David Williams wrote:
Wow, very cool! This has to be one of the most complete projects we've seen using PolyVox. I downloaded the game and had a play around. There were a few issues running the it (I'll email you) but this does look very promising. I didn't get very far though (blew up some terrain and collected some resources... but I didn't get much built).
Thank you for the feedback. There have been some unexpected problems with the game and I am working on solving them as soon as possible. This is my first release of a game to the public and I guess there are always some problems to fix immediately following the release, especially for an independent developer with very limited resources for testing the game on different setups.
David Williams wrote:
I'd be interested how PoyVox is working out for you, and any problems/limitations you are having with it?
PolyVox has been working without any major problems since the last recommended release (11th December, 2011). Before that I had problems with crashes with the surface extraction etc., but I would say that the library has matured and is now stable enough to be used in a full production.
I have experienced three major limitations when implementing the voxel terrain: Speed of surface extraction, graphics batch count, and polygon count.
Most of these problems have been solved by working with relatively small volumes (The standard planet is 160x160x160 voxels). Each planet is divided in regions of 10x10x10 voxels, with each of these regions combined into larger regions of 40x40x40 voxels in a low priority thread. This is necessary to reduce the batch count, now limited to a maximum of 4x4x4=64 batches.
In order to achieve real-time speed of the surface extraction, multi-threading is almost unavoidable. In the game, there are three thread types: the main game thread and threads for small and large regions respectively. Each surface extraction event is added to a thread pool and executed on one of the available cores. This means that the speed scales very well for multi-core processing. At least a dual-core processor is recommended to run the game, although it will still run on a single-core processor.
Ideally I would like to reduce the polygon count of the planets as well, but the performance hit and artifacts introduced by surface decimation excludes that option. The game will, however, even with the relatively high polygon count run fast enough on most relatively modern graphic cards.
The game is very well suited for voxel rendering as I get a natural "paging" by the large separation between different planets, which means that only one planet needs to be rendered at once. This means that an unlimited view distance, without any fog, may be used in the game.
Another interesting use of voxels is for the spaceships, which allows them to be fully customizable and destructible. This should open up for som interesting space battles in future updates.
David Williams wrote:
Also, can I grab some screenshots for the
Projects using PolyVox page?
I will send a pack of screenshots to your mail that you may use. Of course, also feel free to take any screenshots you want from inside the game.
I think cross-promotion is a very good idea. Would you like me to add the PolyVox logo to the splash screen of Masterspace and a link to
http://www.volumesoffun.com on the Masterspace web site?
ker wrote:
oh wow, this is awesome.
I'll test it as soon as I can find a windows machine xD
Are you by any chance a clonk player?
The round resources and Sven2 posting in your forum kinda tipped it off xD
Nice to see that some influences from Clonk may be spotted in the game.
I have been playing quite a bit of Clonk and it is one of the main influences for Masterspace, as I always wanted to play a Clonk-like game in full 3D. Sven2 has been really helpful in troubleshooting the game thus far and I guess he registered the game after some promotion on the OpenClonk website.
