With this post, I’ll be wrapping up a multi-part discussion about measuring video quality. In the first part, I described Mean Opinion Scores and how these scores provide a quantitative value of multimedia quality. Last time, I spoke about “full-reference” and “reference-free” measurement techniques, and gave some advantages of each.

Today, we discuss ways that mobile video quality problems may be introduced, some of the more common types of quality issues, and some of the causes behind these.

Most video quality problems may be introduced at several points along the video distribution chain:

During video creation. While most professionally-produced video content is created by trained technicians using expensive devices, most user-generated video content is recorded with poor-quality cameras. In addition, the person doing the recording will often have shaky hands. This leads to quality problems that are introduced as soon as the video is created.

During transcoding. While today’s algorithms for encoding and decoding video content are quite sophisticated and very good at preserving quality, the nature of compression causes some loss of quality each time content goes through a coding cycle. Therefore, the number of encodes and decodes should be kept to a minimum.  However, in the real world, content must often be translated (or “transcoded”) between algorithms.  This can be because different devices support different algorithms, or because one algorithm might be great for display quality while another is better for transport efficiency.  In today’s telecommunications networks, it is common for video content to be transcoded several times between creation and delivery; each of these transcodes can contribute to reducing video quality.

During video transmission. IP networks suffer from impairments such as packet loss and jitter. While these problems have affected voice services for years, the human ear can recover from communication gaps better than the human eye. Even a small glitch in a video communications stream can cause distraction for a viewer.

When displaying video on a device. Even the best video content can be unsatisfying if it is displayed on a poor viewing device. Most of today’s mobile devices were designed to provide a high-quality voice experience, while video capabilities may be limited. Higher-quality video displays require a lot of power, so device makers may consider tradeoffs like offering extended battery life by sacrificing display quality in a mobile device.

While the previous paragraphs discussed how quality problems may be introduced during the process of video creation and distribution, the next will describe several of the more common types of problems that may be experienced (and some of the potential causes of each):

  • Video blockiness is an impairment in which the image contains artifacts that resemble small blocks of a single color. Blockiness usually occurs when the video encoder software is not able to process the video data stream accurately and instead uses a single “mean value” to represent all pixels in a large area.
  • Video blurriness occurs either when the video is being recorded (e.g. poor camera settings) or during encoding. It is commonly caused by bandwidth constraints in the network – the number of bits available is less than the number required to process the video accurately. This is especially common in mobile video when the content of the video contains fast action or many scene changes. (You can see a demonstration of this impairment on Dialogic’s video demonstration portal.)
  • Video freezing is usually caused by insufficient bandwidth. The video image will just suddenly appear “frozen” in time. On a laptop, the viewer may see an indication that the video is “buffering”; on a mobile device, the video may just freeze temporarily until enough bits are received to continue the stream.
  • Video jerkiness is a situation that occurs when a video stream does not flow smoothly but has quick starts and stops. There are many potential causes of video jerkiness: limited bandwidth (so the stream will freeze quite often for very short periods), network jitter problems that cause problems with buffering algorithms, or even a reduction in the frame rate of the displayed video.  (Note that frame rate reductions may be a natural response of compression algorithms when bandwidth is reduced; rather than keep a high frame rate with frequent buffering, the algorithms can be designed to stream the video continually but at a lower frame rate.)
  • Video blackout is a sudden and complete loss of signal, resulting in a black screen. This is most frequently caused either by some catastrophic network failure, a dropped call, or an improper handoff between cell sites.
  • Audio/video synchronization problems, or “lip sync” issues, can be very annoying to viewers. The soundtrack of the video is out of sync with the visual track. The audio track and the video track are often sent separately to the mobile device, and bandwidth constraints can slow down one track and not the other. Another potential cause is poor video capture (e.g. a microphone not working at time of video creation).

I hope that these last few posts have been helpful as a reference to any of our customers (or potential customers) that are considering tools and systems for measuring video quality. The increasing focus on video quality, and particularly the quality of mobile video, is an important step forward for our industry. At Dialogic, we welcome this increased attention to quality, as our business is focused on providing our customers with the highest-quality, highest-performance video solutions. Please stay tuned to this space in the next few months, and you’ll be hearing more about our efforts related to video quality measurement and improvement.

In the meantime, if you would like to learn more about this topic, we have a white paper available for free download here. And please take a moment to let us know what you are thinking about this subject – I’d enjoy the opportunity to continue the discussion!