Saturday, July 27, 2013

Redux

So, my opus magnum is the game The Little Crane That Could. Every day, people tell me that they want more levels for it. The truth is, my inspiration for new tasks/puzzles to be solved with a crane is now exhausted. That is why I released a level editor for the game. Still, people want more. Others suggest a sequel to the game. But I think there can only be a sequel if there is a significant improvement to the current game. So let's talk about this sequel. First, how to name it?

  • The Little Crane Reloaded
  • The Little Crane: Redux
  • The Little Crane Strikes Back
  • The Little Crane 2014 edition
  • The Little Crane That Could: The Diggening (thanks Matt)

With whatever name it ends up, some serious new tech needs to be part of it. A logical next step would be to add deformable terrain to it. It needs soil that you can dig into, pile up, move around, maybe even compact. Doing a convincing simulation of soil digging is incredibly hard. I don't think anyone has actually done this properly, and certainly not at a smooth 60fps real time manner. Can I do it? Probably not, but it would sure be fun to try. Let me list a few attempts by others.


Digger Simulator 2011 does not look too convincing at timestamp 27sec. The loading of the scoop is too contrived. Whatever approach they took, it is not worth pursuing judging from this video.

This paper from the university of Aachen looks more promising. However, since they incorporate height fields into their solution, it makes tunnels and overhangs impossible. The video is impressive. And their model supports compacting of the soil.

Atomontage is definitely one of the most impressive demos out there. The rock sections show cliffs and overhangs, but it is hard to judge whether the sand can do overhangs as well. Tyre impressions are done well, could it do digging, tunnelling?

Spin Tires looks pretty, but when closely examining the soil, you will see that they are using a simple height field. Again, this means no overhangs, no tunnelling, etc.

This video of Cubiquity Voxel Engine (using PolyVox) is not a soil simulator. But a powerful voxel engine could possible be a great starting point for simulating soil.

For now, I think I will study PolyVox, and see what is possible with this. If you have come across a good soil simulator (or deformable terrain algorithm) that is convincing and suitable for games, please drop a link in the comments. And thank you for reading.

No comments:

Post a Comment