Recherche avancée

Sur d’autres sites (4248)

• H.264 and VP8 for still image coding : WebP ?

  1er octobre 2010, par Dark Shikari — H.264, VP8, google, psychovisual optimizations

  JPEG is a very old lossy image format. By today’s standards, it’s awful compression-wise : practically every video format since the days of MPEG-2 has been able to tie or beat JPEG at its own game. The reasons people haven’t switched to something more modern practically always boil down to a simple one — it’s just not worth the hassle. Even if JPEG can be beaten by a factor of 2, convincing the entire world to change image formats after 20 years is nigh impossible. Furthermore, JPEG is fast, simple, and practically guaranteed to be free of any intellectual property worries. It’s been tried before : JPEG-2000 first, then Microsoft’s JPEG XR, both tried to unseat JPEG. Neither got much of anywhere.

  Now Google is trying to dump yet another image format on us, “WebP”. But really, it’s just a VP8 intra frame. There are some obvious practical problems with this new image format in comparison to JPEG ; it doesn’t even support all of JPEG’s features, let alone many of the much-wanted features JPEG was missing (alpha channel support, lossless support). It only supports 4:2:0 chroma subsampling, while JPEG can handle 4:2:2 and 4:4:4. Google doesn’t seem interested in adding any of these features either.

  But let’s get to the meat and see how these encoders stack up on compressing still images. As I explained in my original analysis, VP8 has the advantage of H.264′s intra prediction, which is one of the primary reasons why H.264 has such an advantage in intra compression. It only has i4x4 and i16x16 modes, not i8x8, so it’s not quite as fancy as H.264′s, but it comes close.

  The test files are all around 155KB ; download them for the exact filesizes. For all three, I did a binary search of quality levels to get the file sizes close. For x264, I encoded with --tune stillimage --preset placebo. For libvpx, I encoded with --best. For JPEG, I encoded with ffmpeg, then applied jpgcrush, a lossless jpeg compressor. I suspect there are better JPEG encoders out there than ffmpeg ; if you have one, feel free to test it and post the results. The source image is the 200th frame of Parkjoy, from derf’s page (fun fact : this video was shot here ! More info on the video here.).

  Files : (x264 [154KB], vp8 [155KB], jpg [156KB])

  Results (decoded to PNG) : (x264, vp8, jpg)

  This seems rather embarrassing for libvpx. Personally I think VP8 looks by far the worst of the bunch, despite JPEG’s blocking. What’s going on here ? VP8 certainly has better entropy coding than JPEG does (by far !). It has better intra prediction (JPEG has just DC prediction). How could VP8 look worse ? Let’s investigate.

  VP8 uses a 4×4 transform, which tends to blur and lose more detail than JPEG’s 8×8 transform. But that alone certainly isn’t enough to create such a dramatic difference. Let’s investigate a hypothesis — that the problem is that libvpx is optimizing for PSNR and ignoring psychovisual considerations when encoding the image… I’ll encode with --tune psnr --preset placebo in x264, turning off all psy optimizations. 

  Files : (x264, optimized for PSNR [154KB]) [Note for the technical people : because adaptive quantization is off, to get the filesize on target I had to use a CQM here.]

  Results (decoded to PNG) : (x264, optimized for PSNR)

  What a blur ! Only somewhat better than VP8, and still worse than JPEG. And that’s using the same encoder and the same level of analysis — the only thing done differently is dropping the psy optimizations. Thus we come back to the conclusion I’ve made over and over on this blog — the encoder matters more than the video format, and good psy optimizations are more important than anything else for compression. libvpx, a much more powerful encoder than ffmpeg’s jpeg encoder, loses because it tries too hard to optimize for PSNR.

  These results raise an obvious question — is Google nuts ? I could understand the push for “WebP” if it was better than JPEG. And sure, technically as a file format it is, and an encoder could be made for it that’s better than JPEG. But note the word “could”. Why announce it now when libvpx is still such an awful encoder ? You’d have to be nuts to try to replace JPEG with this blurry mess as-is. Now, I don’t expect libvpx to be able to compete with x264, the best encoder in the world — but surely it should be able to beat an image format released in 1992 ?

