We identified the following scenarios which anyone who has manned a Computing Services' Advisory Desk will no doubt recognise. In fact the scenarios probably have analogues in any advisory situation and are not necessarily only related to computing.
The videoconferencing facilities in this case can be used in such a way that the user shows the expert via the video link what is wrong and the advisor then uses the audio/video facilities to give advice. It may, of course, be appropriate for the expert to use the whiteboard or run the application which created the problem as a shared application so that the way the problem was created can be ascertained. However in the pilot study we wished to determine whether the basic video/audio capability was useful in its own right.
This is an obvious case for use of the shared application facility in which typically, following initial contact using the video/audio component, the user demonstrates the problem by launching the relevant application and re-creates the problem while the expert is watching. The expert then takes control of the application and shows the user the correct way of carrying out the task.
This sort of query can be handled by the user sketching out his problem on a shared whiteboard and the expert also using the board to then confirm/deny his/her impressions of what is required. When the participants are agreed on the requirement the expert can optionally use the shared application component to demonstrate how to achieve the effect with the appropriate application.
In the following we have tried to organise our findings into well-defined areas as much as possible to try to extract separate conclusions. In the study, however, all these different issues were all interacting to give an overall level of performance and 'feel' for how useful the tools were. We therefore also give some overall impressions at the end of the section.
The pilot study was carried out in two ways. Some experiments were carried out in adjacent offices. This helped greatly at first since we did find it necessary to run back and forth in order to sort out in person various minor problems we encountered with the software. This reflected partly our inexperience with Showme but is also a reminder of the superior efficiency of human contact. Subsequent experiments were carried out between the advisory desk and an office in a remote building. It is interesting to note that this was necessary in order to simulate real conditions since the temptation to pop one's head around the door when the experiments were done in adjacent offices was almost irresistible!
The address book features, when used in a LAN environment required the workstation address of the person being contacted to be coded into a file and hence would not be of much use in circumstances where users move around from workstation to workstation. This, of course, is the case in establishments where publically accessible workstations and PCs are available.
Showme also provides good monitoring and control of the parameters of video transmission, for example, the frame rate and number of colours in the picture. Adjusting these parameters can reduce the load on the network whilst maintaining an albeit inferior quality of live video.
If there is a requirement for showing documents to participants (as is the case for our first scenario) then a fixed, vertically-mounted camera with adjustable focus, under which documents could be placed for viewing, would be a very useful asset. Using the standard display-mounted camera is possible. However, such cameras have wide angle lenses and fixed focus, and are therefore of limited use for this purpose. If a second camera is to be used then the video board must accept two video inputs. This is in fact a feature of the Sunvideo board and Showme does have software controls for switching between the two inputs. For our study we did not actually set this up but it would be a fairly high-priority requirement for any Helpdesk set-up.
It should be noted that video boards associated with PC/Windows products that we have seen do not cater for two video input channels. It may be possible to overcome this limitation by some sort of hardware switch between two cameras and a workstation or by mounting the camera on a pedestal with a universal joint which would allow the camera to be easily redirected from the conferee to a fixed document.
We found that camera position was important both in terms of the framing of the person being viewed and in terms of the lighting. Some adjustments were necessary in order to cope with different times of day and lighting conditions. Lighting can vary from bright sunlight to darkness over a working day depending on both geographical orientations of different offices and weather conditions. This underlined the fact that a good system would provide flexible controls to control camera angles and contrast and brightness of video pictures. The Showme software did provide such flexibility.
The difficulties with the audio performance mean that some sort of dedicated audio link (telephone with hands-free operation or intercom) would be highly desirable for any live set up if a LAN was being used as the base carrier for conferencing. This would obviously add to the cost of the overall set-up and is somewhat unsatisfactory since synchronisation of the video and audio streams is then not possible.
The above problems emphasise one of the advantages of using ISDN as the base carrier. With ISDN video and audio traffic can be sent on separate 64 Kbit channels which are dedicated to the conference participants. Although the video performance with ISDN is relatively poor (compared with a LAN under favourable conditions), the sound quality is muchbetter than that obtained with ordinary analogue telephone lines.
We found that we soon developed short-cuts to communication using the various standard icons provided. The drawing tools were fairly primitive but allowed simple sketches to be drawn quickly. Of course if more complex or, say, accurate 3D models need to be shown then conference participants can cut and paste from a running application's window. For more complex interactions it is possible to launch a shared application such as, say, AutoCAD. In practice, however, we often found that this was impractical due to the excessive time it took to launch a shared application. This, in some ways, is a reflection on the current performance of our system, but it also raises some user interface issues which are discussed later.
We also found that applications were sometimes slow to start up and occasionally would not start at all. Such problems can obviously ruin a conference with a remote person. It is at times like these that the audio and video components can be very important since they give the conference participants good feedback on what is happening and, possibly, reassurance that something is being done about any problems that have arisen.
We experienced flickering and poor performance when running certain applications (AutoCAD in particular). This occurred on our SUN IPX workstations as the graphics cursor was moved over the screen. We also experienced colour flashing which is caused by reallocation of colour table entries by the Shared Application component. These are documented problems (in the Showme manual) associated with the windowing system (in our case X11) and its use of colours from the colour palette. They seem to indicate the inadequacy of the X-Windows environment to cope with this form of working. Such problems obviously greatly reduce the effectiveness of videoconferencing and, in our opinion, mean that application sharing will not be for the masses until the problems are sorted out.
The Shared Application component is arguably the potentially most powerful feature of videoconferencing suites but it is essential that its implementation presents a smooth, seamless integration with the other components of the software if it is to be used by non-specialists.
