Measuring video quality is done daily around the world between two video assets. But what happens when you want to take the aggregate quality of a whole manifest? With VMAF being a well regarded metric, how can we use that in an automatic way to get the overview we need?
In this talk, Nick Chadwick from Mux shares the examples and scripts he’s been using to analyse videos. Starting with an example where everything is equal other than quality, he explains the difficulties in choosing the ‘better’ option when the variables are much less correlated. For instance, Nick also examines the situations where a video is clearly better, but where the benefit is outweighed by the minimal quality benefit and the disproportionately high bitrate requirement.
So with all of this complexity, it feels like comparing manifests may be a complexity too far, particularly where one manifest has 5 renditions, the other only 4. The question being, how do you create an aggregate video quality metric and determine whether that missing rendition is a detriment or a benefit?
Before unveiling the final solution, Nick makes the point of looking at how people are going to be using the service. Depending on the demographic and the devices people tend to use for that service, you will find different consumption ratios for the various parts of the ABR ladder. For instance, some services may see very high usage on 2nd screens which, in this case, may take low-resolution video and also lot of ‘TV’ size renditions at 1080p50 or above with little in between. Similarly other services may seldom ever see the highest resolutions being used, percentage-wise. This shows us that it’s important not only to look at the quality of each rendition but how likely it is to be seen.
To bring these thoughts together into a coherent conclusion, Nick unveils an open-source analyser which takes into account not only the VMAF score and the resolution but also the likely viewership such that we can now start to compare, for a given service, the relative merits of different ABR ladders.
The talk ends with Nick answering questions on the tendency to see jumps between different resolutions – for instance if we over-optimise and only have two renditions, it would be easy to see the switch – how to compare videos of different resolutions and also on his example user data.
How does the move to OTT delivery impact the traditional platforms? Are there too many streaming services? This session looks at the new platforms, the consumer experience, the role of aggregation and the way that operators have been involved in de-aggregation and then re-aggregation of channel packages both in competition and in cooperation.
How many subscription services are too many for a household? There’s some thinking that 3 may be the typical maximum when people tend to switch to a ‘one in, one out’ policy on subscription packages. Colin Dixon says the average is currently 2 in the UK and Germany. The panel asks whether we should have as many and compares the situation with audio where ‘super aggregation’ rules. Services like Apple Music and Spotify rely on aggregating ‘all’ music and consumers don’t subscribe separately to listen to Sony artists one on service and EMI on another, so what is it that drives video to be different and will it stay that way?
The topic then switches to smart TVs discussing the feeling that five to eight years ago they had a go at app stores and ended up disappointing. Not only was it often clunky at the time, but support has now gone on the whole from the manufacturers. Is the current wave of smart TVs any different? From BT’s perspective, explains Colin Phillips, it’s very costly to keep many different versions of app up to date and tested so a uniform platform across multiple TVs would be a lot better.
The talk concludes looking at the future for Disney+, Netflix and other providers ahead of discussing predictions from industry analysts.
What’s implementing SMPTE ST-2110 like in real life? How would you design your network and what were the problems? In this case study Ammar Latif from Cisco Systems presents the architecture, best practices and lessons learned they gleaned in this live IP broadcast production facility project designed for a major US broadcaster. Based on SMPTE ST-2110 standard, it spanned five studios and two control rooms. The central part of this project was a dual Spine-Leaf IP fabric with bandwidth equivalent of a 10,000 x 10,000 HD SDI router with a fully non-blocking multicast architecture. The routing system was based on Grass Valley Convergent broadcast controller and a Cisco DCNM media controller.
As the project was commissioned in 2018, the AMWA IS-04 and IS-05 specifications providing an inter-operable mechanism for routing media around SMPTE 2110 network were not yet available. Multicast flow subscription was based on a combination of IGMP (Internet Group Management Protocol) and PIM (Protocol Independent Multicast) protocols. While PIM is very efficient and mature, it lacks the ability to use bandwidth as a parameter when setting up a flow path. Ammar explains how Non-Blocking Multicast (NBM) developed by Cisco brings bandwidth awareness to PIM by signalling a type of data (video, audio or metadata).
The talk continues by discussing PTP distribution & monitoring, SMPTE 2022-7 seamless protection switching and remote site production. Ammar also lets us see how the user interfaces on the Cisco DCNM media controller were designed which include a visualisation of multicast flow, network topology and link saturation of ports.
There are many video fundamentals in today’s video looking at how we see light and how we can represent it in a video signal. Following on from last week’s look at analogue 525-line video we take a deeper dive in to light and colour.
The video starts by examining how white light can be split into colours, primaries, and how these can be re-combined in different amounts to create different colours. It then moves on to examining how the proportion of colours which create ‘white’ light isn’t as even as you might imagine. This allows us to understand how to create brighter and dimmer light which is called the luminance. We’re introduced to the CIE 2d and 3d colour graphs helping us to understand colour space and colour volume
Modern video, even if analogue, is acquired with red, green and blue as separate signals. This means if we want a grey-scale video signal, i.e. luminance only, we need to combine using the proportions discussed earlier. This biased version of luminance is what is called ‘luma’ explains the video from the Displaced Gamers YouTube Channel.
On top of human perception, much of the 20th century was dominated by CRT (Cathode Ray Tube) TVs, which don’t respond linearly to electrical voltage, meaning if you double the voltage, the brightness doesn’t necessary double. In order to compensate for that, ‘gamma correction’ is applied on acquisition so that playback on a CRT produces a linear response.
Pleasantly, an oscillator is wheeled out next looking at a real analogue video waveform demonstrating the shape of not only the sync pulses but the luminance waveform itself and how the on-screen rendition would be seen on a TV. The video then finishes with a brief look at colour addition NTSC, PAL, SECAM signals. A prelude, perhaps, to a future video.