  Earth to Google : make the encoder good first, then promote it as better than the alternatives. The reverse doesn’t work quite as well.

  [155KB]

• Handling correctly the ffmpeg & ffprobe with php

  29 septembre 2014, par cocco

  Handling correctly the ffmpeg & ffprobe with php

  maybe not relevant final goals :

  1. upload clip with ajax
  2. get ajax info from ffprobe using php as json executing ffprobe once only (no ffmpeg)
  3. handle all calculations with javascript
  4. maybe an extra php script tool that can create gifs, extract frames(thumbs), or a video grid preview
  5. when rdy ajax the conversion info to the final php conversion script executing ffmpeg once only (just the final ffmpeg string.).

  I’m trying to write my own ffmpeg local web video editor that converts all formats to mp4 automatically. As mp4 is the most compatible container now and the h264+aac/+ac3 is also one of the best compressions. I also want to be able to cut, crop, resize, remove streams, add streams and more. I’m stuck on some simple problems :

  1. HOW TO GET THE INFO ?

  I’m using ffprobe to get the file information as json with the following command :

  ffprobe -v quiet -print_format json -show_format -show_streams -show_packets '.$video

  this gives you a lot of information, but some relevant stuff is not always present. I need the duration (in milliseconds),the fps (as a float) and the total frames (as an integer).

  i know that these values can sometimes be found inside this array :

  format.duration //Total duration
  
  streams[0].duration //Video duration
  
  streams[1].duration //Audio duration
  
  
  
  streams[0].avg_frame_rate //Average framerate
  
  streams[0].r_frame_rate //Video framerate
  
  
  
  streams[0].nb_frames //Total frames

  but most of the time nb_frames is missing, also avg_frame_rate differs from r_frame_rate, which is also not always available.

  I know that i could use multiple commands to increase the chance to get the correct values.. but srsly ???

  //fps
  
  ffmpeg -i INPUT 2>&1 | sed -n "s/.*, \(.*\) fp.*/\1/p" 
  
  //duration
  
  ffmpeg -i INPUT 2>&1 | awk '/Duration/ {split($2,a,":");print a[1]*3600+a[2]*60+a[3]}'
  
  //frames
  
  ffmpeg -i INPUT -vcodec copy -f rawvideo -y /dev/null 2>&1 | tr ^M '\n' | awk '/^frame=/ {print $2}'|tail -n 1

  I don’t want to execute ffmpeg 3 times to get this information ; I’d prefer to just use ffprobe.

  So... is there an elegant way to get the extra info that is not always present inside the ffprobe output (fps, frames, duration) ???

  In the preview i want to be able to jump correctly to a specific frame (NOT TIME). if the above parameters are aviable i can do that using this command.

  ffmpeg -i INPUT -vf 'select=gte(n\,FRAMENUMBER)' -vframes 1 -f image2 OUTPUT

  using the above command by setting the framenumber to the last frame always returns a black frame.
  if there are 50 frames (for example) the range is 1-50 — correct ? Frame 50 is black, frame 1 is ok, frame 0 returns an error...

  ---

  2. WHILE READING THE LOG HOW TO SKIP ERRORS AND DETERMINE IF THE CONVERSION IS FINISHED ?

  I’m able to upload one single video per time (per page) and i can read the current progress from the ffmpeg generated output log until i don’t close the page. more control/multiple conversions would be nice.

  i’m reading the last line of the log with a custom tail function but as this is a log that also includes errors i don’t always get a nice line containing the desidered values. btw to check if the progress is complete i check if the last line CONTAINS the WORD frame ....

  How can i find out when the conversion progress is finished ?

  maybe a way to delete the log with ffmpeg command ??And skip/log the errors ??

  i’m using server sent events to read the log...
  here is the php code

  <?php
  
  setlocale(LC_CTYPE, "en_US.UTF-8");
  
  function tailCustom($filepath,$lines=1,$adaptive=true){
  
   // a custom function to get the last line of a textfile.
  
