Someone I follow on twitter—I think it must have been Nicolas Holzschuch—retweeted the following tweet, which grabbed my attention:

Markus Schütz: Just uploaded a preprint of "Real-Time Continuous Level of Detail Rendering of Point Clouds"…

The third image from that tweet struck a chord in my mind. It resonated with a couple of ideas that had been bouncing around in there for a while: flying motion through convoluted 3D architectural spaces (not unlike in Descent); a concept that the world is only visible only when you look at it, as a performance optimisation (lingering from reading Greg Egan’s novel Permutation City many years ago); and after connecting with these ideas, the image also suggested a visual aesthetic:

An atrium under construction, with flying staircases connecting the four storeys, and temporary railings lining the edges of the concrete slab floors. It's rendered with millions of voxels, tiny squares that, for visual effect, are drawn at increasingly large sizes until at the edges of the image they're just large square blobs of colour, no longer forming a coherent image.

I was particularly struck by the railing in the upper right of the picture, where the voxels are not small and precise enough to seem solid, but have not yet degenerated into an incoherent mess as they do further out from the centre. I suddenly had the glimmering of a gameplay and aesthetic combination that I wanted to explore.

Of course the use of point clouds and voxels for rendering is nothing new. Memories of Comanche: Maximum Overkill from way back in 1992 also resurfaced and connected with this soup:

A view from the cockpit of an attack helicopter. It's flying low over a river-filled canyon. The ground is rendered with large, crude cubical voxels.

This started some wheels turning in my head. And I tweeted:

Just saw something that gave me an idea I want to try coding up … and I want to ride that inspiration … but with this housing sword hanging over my head, and inability to focus most days, I’m afraid to start.

Because I’m afraid that (as has nearly always happened in the past) something that gets started then immediately disrupted without getting any momentum ends up aborted, and so buried under a pile of negative emotions that I can’t resume it.

This might be different if I had a solid toolkit to use, to reach momentum faster. But in games stuff I’ve not yet found any such toolkit that was viable; and writing one is a big project with the same momentum problem and a whole herd of big hairy yaks too.

Maybe I just need to find a way to spend six months in an environment conducive to focus just to scrape together such a toolbox?

Some place, between a forest and a lake; with IRC and access to code/research-related internet sites, but without twitter and such; and ideally with one or two or perhaps three likeminded folks working on a similar sort of problem…

Seems like a pipedream though.