opt/compiler/gcc-4.8.2. I compiled FFmpeg successfully with gcc-4.8.2. When I tried to compile OpenCV, I met with some problems.

  In linking stage, there were some libraries can not be found by /opt/compiler/gcc-4.8.2/bin/ld. The error message is as follows:

  The missing libraries are related with FFmpeg, and I have installed FFmpeg in ~/local/lib. Then I checked the file module/core/CMakeFiles/opencv_pref_core.dir/link.txt, It shows as follows:

  The ld just did not search the path where I installed FFmpeg. I tried two methods:

  1. add FFmpeg path to env: export LD_LIBRARY_PATH=~/local/lib:$LD_LIBRARY_PATH;
  2. add -D FFMPEG_INCLUDE_DIRS=~/local/include -D FFMPEG_LIBRARAY_DIRS=~/local/lib to cmake options

  They did not work at all.

• javacv FFmpegFrameGrabber http Authorization Required

  6 juin 2017, par Anastasia Sisordia

  I use the javacv library and http protocol for connect to camera. I have the adress (http://123.456.78.90/ for example) and login and password.

  ( For example: admin 12345 )

  I write

  FFmpegFrameGrabber grabber = new FFmpegFrameGrabber("http://123.456.78.90/");
  

  but I dont know where I need to use login+password. If I don`t use, it catch exeption: HTTP error 401 Authorization Required.

  Anyone can say me how I can send login/password. I tried http://login:password@123.456.78.90/ and this don`t work.

• How to export video as .mp4 using openCV ?

  6 juin 2017, par Riko

  I am trying to export video as .mp4 with openCV. I have tried several codecs but for now I had no success.

  This is a function that constructs a video from frames:

  def create_movie(self, out_directory, fps, total_frames):
      img1 = cv2.imread("temp/scr0.png")
      height, width, layers =  img1.shape
      codec = cv2.cv.CV_FOURCC('X','V','I','D')
      video = cv2.VideoWriter(out_directory, codec, fps, (width, height))
  
      for i in range(total_frames):
          img_name = "temp/scr" + str(i) + ".png"
          img = cv2.imread(img_name)
          video.write(img)
  
      video.release()
      cv2.destroyAllWindows()
  

  I usually get next error message, using different codecs:

  Tag XVID/0x44495658 incompatible with output codec id '13'
  

  Is is possible to do this and how?

• ffmpeg convert AVFrame from yuv420p to argb

  6 juin 2017, par Star丶Xing

  I'm tring to decode h.264 stream buffer on android using ffmpeg 3.3. I got an AVFrame structure from the stream successfully, it's format is AV_PIX_FMT_YUV420P. In order to display the image on android, I need to convert it to AV_PIX_FMT_ARGB or AV_PIX_FMT_RGB565. I tried many times, finally I find that I can convert the frame to AV_PIX_FMT_NV21 but cannot convert it to argb or rgb565 using almost same code except some arguments.Here is my code looks like:

  //here i got srcFrame from h264 stream, it's format and size was set by decoder correctly
  
  AVFrame *dstFrame = av_frame_alloc();
  dstFrame->format = AV_PIX_FMT_ARGB;
  dstFrame->width = srcFrame->width;
  dstFrame->height = srcFrame->height;
  
  struct SwsContext *swsContext = sws_getContext(srcFrame->width, srcFrame->height, srcFormat,
                                                 dstFrame->width, dstFrame->height,
                                                 (enum AVPixelFormat) dstFrame->format,
                                                 SWS_FAST_BILINEAR, NULL, NULL, NULL);
  
  int dstBufferSize = av_image_get_buffer_size((enum AVPixelFormat) dstFrame->format, ctx->width,
                                               ctx->height, 1);
  uint8_t *dstBuffer = (uint8_t *) av_malloc(
          sizeof(uint8_t) * dstBufferSize);
  
  av_image_fill_arrays(dstFrame->data, dstFrame->linesize, dstBuffer,
                       (enum AVPixelFormat) dstFrame->format, dstFrame->width,
                       dstFrame->height,
                       1);
  
  sws_scale(swsContext, (const uint8_t *const *) srcFrame->data,
            srcFrame->linesize, 0, srcFrame->height,
            dstFrame->data, dstFrame->linesize);
  

• what's the theory of mix audio in the ffmpeg

  6 juin 2017, par from_mars

  when I wanted to mix two file into one, I found the toolkit.when I did this after following the command, everything was fine,just like:

  ffmpeg -i 1.wav -i 2.wav -filter_complex amix=inputs=2:duration=longest:dropout_transition=2 out.wav

  but there is something I want to know.

  1. what's the theory of mix audio in the ffmpeg.

     Do you just calculate the average of the input file?

  2. what's the dropout_transiton.

     and I found the reference in the website: dropout_transition The transition time, in seconds, for volume renormalization when an input stream ends. The default value is 2 seconds.. but I don't know what's the mean.

     if dropout_transiton=0, what would happened?

  Thanks!(sorry for my poor english)