  }
  
  function send($data){
  
   echo "id: ".time().PHP_EOL;
  
   echo "data: ".$data.PHP_EOL;
  
   echo PHP_EOL;
  
   ob_flush();
  
   flush();
  
  }
  
  header('Content-Type: text/event-stream');
  
  header('Cache-Control: no-cache');
  
  while(true){
  
   send(tailCustom($_GET['log'].".log"));
  
   sleep(1);
  
  }
  
  ?>

  And here the SSE js

  function startSSE(fn){
  
   sse=new EventSource("ffmpegProgress.php?log="+encodeURIComponent(fn));
  
   sse.addEventListener('message',conversionProgress,false);
  
  }
  
  function conversionProgress(e){
  
   if(e.data.substr(0,6)=='frame='){
  
    inProgress=true;
  
    var x=e.data.match(/frame=\s*(.*?)\s*fps=\s*(.*?)\s*q=\s*(.*?)\s*size=\s*(.*?)\s*time=\s*(.*?)\s*bitrate=\s*(.*?)\s*$/);
  
    x.shift();x={frame:x[0]*1,fps:x[1]*1,q:x[2],size:x[3],time:x[4],bitrate:x[5]};
  
  var elapsedTime = ((new Date().getTime()) - startTime);
  
  var chunksPerTime = timeString2ms(x.time) / elapsedTime;
  
  var estimatedTotalTime = duration / chunksPerTime;
  
  var timeLeftInSeconds = Math.abs(elapsedTime-(estimatedTotalTime*1000));
  
  var withOneDecimalPlace = Math.round(timeLeftInSeconds * 10) / 10;
  
    conversion.innerHTML='Time Left: '+ms2TimeString(timeLeftInSeconds).split('.')[0]+'<br />'+
  
    'Time Left2: '+(ms2TimeString(((frames-x.frame)/x.fps)*1000)+(timeString2ms(x.time)/(duration*1000)*100|0)).split('.')[0]+'<br />'+
  
    'Estimated Total: '+ms2TimeString(estimatedTotalTime*1000).split('.')[0]+'<br />'+
  
    'Elapsed Time: '+ms2TimeString(elapsedTime).split('.')[0];
  
   }else{
  
    if(inProgress){
  
     sse.removeEventListener('message',conversionProgress,false);
  
     sse.close();
  
     sse=null;
  
     conversion.textContent='Finished in '+ms2TimeString((new Date().getTime()) - startTime).split('.')[0];
  
     //delete log/old file??
  
     inProgress=false;
  
    }
  
   }
  
  }

  EDIT

  HERE IS A SAMPLE OUTPUT after detecting h264 codec in a m2ts with ac3 audio

  As most devices can already read h264 i just need to convert the audio in aac and copy the same audio AC3 as second track. and put everything inside a mp4 container. So that i have a Android/chrome/ios & more browsers compatible file.

  $opt="-map 0:0 -map 0:1 -map 0:1 -c:v copy -c:a:0 libfdk_aac -metadata:s:a:0 language=ita -b:a 128k -ar 48000 -ac 2 -c:a:1 copy -metadata:s:a:1 language=ita -movflags +faststart";
  
  
  
  $i="in.m2ts";
  
  $o="out.mp4";
  
  $t="title";
  
  $y="2014";
  
  $progress="nameoftheLOG.log";
  
  
  
  $cmd="ffmpeg -y -i ".escapeshellarg($i)." -metadata title=".$t." -metadata date=".$y." ".$opt." ".$o." null >/dev/null 2>".$progress." &";

  if you have any questions about the code or want to see more code just ask...

• Make better marketing decisions with attribution modeling

  19 décembre 2017, par InnoCraft

  Do you suspect some traffic sources are not getting the rewards they deserve ? Do you want to know how much credit each of your marketing channel actually gets ?

  When you look at which referrers contribute the most to your goal conversions or purchases, Matomo (Piwik) shows you only the referrer of the last visit. However, in reality, a visitor often visits a website multiple times from different referrers before they convert a goal. Giving all credit to the referrer of the last visit ignores all other referrers that contributed to a conversion as well.

