24.7.3 Optimize video files
By way of introduction it should be mentioned that the optimization of video files only works with the MP4 format with sub-variant 'h264' and the program should only be called if there is a sound knowledge of how the h264 format itself is structured or which optimization is useful. If the corresponding video files are optimized with wrong or too 'aggressive' parameters, the result will be destroyed video files.
But what is optimization all about? MP4 files can be created and distributed in completely different resolutions or quality levels. First of all, there is the frame rate, i.e. the number of frames that are played per second. Here there are for example 20,24,25,30,50,60 or even more frames that can be contained in the files. The more images are stored per second, the larger the corresponding video files will be. It makes little sense to optimize from the bottom up. It makes little sense to upscale from 25 to 50 frames. However, in the opposite case, when scaling down from 50 to 25 frames, files are created that only require plus/minus half of the original storage space. A video file with 50 frames per second can take up to 6 GBytes even in HD quality (1280x720). Optimizing to 25 frames saves many GBytes of data as the number of files increases.
It can also be useful to scale from 4K (3840x2160) to FullHD (1920x1080) or even HD (1280x720. A FullHD video file, which is about 6 GBytes, even usually requires significantly less than half of the original space after reduction to HD quality. Of course, the question can be asked here what sense it makes to reduce the resolution. However, many TV stations now publish content in FullHD, even if they are older broadcasts that were recorded at 720x576 at the time. A reduction at least to HD (1280x720 always makes sense, because FullHD quality does not exist, it is simply used at the moment because the broadcasters themselves broadcast their programs either in HD or FullHD.
The same question arises with the quality itself. The so-called 'crf' (Constant Rate Factor) specifies how much loss should be applied to individual pixels. Value '0' in the optimization means that the quality will not be changed. Values between 1 and 50 mean that the lower the value, the higher the quality, but also the larger the file will be. Values above 23 mean that the quality will be worse than before. Values above 23 can hardly be recommended. However, it can be useful to optimize extremely large video files with the value '23' to a tolerable level. The file size can become smaller by several factors.
Finally, the value 'Buffer' is necessary if video files cannot be optimized at all without this value. After processing several 10'000 files, it can be added here, for about 1 to 3 per thousand of the files a buffer value had to be specified.
After the start of the optimization the desired files are processed. This process is extremely computationally intensive. Even with ArchivistaEverest, 'only' about 1000 hours can be rasterized per day. With ArchivistaK2 it is approx. 250 hours and with ArchivistaDom it is still approx. 80 to 100 hours. On feature film length, this results in approx. 700 films for ArchivistaEverst, approx. 150 for ArchivistaK2 and still approx. 50 films for ArchivistaDom.
A backup copy before starting the optimization is highly recommended, as the results of the optimization can only be examined when the corresponding video file is now available in an optimized version. The originals of the optimization are kept with the same name in the directory 'archhigh' in the path '/home/data/archivista'. This means that the optimization can only be performed if there is enough space for the originals and reduced files during the optimization. The files in the 'archhigh' directory must be deleted manually.