|
Author Topic:   Terrain Code
MikeC
Member
posted November 13, 1999 04:58 PM            
Well, I have attempted to implement a Terrain based on the Continuous Level of Detail for Height Fields paper that Chris C had mentioned previously. The source and a small program are available at my web site: http://www.salug.org/~mcoker. I still see cracks from time to time, so it is not perfect, but I thought some people might find it useful. This land code is the only piece of source in the zip file. The rest of the code partly belongs to another person, so I cannot distribute his stuff at this time.

Anyway... I hope this is useful.

MIkeC

IP:

MikeC
Member
posted November 13, 1999 08:15 PM            
Umm.. Beware of the period at the end of the hyperlink... http://www.salug.org/~mcoker


mike

IP:

Chris C
Member
posted November 23, 1999 10:32 PM         
Hi Mike,

Congrats on getting your implementation finished! I was just wondering, would you or anyone else be interested in the source to my Quadtree/ROAM implementation? It runs pretty fast and allows you to flip between quadtree and ROAM algorithms instantaneously. I'm not going to do any more work on it, so if you don't mind the messy code (heaps of commented out stuff!) in the ROAM part, I'll put it up for download, with an executable.

Cheers,

Chris

IP:

MikeC
Member
posted November 24, 1999 08:21 AM            
Chris,

Yeah that'd be cool. I was looking at doing a ROAM implementation myself, but I wasn't sure if it was worth it. Of the two algorithms, which would you say is better? Or do you really see that much of a difference? I know with the ROAM algorithm, it seems that you CAN'T get cracks, but there is a small price to pay in terms of speed.

Anyway, I don't mind messy code, since most of mine is probably messy too

thanks
mike

IP:

MarkBatten
Member
posted November 24, 1999 04:27 PM            
Yes, definitely, those of us interested in these algorithms would love to take a look at your code (and thanks). That's especially true of attempts to implement ROAM, as I (at least) found the paper seriously lacking on a number of key details.

IP:

Chris C
Member
posted November 25, 1999 03:20 PM         
Hi guys - thanks for the response! I'm just uploading the program & source at the moment, and I'll post a new topic to announce it where I'll give out the URL.

Quadtrees and ROAM - what's the best? Hmm. I'd personally go for quadtrees because on the whole they are easier to implement, but a frame coherent implementation of ROAM would still be faster. A frame incoherent version of ROAM (such as in Tread Marks) would probably be on a par with the quadtree algorithm. I haven't done enough research to conclude which is the best after both have been optimized.

Cheers,
Chris

IP: