从Azure Media Services中的现有资产重新创建缩略图

Question

I am using Azure Media Services and Azure Functions to build a VOD element for a website. Basically, when the source video is uploaded a blob trigger starts off a DurableOrchestration to create an asset and then encode the video. It also generates 3 different size thumbnails using the default {Best} frame. So far so good.

What I want to do now is allow the user to select a frame from the encoded video and choose that to be the poster thumbnail.

I have an HttpTrigger which takes the asset id and the frame timestamp and kicks off another durable function which should recreate the thumbnails at the specified frame.

But it isn't working.

I originally got 3 blank images in a new asset and when I tried to force it to put the images back into the original asset, I got nothing.

This is the code I'm using to try to achieve this. It's pretty much the same as the code to create the original asset. The only real difference is that the json preset only has instruction for generating thumbnails, the asset already has 6 encoded videos, 3 thumbnails and associated meta files in it, and I'm not passing the original source video file to it (because I delete that as part of the clean-up once the original encoding is complete).

        PostData data = inputs.GetInput<PostData>();

        IJob job = null;
        ITask taskEncoding = null;
        IAsset outputEncoding = null;
        int OutputMES = -1;
        int taskindex = 0;
        bool useEncoderOutputForAnalytics = false;
        MediaServicesCredentials amsCredentials = new MediaServicesCredentials();
        try
        {
            AzureAdTokenCredentials tokenCredentials = new AzureAdTokenCredentials(amsCredentials.AmsAadTenantDomain,
                                                    new AzureAdClientSymmetricKey(amsCredentials.AmsClientId, amsCredentials.AmsClientSecret),
                                                    AzureEnvironments.AzureCloudEnvironment);

            AzureAdTokenProvider tokenProvider = new AzureAdTokenProvider(tokenCredentials);

            _context = new CloudMediaContext(amsCredentials.AmsRestApiEndpoint, tokenProvider);
            IAsset asset = _context.Assets.Where(a => a.Id == data.assetId).FirstOrDefault();

            // Declare a new encoding job with the Standard encoder
            int priority = 10;
            job = _context.Jobs.Create("CMS encoding job", priority);

            foreach (var af in asset.AssetFiles)
            {
                if (af.Name.Contains(".mp4)"))
                    af.IsPrimary = true;
                else
                    af.IsPrimary = false;
            }

            // Get a media processor reference, and pass to it the name of the 
            // processor to use for the specific task.
            IMediaProcessor processorMES = MediaServicesHelper.GetLatestMediaProcessorByName(_context, "Media Encoder Standard");

            string preset = null;

            preset = "MesThumbnails.json";  // the default preset

            string start = data.frame;

            if (preset.ToUpper().EndsWith(".JSON"))
            {
                // Build the folder path to the preset
                string presetPath = Path.Combine(System.IO.Directory.GetParent(data.execContext.FunctionDirectory).FullName, "presets", preset);
                log.Info("presetPath= " + presetPath);
                preset = File.ReadAllText(presetPath).Replace("{Best}", start);
            }

            taskEncoding = job.Tasks.AddNew("rebuild thumbnails task",
               processorMES,
               preset,
               TaskOptions.None);

            // Specify the input asset to be encoded.
            taskEncoding.InputAssets.Add(asset);
            OutputMES = taskindex++;

            string _storageAccountName = amsCredentials.StorageAccountName;
            outputEncoding = taskEncoding.OutputAssets.AddNew(asset.Name + " MES encoded", _storageAccountName, AssetCreationOptions.None);

            asset = useEncoderOutputForAnalytics ? outputEncoding : asset;

            job.Submit();
            await job.GetExecutionProgressTask(CancellationToken.None);

My question is whether what I am trying to do is actually possible, and if so what is wrong with the approach I'm taking.

I've searched quite a bit on this topic but can always only find reference to generating thumbnails whilst encoding a video, never generating thumbnails from encoded videos after the event.

Answer 1

I'm not passing the original source video file to it That's likely why you are running into the problem. The output of your Adaptive Streaming Job, as you have seen, contains multiple files. There are some additional flags needed to tell the thumbnail generation Job to focus on just one file (typically the highest bitrate file). The preset below should do the trick.

1. Note how the preset starts with a Streams section which tells the encoder to pick the highest/top bitrate for video and audio
2. Note that the Step is set to 2, but range is 1, ensuring only one image is generated in the output

{
  "Version": 1.0,
  "Sources": [
    {
      "Streams": [
        {
          "Type": "AudioStream",
          "Value": "TopBitrate"
        },
        {
          "Type": "VideoStream",
          "Value": "TopBitrate"
        }
      ]
    }
  ],
  "Codecs": [
    {
      "Start": "00:00:03:00",
      "Step": "2",
      "Range": "1",
      "Type": "JpgImage",
      "JpgLayers": [
        {
          "Quality": 90,
          "Type": "JpgLayer",
          "Width": "100%",
          "Height": "100%"
        }
      ]
    }
  ],
  "Outputs": [
    {
      "FileName": "{Basename}_{Index}{Extension}",
      "Format": {
        "Type": "JpgFormat"
      }
    }
  ]
}