Testing an AG's Video

Methods to easily test and improve the quality of your AG video.


For remote participants to effectively interact with people at your Access Grid node, it is essential to produce high-quality outgoing video. This section of the guide demonstrates simple methods to test and improve AG video quality. A number of video problems are commonly observed in Access Grid sessions. These include, but are not limited to the following:

  • Poor Camera positioning of subjects; and

  • Poor picture quality

Generally speaking, most Access Grids produce reasonable quality video. This is largely due to feedback - operators are used to viewing their outgoing video streams in order to judge camera-positioning, zoom level, etc. (It's worth asking, why don't operators similarly monitor audio? Sound is arguably more important.) In order to fix problems, we must be able to observe them.

Video Quality Testing

To test your video quality, bring up video windows showing your outgoing streams. For AG nodes that are (i) producing video with the "vic" video tool, and (ii) running both video producers and consumers on the same machine as a single Access Grid node, you will have to select and un-mute the video stream video producer service to be able to view outgoing video.


Problems and Possible Solutions

Poor Camera Positioning;

Proper camera positioning - "framing" in broadcast parlance - is easily observed by watching a typical news-style show on television. Note how cameras are typically positioned to show "head and shoulder" shots of the presenters. Node operators would do well to emulate this aspect of TV production.

Unfortunately, it is quite common to attend AG sessions where a single camera is used to show 3 or more people. With low-resolution video, the result is that relatively few pixels are used for each face. This makes it difficult for remote viewers (your audience) to discern facial expressions, akin to having a conversation with someone standing 6 meters away. If groups must be shown, people should sit close together with a maximum of 2-3 people per camera.

Further, it is recommended that a remote-controlled camera be used for the “Presenter” video stream so that different camera angles and zoom levels can be used without disturbing the session. The images below provide examples of a video stream with poor and good video framing of a presenter.


Note, in addition to close-up views, it is useful to have a wide-angle stream showing the entire room, including the entrance/exit (see below). This allows remote viewers to have a mental model of how the close-up shots relate to one another. It also allows viewers to know if and when someone enters or exits the AG node.


Poor picture quality;

Picture quality is affected by a number of common issues:

  • focus

  • lighting (both subject and overall room lighting)

  • flicker

  • contrast

The contrast between the background and speakers in the foreground should be good, i.e. the presenter should easily be recognised and the lighting should be acceptable so that the face is highlighted and not too dark or washed out because of the background causing the presenters to fade into the background.

Presenters should not be placed in front of bright backgrounds (such as those with a projected image in the background), as the system may try to overcompensate and cause the presenters' images to become too dark. Alternatively the picture quality on some cameras will cause the focus to "hunt" thereby causing the picture image to continually focus in and out. Images should also be stable and not flicker.

In the image below, the video is out of focus. While this is an extreme example, it is suprising how often "out of focus" cameras are encountered. Auto-focus cameras can often fail if the room lighting is too low. Manual focus works well but requires operator attention. As a node operator, you should frequently check the quality of your cameras - before, and during, a session.


In the images below, the room lighting is very uneven. Bright spots appear on the (reflective) desktop while the rest of the room is in relative darkness. This may be comfortable for room participants but is far from ideal for remote viewers. This would be highly problematic if people were seated behind the primary presenter. Remember that you are both the audience and presenter - remote viewers need to see you clearly.



Cameras typically have a very restricted dynamic range. In the image below, the bright natural light streaming through a gap in the curtains affects the video contrast. (A separate question is the utility of this medium-range shot. The camera is too far for person-to-person conversation but too close to give us an overall sense of the space.)


The image below is a great example of how a bright background can cause the presenter’s image to be dark and difficult to see. Avoid this scenario.



login or register to post comments