L3DT users' community
Large 3D terrain generator

Randomness into Climate Variables

It doesn't hurt to ask...

Randomness into Climate Variables

Postby Kafetist » Wed Feb 22, 2006 3:45 am

How about a little jitter in the climates? I like it how MinAlt1 and 2 kinda blend the textures, but small randomness could make some maps way more realistic looking.

ie. when calculating texture scores "#WaterCoeff 1.2,1.6" would give score from 1.2 to 1.6 when at water. (or something, I'm not sure how the exact calculations are done)

If you added it to all of the variables there (except filenames and texturenames and detailmap colors heh heh) we wouldn't need any "fractal coverage" mentioned in earlier post.
over - f00nk! fTW-!!! ''Oh, it was my coffee ...'' - and out
Kafetist
Member
 
Posts: 10
Joined: Tue Feb 21, 2006 5:52 pm
Location: finland

Postby Aaron » Wed Feb 22, 2006 4:49 am

Hi Kafetist,

I think I prefer the 'fractal coverage' option, as it's already implemented in L3DT (see other post). Having value ranges as you suggest would make the calculation slower (particularly since I'd have to use a noise map to vary the values), and editing the climate files would also be more difficult.

Cheers,
Aaron.
User avatar
Aaron
Site Admin
 
Posts: 3696
Joined: Sun Nov 20, 2005 2:41 pm
Location: Melbourne, Australia

Postby Kafetist » Wed Feb 22, 2006 5:09 am

Noise map would do a crappy result unless u generated the noisemap for each of the variables differently, and for each texture too.

But it wouldn't make writing the climate any harder if there was an optional randomness there. You could still be happy with "#WaterCoeff: 1.4"

It would be different from fractal coverage. This option would enable making the fractal coverage apply only to spesific things, like apply to the ground normally, but in slopes it'd be random.

I guess it wouldn't take THAT much extra time, as only defined variables would have to be noisegenerated, and have you ever flooded seas and lakes? Oh man, it takes SO much more time than attribute map generating. I've been waiting this for like Hours now.

Edit: Tho, if it takes too much, it takes too much... I guess I can do 2 layers of the same texture if I want the slope-texture to be more fractally covered than flat ground texture.

Edit2: BTW, it's been ~3.5 hours, and "t3h pr0gr3ssb4r" is just about to reach the left side of the cancel button, so I really can't claim to be half-way there yet ; D ((ooh, and I'm sooo waiting for it .. it's 7:18 AM here and I'm friggin' tired too hehe))
over - f00nk! fTW-!!! ''Oh, it was my coffee ...'' - and out
Kafetist
Member
 
Posts: 10
Joined: Tue Feb 21, 2006 5:52 pm
Location: finland

Postby Aaron » Wed Feb 22, 2006 6:02 am

Hi Kafetist,

Kafetist wrote:Noise map would do a crappy result unless u generated the noisemap for each of the variables differently, and for each texture too.


The trick is to use the same noise map, but give each land type a random coordinate offset within the noisemap. If you want to be a bit fancy, you can mess with the x-y coordinate mapping functions for each land type too.

Anyway, the reason I say a proper noise map would be required is because you need to have some continuity from one pixel to the next, as otherwise the output will look too much like static noise. If you have a smoother noise map (eg perlin, fractal), you get a tuneable spatial frequency (ie. feature size).

This approach is used in the new texture-mapping algorithm with per-pixel land-types.

Kafetist wrote:But it wouldn't make writing the climate any harder if there was an optional randomness there. You could still be happy with "#WaterCoeff: 1.4"


Okay, you're right.

Kafetist wrote:It would be different from fractal coverage. This option would enable making the fractal coverage apply only to specific things, like apply to the ground normally, but in slopes it'd be random.


But would that be useful? I don’t know, but I'll put it on my list of things to test.

Kafetist wrote:I guess it wouldn't take THAT much extra time, as only defined variables would have to be noisegenerated.


Maybe. I'll try it out.

Kafetist wrote:and have you ever flooded seas and lakes?


Yes ;)

Kafetist wrote:Oh man, it takes SO much more time than attribute map generating. I've been waiting this for like Hours now.


Thanks for the reminder - I need to decrease the default lake-flooding strength. It's absurd when you use 50% on the ‘lakes’ slider. If you use a value of 10% you'll get about the same results (because there are only so many places a lake can be on a map), but it will take a lot less time.

Anyhoo, I'll drop the extra noise parameters on the to-do list for L3DT 2.4a (or so), when I plan to upgrade the climate system.

Cheers,
Aaron.
User avatar
Aaron
Site Admin
 
Posts: 3696
Joined: Sun Nov 20, 2005 2:41 pm
Location: Melbourne, Australia


Return to Feature requests

Who is online

Users browsing this forum: Google [Bot] and 4 guests

cron