Recherche avancée

Sur d’autres sites (7620)

• How to merge the raw pcm data of two channels ?

  15 mai 2019, par maximedupre

  I have two Float32Arrays containing the raw pcm data of the left and right channel. Is it possible to create a Float32Array that combines both channel ? If so, how would I do that ? Can I simply concatenate the arrays ?

  Cheers !

• Mapping v4l2 pixel format codes to ffmpeg pixel format names

  12 décembre 2019, par DekiChan

  I’m building an app that is capturing frames from any v4l2 supported webcam. I’m controlling the camera with v4l2 (golang package https://github.com/blackjack/webcam) and one of the things I need is pixel format. However, what I get is uint32 representing the format in v4l2.

  I do the streaming by wrapping ffmpeg command and reading from its stdout :

  args := []string{
  
      "-f", "v4l2",
  
      "-c:v", "rawvideo",
  
      "-pix_fmt", pixFmt,
  
      "-s", size,
  
      "-i", input,
  
      "-f", "image2pipe",
  
      "-c:v", "mjpeg",
  
      "-qscale:v", "5",
  
      "-", // pipes to stdout
  
  }
  
  
  
  cmd := exec.Command("ffmpeg", args...)

  Now the problem is that -pix_fmt option requires format name in string that is recognized by ffmpeg (for example 8-bit Bayer RGGB8 corresponds to bayer_rggb8, while this format in v4l2 has code 1111967570).

  I found the mapping code for it in ffmpeg’s libavdevice : https://github.com/FFmpeg/FFmpeg/blob/master/libavdevice/v4l2-common.h includes the functions I need, but somehow I’m not able to include <libavdevice></libavdevice>v4l2-common.h> in CGo (no problems with <libavdevice></libavdevice>avdevice.h>) :

  package conv
  
  
  
  /*
  
      #cgo pkg-config: libavdevice
  
      #include <libavdevice></libavdevice>avdevice.h>
  
      #include <libavdevice></libavdevice>v4l2-common.h>
  
  */
  
  import "C"
  
  
  
  type (
  
  )
  
  
  
  // PixFmtToFF takes v4l2 pixel format code and converts it to a string
  
  // recognized by ffmpeg command
  
  // example: 1111967570 to "bayer_rggb8"
  
  func PixFmtToFF(v4l2code uint32) (string, error) {
  
      // should work something like this, couldn't test
  
      codecID := C.ff_fmt_v4l2codec(v4l2code)
  
      pixFmt := C.ff_fmt_v4l2ff(vfd.pixelformat, codec_id)
  
      return C.av_get_pix_fmt_name(pixFmt)
  
  }

  Build output returns :

  go build \
  
          -mod=vendor \
  
          -o ./bin/camera-api \
  
          ./cmd/camera-api
  
  # kca/internal/camera/conv
  
  internal/camera/conv/convert.go:6:11: fatal error: libavdevice/v4l2-common.h: No such file or directory
  
    #include <libavdevice></libavdevice>v4l2-common.h>
  
             ^~~~~~~~~~~~~~~~~~~~~~~~~~~
  
  compilation terminated.
  
  make: *** [Makefile:13: build/camera-api] Error 2
  
  make: *** [builder/camera-api] Error 2

  Output of pkg-config --libs libavdevice is -L/usr/local/lib -lavdevice.

  So, I have two questions :

  1) How could I include v4l2-common.h so I can use the mapping functions ?

  2) Is there a better way ? I need ffmpeg wrapping so we can support wide range of formats for captured frames.

• cut hls files in to many small ".ts" files or keep one big ".ts" file. which is better ?

  5 août 2022, par Evan Lee

  I need to stream my videos using HLS bytes range HTTP requests.

  


  FFmpeg has an option to keep all the ts file to one single large ".ts" file.

  


  Any pros and cons of split the ts files or keep a big ts file ?

  


  Is a big ts file make the request slower ? because of HDD fseek is slow ?