Recherche avancée

Sur d’autres sites (6341)

• Encode video of powerpoint presentation for HTML5 playback

  17 avril 2013, par user2291446

  We have a number of powerpoint presentations that have been converted to 16:9
  aspect ratio and then converted into mp4 "master videos" with an "apple TV" 720p
  profile. These powerpoint presentations are voice annotated. So in essence, we
  show a slide and then let the annotation sound play for a while, then go to the
  next slide, and so on. The resulting mp4 master video is somewhere around 900MB
  on average.

  Here is an example of the master video

  Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'input.mp4' :
  Metadata :
      major_brand : isom
      minor_version : 512
      compatible_brands : isomiso2avc1mp41
      creation_time : 1970-01-01 00:00:00
      encoder : Lavf52.104.0
    Duration : 02:00:57.65, start : 0.000000, bitrate : 970 kb/s
      Stream #0:0(und) : Video : h264 (Main) (avc1 / 0x31637661), yuv420p, 1280x720 [SAR 1:1 DAR 16:9], 836 kb/s, 29.97 fps, 29.97 tbr, 30k tbn, 59.94 tbc
      Metadata :
        creation_time : 1970-01-01 00:00:00
        handler_name : VideoHandler
      Stream #0:1(und) : Audio : aac (mp4a / 0x6134706D), 44100 Hz, stereo, fltp, 127 kb/s
      Metadata :
        creation_time : 1970-01-01 00:00:00
        handler_name : SoundHandler
  

  We are trying to get these presentations to play on the web on as many
  devices/browsers as possible including some that don't do HTML5 (IE7/IE8). We
  have narrowed down our player of choice which is mediaElement and have extracted
  some "cue points" from the powerpoint presentation that mark where the slides
  are changing. We have also captured thumbnails for those cuepoints such that we
  now have a nice list of thumbnails for each slide and an associated cuepoint in
  the video where the particular slide begins.

  Here comes the problem...due to the large size of the master video it is not
  practical for us to use the master video with our mediaElement player. We do
  need to transcode the master video to mp4 and ogv in order to get decent
  device/browser coverage.

  We do not seem to be able to find a suitable transcoding strategy to reduce the
  size of the video. We have played with numerous ffmpeg settings and were able to
  reduce the size but when we do so we compromise the ability to jump to specific
  cue points.

  It works well for browsers that do HTML5 video natively (Chrome and Firefox) but
  not for the flash fallback of mediaElement (IE7/IE8) which uses the mp4 file and
  seemingly is very tied to the number and frequency of key frames in the video in
  order to allow for clean seeking and skipping using the cue points.

  Seeing that we are talking about a video that has only slides (practically 90
  static images per presentation) and some sound we imagine it must be possible to
  transcode as such that the keyframes fall at the cue points or near the
  cuepoints, and that the size of the video could be drastically reduced while
  still allowing for smooth seeking and skipping.

• Is 'Android+FFMpeg' friendship really available ?

  26 janvier 2016, par vold_by

  The question does not mean that I’m interested if ffmpeg code can be used on Andoid. I know that it can. I’m just asking if somebody has the real performance progress with that stuff.
  I’ve created the question after several weeks of experiments with the stuff and I’ve had enough...
  I do not want to write to branches where people even do not say what kind of video they decode (resolution, codec) and talk only about some mystical FPS. I just don’t understand what they want to do. Also I’m not going to develop application only for my phone or for Android 2.2++ phones that have some extended OpenGL features. I have quite popular phone HTC Desire so if the application does not work on it, so what’s next ?

  Well, what do I have ?

  1. FFMpeg source from the latest HEAD branch. Actually I could not buld it with NDK5 so I decided to use stolen one.

  2. Bambuser’s build script (bash) with appropriate ffmpeg source ([web] : http://bambuser.com/r/opensource/ffmpeg-4f7d2fe-android-2011-03-07.tar.gz).
     It builds well after some corrections by using NDK5.

  3. Rockplayer’s gelded ffmpeg source code with huge Android.mk in the capacity of build script ([web] : http://www.rockplayer.com/download/rockplayer_ffmpeg_git_20100418.zip).
     It builds by NDK3 and NDK5 after some corrections. Rockplayer is probably the most cool media player for Android and I supposed that I would have some perks using it’s build.

  I had suitable video for a project (is not big and is not small) : 600x360 H.264.

  Both libraries we got from clauses 2 and 3 provide us possibility to get frames from video (frame-by-frame, seek etc.). I did not try to get an audio track because I did not need one for the project. I’m not publishing my source here because I think that’s traditional and it’s easy to find.

  Well, what’s the results with video ?
  HTC Desire, Android 2.2
  600x360, H.264
  decoding and rendering are in different threads

  1. Bambuser (NDK5 buld for armv5te, RGBA8888) : 33 ms/frame average.
  2. Rockplayer (NDK3 build for neon, RGB565) : 27 ms/frame average.

  It’s not bad for the first look, but just think that these are results only to decode frames.
  If somebody has much better results with decoding time, let me know.

  The most hard thing for a video is rendering. If we have bitmap 600x360 we should scale one somehow before painting because different phones have different screen sizes and we can not expect that our video will be the same size as screen.

  What options do we have to rescale a frame to fit it to screen ?
  I was able to check (the same phone and video source) those cases :

  1. sws_scale() C function in Bambuser’s build : 70 ms/frame. Unacceptable.
  2. Stupid bitmap rescaling in Android (Bitmap.createScaledBitmap) : 65 ms/frame. Unacceptable.
  3. OpenGL rendering in ortho projection on textured quad. In this case I did not need to scale frame. I just needed to prepare texture 1024x512 (in my case it was RGBA8888) containig frame pixels and than load it in GPU (gl.glTexImage2D). Result : 220 ms/frame to render. Unacceptable. I did not expect that glTexImage2D just sucked on Snapdragon CPU.

  That’s all.
  I know that there is some way to use fragment shader to convert YUV pixels using GPU, but we will have the same glTexImage2D and 200 ms just to texture loading.

  But this is not the end. ...my only friend the end... :)
  It’s not hopeless condition.

  Trying to use RockPlayer you definitely will wonder how they do that damn frame scaling so fast. I suppose that they have really good experience in ARM achitecture. They most probably use avcodec_decode_video2 and than img_convert (as I did in RP version), but then they use some tricks (depends of ARM version) for scaling.
  Maybe they also have some "magic" buld configuration for ffmpeg decreasing decoding time but Android.mk that they published is not THE Android.mk they use. Dunno...

  So, now it looks like you can not just buld some easy JNI bridge for ffmpeg and than have real media player for Android platform. You can do this only if you have suitable video that you do not need to scale.

  Any ideas ? I hope for you ;)

• Revision b79d5b62bd : Fixed VP9 denoiser COPY_BLOCK case Now copies the src to the correct location i

  14 juin 2014, par Tim Kopp

  Changed Paths :
   Modify /vp9/encoder/vp9_denoiser.c

  
  
  Fixed VP9 denoiser COPY_BLOCK case

  Now copies the src to the correct location in the running average buffer.

  Change-Id : I9c83c96dc7a97f42c8df16ab4a9f18b733181f34