  You can now push your marketing analysis to the next level with attribution modeling and finally discover the true value of all your marketing channels. As a result, you will be able to shift your marketing efforts and spending accordingly to maximize your success and stop wasting resources. In marketing, studying this data is called attribution modeling.

  Get the true value of your referrers

  Attribution is a premium feature that you can easily purchase from the Matomo (Piwik) marketplace.

  Once installed, you will be able to :

  • identify valuable referrers that you did not see before
  • invest in potential new partners
  • attribute a new level of conversion
  • make this work very easily by filling just a couple of form information

  Identify valuable referrers that you did not see before

  You probably have hundreds or even thousands of different sources listed within the referrer reports. We also guess that you have the feeling that it is always the same referrers which are credited of conversions.
  Guess what, those data are probably biased or at least are not telling you the whole story.
  Why ? Because by default, Matomo (Piwik) only attributes all credit to the last referrer.

  

  It is likely that many non credited sources played a role in the conversion process as well as people often visit your website several times before converting and they may come from different referrers.

  This is exactly where attribution modeling comes into play. With attribution modeling, you can decide which touchpoint you want to study. For example, you can choose to give credit to all the referrers a single visitor came from each time the user visits your website, and not only look at the last one. Without this feature, chances are, that you have spent too much money and / or efforts on the wrong referrer channels in the past because many referrers that contributed to conversions were ignored. Based on the insights you get by applying different attribution models, you can make better decisions on where to shift your marketing spending and efforts.

  Invest in potential new partners

  Once you apply different attribution models, you will find out that you need to consider a new list of referrers which you before either over- or under-estimated in terms of how much they contributed to your conversions. You probably did not identify those sources before because Matomo (Piwik) shows only the last referrer before a conversion. But you can now also look at what these newly discovered referrers are saying about your company, looking for any advertising programs they may offer, getting in contact with the owner of the website, and more.

  Apply up to 6 different attribution models

  By default, Matomo (Piwik) is attributing the conversion to the last referrer only. With attribution modeling you can analyze 6 different models :

  • Last Interaction : the conversion is attributed to the last referrer, even if it is a direct access.
  • Last Non-Direct : the conversion is attributed to the last referrer, but not in the case of a direct access.
  • First Interaction : the conversion is attributed to the first referrer which brought you the visit.
  • Linear : whatever the number of referrers which brought you the conversion, they will all get the same value.
  • Position Based : first and last referrer will be attributed 40% each the conversion value, the remaining 60% is divided between the rest of the referrers.
  • Time Decay : this attribution model means that the closer to the date of the conversion is, the more your last referrers will get credit.

  Those attribution models will enable you to analyze all your referrers deeply and increase your conversions.

  Let’s look at an example where we are comparing two models : “last interaction” and “first interaction”. Our goal is to identify whether some referrers that we are currently considering as less important, are finally playing a serious role in the total amount of conversions :

  

  Comparing Last Interaction model to First Interaction model

  Here it is interesting to observe that the website www.hongkiat.com is bringing almost 90% conversion more with the first interaction model rather than the last one.

  As a result we can look at this website and take the following actions :

  • have a look at the message on this website
  • look at opportunities to change the message
  • look at opportunities to display extra marketing messages
  • get in contact with the owner to identify any other communication opportunities

  The Multi Channel Attribution report

  Attribution modeling in Matomo (Piwik) does not require you to add any tracking code. The only thing you need is to install the plugin and let the magic happen.
  Simple as pie is the word you should keep in mind for this feature. Once installed, you will find the report within the goal section, just above the goals you created :

  

  The Multi Attribution menu

  There you can select the attribution model you would like to apply or compare.

  Attribution modeling is not just about playing with a new report. It is above all an opportunity to increase the number of conversions by identifying referrers that you may have not recognized as valuable in the past. To grow your business, it is crucial to identify the most (and least) successful channels correctly so you can spend your time and money wisely.

  Learn more about Multi Channel Conversion Attribution

  The post Make better marketing decisions with attribution modeling appeared first on Analytics Platform - Matomo.