How we built the 3D Trail

Grant

Watch Grant's blog

Designing a web tour system that provides the heritage industry with a dynamic way to display visual content was always something that I had in the back of my mind from an early stage. Archaeology is a unique area that often presents challenges and restrictions that may not be necessarily found in the more commercial visualisation sectors, it is also a field where every project can be wildly different ranging from maritime, to terrestrial, politically sensitive, or physically in danger of rapid degeneration. To this end what is expected from a digital output and how it impacts both the local and wider communities has ranging consequences in relation to ownership, privacy and cultural sensitivity. Additionally, budget, time considerations and the agreed information that can be projected to the public are also areas that need to be frequently discussed.

As a 3D artist and trained archaeologist, or ‘digital archaeologist’ there are a number of reasons why I wanted to go with a HTML5 web based environment driven by image sequences, video and 3d model plugins: of which we hope to soon develop our own to move away from SketchFab. A huge consideration is quality and after working with 3DS Max and creating both still and animated CGI over a number of years, it has become clear that the level of output provided by renderers such as VRay can not only be truly photorealistic, but that this is achievable by small, experienced teams of people, reasonably quickly. Individual artists, more so commercially than archaeologically are rivalling what studios can create and compared to the relative level of what gamers are dealing with on a daily basis, the relative curve seems more achievable within heritage. It should be remembered that the models created for these web tours are not mutually exclusive, they are resources than can be committed to more interactive platforms (such as Unity or Unreal) in the future, but to do so would potentially require a fair amount more time investment, cost and development, something that isn’t always possible on every project. The interaction might be there, but the polish and top level custom models outside of premade kits purchasable on the Unity/Unreal marketplace require considerable investment to do well and then present. Our system works as a perfect alternative to projects where the budget, team size and time investment is not going to be good enough to achieve a very high end gaming output where the step to commit from the 3D modelling package is a few stages before the required investment to make objects/assets Unity or Unreal ready.

Focusing on achieving a polished output was our goal and to this end the web tour provides a very good platform to present high fidelity graphics in a way that has not yet been fully explored to its fullest (but will hopefully be in time by us) without requiring any special downloads and just an internet connection. Schools, museums, universities and libraries often have externally managed computer systems that may not even allow for the installation of the most basic of plugins or apps and we didn’t want to take this for granted. With our system this is not a worry and it works across all platforms equally, it also crucially provides an environment where the archaeologist can work closely with us to build the right tour for their needs, especially when information may be lacking in certain areas and the communication of transparency is a concern and a more structured experience is preferred. Having total freedom can often be a double edged sword with many sandbox experiences feeling lacking in narrative, or engagement. In this regard, the ability to quickly create sequences from inside of 3DS Max/Maya/Blender etc to not just supplement information (maps/colour coded overlays etc), but help to build narrative is unparalleled and by association any popular plugins for these platforms are also naturally integrated (such as ForestPack/Rayfire/PhoenixFD etc). The use of specific renderers such as unbiased or spectral renders is also possible.

The rise of photogrammetry sweeping through the heritage world is also a factor as working scans through retopology processes can not only take considerable time, but also requires an experienced user of software packages, especially to extract high levels of detail and then project those back onto more lightweight models. Bringing these assets into 3DS Max with displacement shaders and relying as little as possible on techniques that ‘fake’ details allows for as much visual information as possible to be squeezed out of the scans without worry about model size, user hardware, or texture resolution. Rendering from 8k or 16k image files is not a problem and producing huge images to allow users to zoom into their content can also be achieved.

The sole concern then is just what traditional animation packages can handle (often a combination of millions of polygons, particle systems and other effects) and rendering these out. There is little reason to create amazing Photogrammetry models only to have them crunched down to a point where much of their detail is lost. Over time with technological developments this process will become more efficient even for higher resolutions such as 4k. Consequently what can be done through this system is only at the very early stages and as we develop the tour, due to its unified system, past projects will benefit from future work and updates as we go.

Stuart

Technically there was a wide range of challenges to overcome when creating the Coronation virtual tour. There are various pre-made frameworks and services out there for hosting and presenting 3d through the web - yet they all focus on single models, not entire environments. The tour needs to function across all devices and network speeds in any situation. Schools, museums, galleries - anywhere the web exists. The entire development of the project was bound by these core principles.

One of the biggest technical difficulties was how to display the 3D environment renders in a controllable way. A true 3D model at this quality would be huge, take too long to download and perform very poorly on older devices. Videos on the other hand have a very small footprint at great quality and require no additional plugins - however they are not controllable in a way a 3D tour should be. Our solution is essentially a rotational video loop of the wreck site with full controls and the ability to drag the video back and forth horizontally. This creates the effect of moving the scene much like you would in a 3D model - the only difference being its restricted to the horizontal axis only.

Another challenge encountered was the mobile support. Apple phones are very restrictive when it comes to web video. Androids were in slightly better shape but older devices on both platforms suffered serious slowdown issues. The solution to this was a separate system that performs exactly the same as the video but uses switching images instead. This same adaptation is also used for older desktop browser versions, which were released before HTML5 video support.

These were just a couple of technical solutions involved in creating the Coronation virtual tour. The end product is an educational experience that works across all devices, naturally through the web browser.