In this section, we would like to address several design issues which we have not yet solved to our full satisfaction.

The convincing power of a VR landscape, in particular when it is about a historical scene, depends a lot on its atmospheric qualities. The atmospheric appearance of a VR scene is determined by the visual aesthetics, including the degree of dirtiness of the generated objects, as well as by the often neglected space acoustics. Computer-generated architectural models are simply too clean - they are only convincing when the story is about new buildings. Otherwise street dirt is missing and the whole range of real life traces of decline caused by time and environment. With current software unfortunately these effects can only be realized with an extremely high effort. In our project the absence of such effects is particularly visible, because for a lot of set constructions we use photographs. The consequence is that new, proper objects are next to visibly old and dirty objects even though they come from the same time period.

The problem of aesthetics mainly concerns scenes of those times before the invention of photography. Imagine that our time machine would really allow us to take brilliant and sharp color shots in the 19th century, and import them into the present. The photographs would certainly not convince us because our collective vision of the 19th century is influenced by paintings. There bright colors and hard contrasts had no place. Therefore for the scenery we try to adapt the visual aesthetics of the 19th century city and landscape painting.

When the problems of appearance quality are solved a time variable VR platform can supply a range of new and interesting options for exhibitions with historical objects and the mediation of historical knowledge. The biggest advantage of VR landscapes is certainly that they can visually supply a historical context for museum objects. In classical exhibitions the context is usually reduced to the literal frame in which the object is presented. The historical context is then available as an explanation text and therefore purely mentally accesible. At best visualizations of the object are supplied in catalogues and can be studied after the museum visit at home.

As opposed to the procedure described above a VR landscape allows us to connect a single historical object with the appropriate urban context and its real life component. In exhibitions with objects related to a historical change this is of particular advantage. The historical context is, so to speak, automatically delivered by the time traveling function as long as the development takes place in the time span covered by the system.

The contextual relation of museum objects supports multiple synergies when the VR platform is used for several exhibitions with diverse issues. This can be accompanied by very practical advantages e.g. when objects of the exhibition with subject A can be used as set for the exhibition with subject B, and vice versa. Therefore we are hoping that the tremendous work effort we have had to put in the set objects of our VR world, could be reduced if a future VR platform was intensely used by museums. In addition, an interesting synergy in content can also be the result. A user who is for instance interested in an exhibition focus "History of consumer products" which is planned by us, will be seduced to also look at information concerning "History of mass media". In this case the connecting element would be the change in product advertisement.

The strong context relation of exhibition objects also has its problematic side. Obviously a connection to a specific every day context demands correct dating and spatial categorization of the object. Since most museum objects are usually presented in isolation, exact information on their life time and distribution areas is not needed. Therefore we had to undertake substantial research to get this missing information for the set objects of our VR landscape.

Another problem pointed out by historians is the historical distortion for which such a VR platform could be blamed. On a pure visual level the differentiation between the museums object and the illustrative set fades for the user. But that is exactly the point. These boundaries are explicitly supposed to vanish: the fading of boundaries is the suggestive power of the VR platform. The illustrative objects are designed in order to recreate history as a whole not only a fragmented piece. It only has to be made clear what they could have looked like that in the chosen time and in the reconstructed urban area in which they could have existed. In contrast, museum objects are defined rigidly. In fact, we could have avoided these accusations by countering that no historian was able to help us when we needed "hard" facts.

Finally, we acknowledge a problem concerning user guidance. Users of the system must be able to receive information on the historical status of an object via adequate navigation systems. The problems of historical framing also demonstrated that an exhibition realized on an online VR platform made greater demands on an interdisciplinary approach than any conventional show. Such an enterprise is unthinkable without a functioning cooperation between archivists, librarians, photographers, historians of diverse special knowledge and a strong willingness to communicate with the involved VR specialists.

As in real life, in virtual reality the question to whom the travel offer should be made, must be addressed. In both situations two opposite types can be found: on the one side the kind of traveler who wants to know before hand what will be expected and who wants to be pampered and permanently informed. On the other side we have the knap sack freak on his search for adventure and authenticity. This candidate would feel uncomfortanle with organized travel and be worried about his freedom.

As in real life both traveling groups could interfere in a negative way. Just think for instance of visual or acoustic hints integrated in the scene which could possibly destroy the knap sack freak's illusion or think of the misuse of the VR platform as a 3-D adventure game which brings the server to a standstill and therefore disturbs a serious traveler on his educational journey. But unlike the real world, the VR-technology has more options available in order to allow both groups to go their own way without interfering. The travelers can switch between free exploration and guided tour at any time. This was the main reason for our decision to offer both possibilities.

The guided tour is addressed to users who have no experience with the navigation in 3D worlds. It is also addressed to users looking for very specific museum information. We are planning to establish guided tours for all three main subjects. With the help of a separate browser window the guided tour will lead the visitor step by step to the historically interesting sites. With every stop additional information to the subject will be supplied (Figure 4). The visitor can decide when the next step will be made. Any time he can switch to the free navigation. For museums which want to use the VR platform for an online exhibition the guided tour is an important design element.

Much more difficult to design is the navigation assistance for users who want to move freely. In fact, there are two problems to be solved which are not directly related to each other. First, the user has to be provided with feedback on where he is moving, virtually, and what time he currently is in. Otherwise he will be a victim of the 'lost in space' phenomenon, which here also means a 'lost in time' syndrome as well. In order to prevent this we planned some kind of pilot display which is inserted as a standard object in a separate browser window. In this window the actual time is indicated, and can be changed by turning on a time wheel. Furthermore, this window includes a city map. On the map the actual location of the visitor as well as all the stops of the guided tour are indicated. With a mouse click on the map the visitor can move to another site. With a click on the guided tour stop he leaves the free navigation mode and latches onto a guided tour.

Secondly, the user needs assistance in order to locate those objects behind which additional information might be hidden. Therefore in the map of the pilot display all feature objects will be indicated as hot spots. In the scene itself only the special arrangement of objects will point out on feature objects. Otherwise the interest in free exploration would be lost. Whenever the set contains very detailed or moving objects, or the lighting and acoustics noticeably differ from the surroundings, the visitor can expect additional information.

Let us undertake another time travel and let us look back on the near future of our VR platform and other similar VR worlds. Considering the speed of technological development, five years from now should be enough. Well, looking at the year 2003, the view is rather foggy. Still, three developments seem to be quite clear. One development concerning the users of VR landscapes, one concerning the museums and one concerning the data both users and museums are dealing with.

We'll find far more activity in development of VR worlds. Museum educators and teachers obviously overcame their mental reservation against DOOM and other computer games, and encouraged the development of multi-user VR worlds populated with avatars.

The old-style museum belongs to the past. Classical museum functions are no longer under the control of a single curator. Many museums have specialized in purchase, filing and conservation of objects.

The copyrights on images including the copyrights on 3-D presentation are mostly hold by big media and telecommunication companies. Some museums turned into data brokers or exhibition agencies. They react to sensational media events like "The sinking of the Titanic" and organize exhibitions of current interest. This is a development which we actually took for granted when we started to develop our online VR systems.

PICS is as dead as Z3950 - even Dublin-Core and Iconclass are seriously ill. The winner is Extra-Medium-Large. We could have used such a meta data solution on the basis of XML, even for our small project. But one look at the confusion in standards discussed by the museums, archives and libraries convinced us to desist from using meta data.

Final result: We can look back optimistically on our future.