A Short Guide to Choosing the Right Export Format For Your Images and Videos: Part 2

Dear tip crunchers, welcome to this week’s tip! Today we’ll finish the mini-series on exporting formats. Last week, we discussed the case of images: we saw some criteria to choose the proper kind of compression, and we reviewed the range of options offered by Amped products. Today it’s about video, so… keep reading!

Sometimes, the goal of your enhancement goes beyond producing a single still frame. Whether you’re working on a fight, a car accident, or a homicide case, it’s vital to see things in motion to understand the flow of events. In these cases, at the end of your enhancement, you have a video, and you need to export it.

The case of Amped Replay has already been discussed in this previous tip, and as of now, Amped Authenticate does not work on videos, so in this post, we’ll focus on exporting videos from Amped FIVE.

When you select the Video Writer filter, you are presented with a long list of possible output formats:

You may have noticed that we actually have some combinations of containers (avi, mkv, mp4, mov, wmv, and flv) and codecs (MSMPEG4v2/v3, MJPEG, MPEG4, H264, Raw Video, and Flash Video). The container will not affect the quality of your pixels but may affect the “compatibility” of your output with consumer players. The codec, instead, is the software that turns your frames into a coded stream so it really makes the difference. Amped FIVE allows you to choose between uncompressed and lossy compressed videos. However, we’ll see below that you can configure lossy compression to be very conservative.

If you need to preserve the exact pixel values, you can opt for the Raw Video codec: as the name says, it will write raw pixels without any compression. But be sure to have many gigabytes available, because uncompressed videos eat up loads of memory (it all depends on how long your video is, and its resolution), to the point that it’s not a viable solution in many cases. If you want something more compact, but still of high quality, then we recommend choosing H264, which enables the other dropdown list for selecting the quality:

If you set the Quality option to Visually Lossless, your compressed video will be very, very similar to the original pixels, meaning that, normally, the average difference in pixels luminance values will be smaller than one (which, when compared to the [0-255] range, gives us 0.4% average variation in pixel values), and the PSNR is normally well above 43 dB. 200 frames of a 1920×1080 video stored in Visually Lossless mode will take roughly 130MB, compared to the 600MBs needed to store the uncompressed pixels. If you need to save more space, you may consider the High quality setting: the average variation in pixels raises to 1.5 (PSNR still above 40), but the size drops to 36.8 MB! If you want to run your own tests, you may take one of the videos in the VISION dataset and use Amped FIVE’s Video Mixer to compare the original and recompressed videos, as explained here.

Ok, all of the above was about the compromise between quality and file size. But there’s another important element that must be considered, which is compatibility. If you’re exporting a video that’s going to another expert, compatibility is not a primary concern: they’ll likely have suitable tools (Amped ones, hopefully!) to decode it. But if the video is to be handed out to a media company or to the press office of your department, then it’s important to ensure the video will play on their computer! That’s the reason why the default output format in the Video Writer is avi-MSMPEG4v2: this combination of container and codec will play virtually on any computer you’ll find out there, even a legacy Windows XP with just Windows Media Player installed. The corresponding quality is similar to the H264High combination (average pixel difference below 2, PSNR around 40 dB), and the size is roughly 120 MB.

So, summarizing:

  • If the only actions you made were to review the video and select a range of frames, then you may choose to use Lossless trim (available inside the Range Selector filter) to export the original pixels into a standard container. You’ll save time and maintain the original stream. You can read more about lossless trim in the previous tip.
  • If you’ve enhanced the video and you are now producing an intermediate result that will be further processed (in Amped FIVE or in different software), then use RAW Video. You’ll likely need to trim a short portion of the video, to avoid eating up all your hard drive.
  • If you’re exporting the video to produce a possible piece of evidence, or to do minor editing (e.g., for demonstrative purposes in a Non-Linear Editor), you may go for H264 + Visually Lossless options.
  • If you need to balance quality and size, e.g. to send the video to a colleague to ask for their opinion on the result, you should be safe with H264 + High quality. This will be compatible with most modern devices, including smartphones and tablets.
  • If compatibility with older devices is of the essence, then the default avi – MSMPEG4v2 is probably the best option. Just be aware that this format may be no longer supported by the most recent portable devices.

And remember: you can always make your own tests using the Video Mixer‘s Similarity tab to compare the original and exported videos.

Before we say goodbye, one last tip: If you always need to send videos to a specific person or agency, then check with that person what they will be using and if possible go and evaluate yourself. You may find that you are exporting great evidence but then they have to transcode again to suit their software or Digital Evidence Management System!

That’s all for today! We hope you’ve found this issue of the Video Evidence Pitfalls series interesting and useful! Stay tuned and don’t miss the next ones. You can also follow us on LinkedIn, YouTube, Twitter, and Facebook: we’ll post a link to every new Tip Tuesday so you won’t miss any!