It should be stressed here that the above comments are based only on experience with our local hardware/software set-up. Such problems will not necessarily occur in other environments. It need hardly be added that the above emphasises how important it is to evaluate any potential products for usability in the exact environment they are going to be used before going too far towards setting up a live system.
For the Showme software, we felt the need to produce crib sheets which listed procedures required frequently but which were not always obvious from the menu systems. This would be especially important in our environment where advisory experts man the advisory desk possibly once a month or less. The sheets are included in Appendix 1.
The video performance, both in terms of speed and clarity, was far below the level at which lip reading could be used for communication. In our environment video frames were dropped frequently by the network and this, of course, rendered lip-reading impossible. The video performance was limited not only by the network technology but also by the Sunvideo card we were using since even when the network was relatively 'empty' the video was far below the quality at which it would be possible to discern lip and facial movements well enough to decipher speech. In fact near broadcast quality video would be required for this. As explained in the previous chapter the Sunvideo card captures and compresses the video stream and each workstation decompresses the video stream totally in software. With today's network bandwidths and the consequent need for high compression ratios for video streams we are a long way away from seeing broadcast quality video pictures in desktop videoconferencing environments. Thus we cannot see that desktop videoconferencing products will be available even in the medium term which would provide sufficient quality of video to support lip reading.
We did find during the study that, even with relatively poor quality video, we began to use a very primitive sign language (such as a thumbs-up and thumbs-down signs) which allowed us to speed up some communication we would otherwise have done via the whiteboard. Formal sign language however requires rapid hand movements and would thus be beyond the capabilities of present desktop videoconferencing systems.
The problems of establishing initial contact in order to start a conference are obviously increased when the remote expert being contacted cannot use the telephone. Thus it is very important to have a fixed time during which the expert is close to a workstation running the videoconferencing software in listening mode and is available for consultation. The Showme software actually pops up a window alerting a potential participant that a conference is being requested, simultaneously emitting an audible alarm. This would not be much good to a person with hearing difficulties who was not watching the screen since they would be unlikely to hear the alarm. The only partial solution to this that we could think of was to have a large brightly flashing signal on the screen which would at least be immediately noticed if the person concerned glanced at the screen occasionally.
Not having the audio component available exacerbates any problems encountered during a session such as difficulties with the shared application part mentioned above. We found in practice though that use of the whiteboard was an excellent, if somewhat slower, substitute to the telephone. In fact we made some use of the icons provided with the standard whiteboard but felt that the ability to expand the icon pallette with one's own icons would be a valuable asset in building up protocols and devising shortcuts which could be used by those with hearing difficulties. Showme does not support this capabaility but the InPerson software does.
For those with partial hearing a set of headphones would obviously be a useful addition to the equipment as it would for anyone contemplating using the system in a noisy environment.
Showme did not use use the mobile phone paradigm used by Proshare (Intel) which is likely to be more familiar than pop-up menu panels to less technically-oriented casual users. The controls offered in Showme for varying the performance factors (video, audio and network) were very useful but we felt that several of them were presented at a fairly technical level and would only be useful for those with technical knowledge. This gave the software a feel of being designed for experimenting with videoconferencing in inadequate environments (which,of course, to some extent, is what we were doing!) and again emphasises the immature state of the technology.
With the Showme product a fair mastery of screen management was required in order to 'drive' the system. Each of the components created one or more windows and the screen soon became very cluttered. Some windows soon became redundant and could be closed, others were useful occasionally and were thus best iconised. Positioning of windows was also a problem and we found that a shared application could easily obscure the shared whiteboard when the latter was still required. All these problems, although irritating, could be coped with by a fairly skilled X-windows user who was forewarned that such things were likely to occur. They would, however, be very off-putting for a casual or occasional user and may be beyond those with little basic training.
There is little doubt that, even in the sort of LAN environment in which we operated, videoconferencing can be used reasonably effectively by trained staff even at the present time even though the video quality was relatively poor. We felt that the video component had a subtle effect on interactions in that we often felt that it was superfluous but somehow found that when it wasn't there the conference was less satisfying. As mentioned previously, the video component was most useful in establishing initial personal contact and was less important subsequently unless something went wrong (e.g. an application failed to start). This supports the notion that current desktop videconferencing is really 'video-supported conferencing'. On the other hand the presence of an additional fixed video camera for showing documents and possibly other objects would always be very useful and so video support in the workstations used would nevertheless be useful.
We found that the audio was definitely not essential and that communication could be carried out quite effectively using text and icons on the whiteboard only, for many types of interaction. This is not to say that audio is not important since it is obviously more convenient than communicating speech by typing on the whiteboard, but this may be a significant observation to those wishing to use low cost videoconferencing over a LAN with public domain software or the less expensive options of commercial products which offer whiteboard (sometimes called notebook or data sharing) only. We would not, however, contemplate using such low-function facilities in a Helpdesk/Advisory environment.
The use of telephone links in LAN environments for the audio component has the disadvantage that several separate calls have to be made to set up a conference. In the case of Showme one would have to first establish the telephone link using the telephone number, then call a video conference using the IP address via the Address Book, and finally call a third conference to establish a Shared Application conference. This is obviously very cumbersome.
Because of the various problems we experienced with the video and audio we did not put the system into live service on the advisory desk. Our experiments showed that the facilities were too unpredictable to justify risking them on our customers at present. However we are optimistic that, if we set up reliable audio links and installed a variable focus camera, as mentioned previously, we should be able to introduce the facilities to Helpdesk staff in the not-too-distant future. We are also optimistic that over the next few years the basic carrier technology will be able to support an adequate level of synchronised audio/video communication which will under-pin the whiteboard and shared application facilities offered by desktop videoconferencing.
Graphics Multimedia Virtual Environments Visualisation Contents