A Circle of Stone, A Circle of Ideas

With the inevitable return to work now looming on the horizon, A-Level results due to be published this week and all the various preparations for the new academic year beginning to make their presence felt, I was keen to squeeze in one more post before the balance of time inevitably shifts back towards work.  For this offering in the 'Season of Scenery', I thought I would share one of the more unusual pieces of terrain I have built for The Silver Bayonet; unusual not so much because of what it is, but because of how it came about.

A little explanation is required here.  There has been much discussion in schools recently about the use of Artificial Intelligence, and I have been sufficiently intrigued by the possibilities to wonder whether it might have a place in my own creative process. Rather than asking it to write a lesson plan or mark a set of essays, I thought I would try something considerably more interesting.  I opened ChatGPT and asked,

 

“I am currently playing Joseph A. McCullough's The Silver Bayonet and I need to design some terrain. I am looking for ideas for spooky rock formations or other related items that I could build. Can you help?”

What followed was, to my surprise, a genuinely exciting creative exchange.  Ideas were thrown backwards and forwards; some were mine, some were suggested by the machine and quite a few of the more extravagant proposals were quietly ushered towards the door.  There was a certain amount of discussion about what would actually work on a gaming table, what could realistically be built and, importantly, what might still look as though it had come from Awdry Towers rather than a computer.

Eventually we settled upon the idea of an ancient stone circle; tall, weathered stones, cracked and fallen in places, with a liberal sprinkling of Gothic horror and enough sinister detail to suggest that perhaps this was not somewhere sensible people should be wandering around after dark.  And then I had a thought.  Could we take these ideas and turn them into something that could actually be 3D printed?

I have been experimenting with a Bambu Lab A1 for about a year now and have enjoyed some modest success; I am still very much in the foothills of the learning curve, but the possibility of taking an idea from a conversation and eventually producing something physical was rather appealing.  So I asked another question, 

"Could ChatGPT help me create some spooky terrain STLs?"  

The initial answer was encouraging.  There was a brief misunderstanding which resulted in links to existing stone circle files, but eventually the conversation moved towards creating something from scratch.  Suggestions became more detailed, the design evolved and, eventually, I was presented with what appeared to be the next stage of the process; code that I could supposedly use to generate the STL files.  This, best beloved, is where matters became somewhat less straightforward.  What I had envisaged as ancient, vine covered rocks with deep fissures and an appropriately sinister appearance initially emerged as six rather immaculate columns; something, clearly, had gone rather awry.

There followed a reassuring exchange about design modifications and adjustments; I was encouraged to leave the process to do its work and return later when my wonderfully detailed STL files would, presumably, be ready.  I returned, nothing.  More questions were asked, more code was produced, more increasingly elaborate solutions were suggested.  Before long I found myself being led rather further into the depths of my laptop than I had originally intended, altering settings and contemplating processes that were beginning to feel suspiciously like actual computer programming.  I was becoming increasingly sceptical, but more importantly, I was becoming increasingly nervous.  Eventually I drew a line under the experiment and asked the question which, in retrospect, I probably should have asked considerably earlier, 

“Is it possible that you are unable to do this?”

There was, I confess, a certain amount of face saving, but ultimately the answer was yes. The parameters of what was possible within the process meant that it could not actually produce the object I had been imagining and yet, somewhat unexpectedly, I didn't feel that the experiment had failed.  By this point I had invested quite a bit of time in the conversation and, perhaps more importantly, there had been a genuinely creative exchange of ideas.  So I changed tack and simply asked for a visualisation of some of the things we had discussed.  The result was a wonderfully Gothic image which immediately gave me something far more useful than another piece of computer code, it gave me an idea I could build.

And so, somewhat inevitably, I returned to the painting table.  The construction itself was wonderfully straightforward. Great chunks of offcut blue foam were hewn to shape and arranged upon an MDF base; the base itself was deliberately cut to an organic outline and its edges bevelled to help it disappear into the tabletop.

I had seen people use crumpled aluminium foil to create convincing rock textures and thought I would give it a try.  It works remarkably well, although I discovered that achieving the desired texture requires rather more force than one might initially expect.  The foam was then carved, battered and textured until the smooth manufactured surface had been persuaded to look considerably more geological.

The stones themselves became the main focus. Cracks and fissures were cut into them and I added a selection of mysterious sigils and devices, not so much to explain what had happened here as to encourage wonder as to what might have happened.

The whole piece was conceived as blocking line of sight, but with access provided by some foam steps could also act as a vantage point.  A simple flagstone area echoed the original visualisation, but I wanted it to feel like a place that had been used for some purpose long before our intrepid troops arrived; somewhere that had accumulated offerings, rituals and perhaps the occasional unfortunate visitor.  Naturally this meant another raid on the bits box.  Weapons, tools, skulls and assorted other detritus were scattered around the stones; little offerings left behind as some form of tribute to whatever gods, spirits or other malign entities might have been worshipped here.

There was also an opportunity to return to an old favourite; hairy wool.  Carefully teased out and attached around the bases of the stones, the fibres make surprisingly convincing roots and creeping vegetation.  Once painted and combined with clump foliage, they help to tie the stones into the surrounding landscape and give the whole thing the appearance of having been sitting here, undisturbed, for considerably longer than is probably healthy.

Painting was, as usual, a mixture of familiar colours and experimentation; browns for the earth, greys for the stones, because, obviously, all rocks are grey.  It was at this point that the rather stark shapes began to disappear and the individual elements started to become a single piece of terrain. Weathering, scattered vegetation, stones, fragments and the occasional unsettling artefact all helped to tell the story.

And there it is.  A piece of terrain which began as a conversation, became a slightly alarming experiment in computer generated geometry, wandered briefly into the mysterious world of code and file generation, and eventually found its way back to where such things belong; on the painting table, surrounded by scraps of foam, bits from the box and a rather unhealthy enthusiasm for Gothic horror.

This wasn't quite what I had in mind when I started thinking about using AI to assist with creativity, but I rather like it.  The machine provided the conversation and, eventually, the image; the rest was left to me.