Dynamic Lighting

One of the things I particularly like about CryEngine 2 is its implementation of dynamic lighting, great for seeing the effects of changing sunlight on the landscape.

This simple video sequence shows a static view of Laxton Castle from sunrise to sunset, compressing a day into 90 seconds.  True, not much happens but the lighting effects at dawn and dusk are great and there is considerable potential for exploring interplay of light and landscape further.

First Steps in Source

Half-life 2 is one of my favourite games, there is just something about the look and feel of the landscapes, the atmosphere and the character models that feels right and I've been keen to try it for archaeological visualisation for a while.

However, the brush based modelling used by Half-life makes use of terrain models derived from greyscale heightmaps problematic. Inspired by discovery of Dan Pichbeck's excellent Chineseroom and his AHRC funded Half-life 2 mods Dear Esther and Korsakovia I felt it was time to grapple with the problem again.  A little bit of searching turned up a really useful tutorial at Interlopers.net and a link to DispGen, a tool for deriving Half-life brush-based displacement maps from greyscale heightmaps.  Initial results are promising, and it proved relatively easy to get our Laxton Castle terrain model into Hammer and on into Half-life 2.  Scaling is a bit of a problem, solved for now by using an NPC as a rough scale.  Early days, but the potential for using Half-life for visualisations based on real terrain is quite exciting...

Laxton's Castle: A Virtual Fieldtrip

Laxton Castle in Nottinghamshire is a fine medieval motte and bailey, once supporting a substantial masonry castle which served as the centre of the administration of Sherwood Forest.  We have surveyed the earthworks over a number of years, and our initial GIS-based visualisations provided much of the impetus for testing out the alternative route of game-based visualisation.

As we've worked on the Laxton model we have begun to think about and explore ways in which the game paradigm can be exploited to encourage exploration as well as simple for visualisation.  A first step is to embed within the terrain model vestiges of the masonry curtain walls and towers, based on the surveyed remains and, realistically hidden in the vegetation.  The first question for a new gamer is "what are you looking at and what evidence can you find for the original form of the castle?"   I feel that we are creating the first draft of a kind of virtual fieldtrip, encouraging users to explore the game landscape and learn from it just as one would confronted with the real thing.

West Burton Deserted Village

Continuing with the aim of using game software for rapid visualisation, here is a very basic rendering of some airborne lidar survey of the deserted medieval village at West Burton, Nottinghamshire.

The original lidar data, a 1m resolution elevation grid, was modelled in ESRI's ArcGIS, filtered to remove buildings and vegetation, and exported as a greyscale heightmap.

The final rendered terrain model in CryEngine has preserved to a remarkable extent the detail in the original lidar data.  Once again, avatar-based exploration gives a new perspective, even to simple landscape models.

Stonehenge Quick Render

As a simple proof of concept we have put together a quick CryEngine based rendering of Stonehenge.
In the absence of any better data the terrain model is simple generic flattish land.

The model of the stones was beautifully crafted by Tom Harvey, using SketchUp and is available from the SketchUp 3D warehouse.  A few edits in Blender and export to CryEngine's cgf format was all that it required. Scaling the model has proven somewhat problematic, and the results is based on what looks and feels about right.

The original textures, which don't conform to the CryEngine requirements, are also distorted, but overall the sense of place is interesting and we quite like running freely through the stones.

Working with CryEngine

Our work is focused on turning landscape-scale surveys comprising many thousands of three-dimensional elevation points into rapidly developed game-based visualisations.  To visualise archaeological surveys the key is to find game software that is capable of quickly generating large landscapes based on real data. We have settled on Crytek’s CryEngine as our usual tool.

After experimentation we have developed a work-flow that copes equally well with fairly small landscapes (of a hectare or so, usually derived from GPS topographical survey) and much larger landscapes (tens of hectares) based on terrestrial laser scanning and airborne lidar.  The in-game landscape models can be dressed with vegetation and trees (accurately located based on survey results) and structures (modelled separately in open-source software) and populated with artificial intelligence - birds that fly away when you approach them. Stick around long enough and the sun will set (and eventually rise again) it may rain, or become foggy, a gale may blow, listen and you will hear the wind in the trees and the birds singing, you are exploring a real archaeological landscape after all.

About Second Site

For almost as long as archaeologists have been using computers they have been exploited to render and visualise archaeological data. As computers have grown in power and graphical capability archaeologists have explored and exploited these new capabilities.

Often our focus has been on visualising objects, their small size and discrete boundaries lent themselves to the relatively limited processing power of early computers, and sometimes down relatively dead ends (for example using VRML in the mid 1990s).

Recently, and under our very noses, a quiet revolution has been played out in the world of computer graphics as computer game software has led the way in exploiting the prodigious capabilities of modern computers; often game and hardware developments go hand in hand, and the complex, beautiful and fiercely interactive worlds of modern games showcase the pinnacle of what is possible using consumer grade computing hardware.  So, what if we could harness the graphical capabilities of game software to visualise archaeological data, not just objects or structures, but entire landscapes?  Such visualisations would have the added advantage of being both immersive and interactive (in the jargon of visualisation the user doesn’t just look at a two-dimensional graphic they are able to enter into and freely explore a fully three-dimensional world) and they would co-opt the familiar game paradigm, no need for users to learn complex software, if you can play FarCry (or whatever is the latest game) you can explore our virtual archaeological world.