Wanted: Deciduous Tree Models

[SpikedHelmet]

It's always polite to ask.

This sounds very exciting Argh, I can't wait to see it in action. What kinds of parameters are you planning on including? ie, feature density, space between each feature, etc, and will it be possible to only generate random features over a specific area? this could be cool for creating urban environments very quickly.

[Argh]

Doesn't guess give you virtually everything he does anyways?

I nearly always ask. It's a gift of Art and Time! Moreover, he knows that I'm always waaay behind skinning what he's gotten modeled, so it's better to say, "hey, now's good".

What kinds of parameters are you planning on including? ie, feature density, space between each feature, etc, and will it be possible to only generate random features over a specific area? this could be cool for creating urban environments very quickly.

I'll be explaining all of this in as much detail as I can when the demo's done, but basically:

1. Density, per volume, user-defined, is already built. That was the easiest thing to do, frankly.

2. Space, per object, is already built, and I'm planning an improvement when I get home on Sunday.

3. Yes, you can generate random stuff over any area, any size, any parameters (so long as they are rectangular- no circles, no diagonals, but you can do a series of rectangles offset X,Z).

4. Urban environments that have random elements are an easy task. With enough logic, and content, I could do Sim City stuff, with what I have now, frankly.

The bigger fish, imo, is making terrains that look like they're full of Nature, which is why I've started with that problem first. That's where the aGorms and LathanStanleys have been trying, very hard, to get us to, and I want to help them, because I want to play the Tree Map someday. At a playable framerate. Or a version of AfterShock that features palms that can actually catch fire, drop coconuts, that sort of thing.

5. Is it totally, 100% perfect yet? No. Will it shine yer shoes? No. Do I have all of the content done? No. Is there a definite release date? No. I keep thinking "yes", then another problem rears its head. However, the main logic is, very thankfully, complete. Now it's mainly content, and deciding how fancy I want the initial demo to be, frankly. I could release it tomorrow, it works at a technical level, but I'm not gonna, to put it one way, I'd like to polish it a bit more first.

[aGorm]

Is there any chance you could send a Dev version of what you have done over? Only, I'd like to see what he code is like for things like tree generation, ect..., see how I can purpose it to what I want. Obviously anything I come up with I'll send back your way. Only without knowing what any of your code looks like, I can't do any work my end on stuff that will fit in to your code... Seeing as your way should be a lot more flexible, it means much of my current stuff will need re-purposing to make the most of your system. (Well, if it does everything you say it dos... )

aGorm

[Argh]

I am going to be releasing something over the next 48 hours, to you and others, with comments and explanations. The code's not rocket-science stuff, I promise, although to do custom work, you're going to need to do some work with both BOS and with LUA. But neither area involves anything that isn't basically cookie-cutter code.

[aGorm]

BOS I can do quite well if i put my mind to it... hopefully I can butcher your LUA to. Very good news, thanks mate.

aGorm

[Argh]

Don't worry, the LUA is very easy, and I'll try to include enough comments that it's obvious what goes where, and some examples of common things people will want to do.

[Argh]

Better, faster solver algorithm. Can generate more natural-looking terrain, too! I guess getting away from the code for a couple of days and doing nothing but thought experiments wasn't a bad thing


Right-click and view picture, to enlarge.

Lurker, I think it's fair to say that I may actually have this issue dealt with. I can pack stuff as tightly as is probably warranted, without overlap, in fact I'm going to have to get creative soon.

aGorm, those are some of your bootiful Features there. A small bird delivered them from On High. I am going to GPL the World Builder core code: may I have the palm trees (note, not "own them", just rights to include them in a GPL'd package)? If not, no biggie, I can emulate what you've done, maybe saving some texture space- that's the biggest issue with them, they're using very large textures. Price of Pure Win, I guess, they're totally gorgeous, and are very efficient at sharing space- it's almost like you were thinking about what I'm doing before I was

[rattle]

Does it avoid water or is it configurable to what feature to use for what terrain?

[aGorm]

I'm fine with you using them... I did want first dibs on using them in a map, but seeing as I now realy realy want to make use of yoru world builder, I suppose you can use them. I'm goona be changing them a bit for aftershock anyway.

Does this mean you have somthing you can send me so I can have a gander?

aGorm

[Argh]

Does it avoid water or is it configurable to what feature to use for what terrain?

Well, here it is, placing palm trees only on lower areas, while planting deciduous trees and my non-tropical plants on hills and higher elevations. And yeah, it avoids water. Actually, that was the easiest part- Spring handles that when you call CreateUnit, but now it's double-dealt-with... I could have underwater plants now


Right-click and view picture, to enlarge.

If I feel like messing with that a bit more, I can probably have it adjust for slope. I need to concentrate on content, though.

As for terrain: originally, my thinking was that mappers would set up "boxes" of solving locations within the main loop, so they'd just say "from here X to here X, from here Z to here Z, you can place <blah>".

To do it in automated fashion is going to take something that it can read, to determine what terrain it's dealing with. Maybe it could read the state of the crude speed-map stuff built into SMF, but that'd require making it operate at cross-purposes with other things.

@aGorm: Thanks!

[lurker]

Lurker, I think it's fair to say that I may actually have this issue dealt with. I can pack stuff as tightly as is probably warranted, without overlap, in fact I'm going to have to get creative soon.

Good. Because this code has not gone well. I'm sure it would be more efficient if you were to pack the map insanely tight, but otherwise...

[Argh]

Right-click to enlarge.

A fully-automated set of "features", generated with World Builder. As you can see, at different terrain heights the algorithm adjusts the foliage generated, so we go from tropical terrain to non-tropical terrain on this map (keep in mind that results will vary, per map, based on the height of the terrain- a map where the lowest ground is at height > 90, for example, will only see deciduous forests).

The only thing left now, and I think it's worth doing, is to code a very simple check for angle changes, so that it won't try placing trees on very steep angles, as even with adjustment of object positions via COB, it doesn't produce the kind of results I think are satisfactory. I'll code that, and then release it for testing by appropriate parties.

[Argh]

Right-click to enlarge.

Final version of the algorithms, I think. Now checks for extreme height variations, to prevent borked placement of objects, has settings that allow for different foliage based on altitude.


Right-click to enlarge.

Final tweaks complete. Now mappers can just adjust a few parameters, and voila, foilage is based on relative altitude. No that's what I call a tree map

[aGorm]

No your only issue is the conflicing art styles.... but then thats down to mappers to fix. When you goona release?

aGorm

[Argh]

The art styles thing is mainly a function of having enough content that is realistic, renders fast, and doesn't suck.

Most of the time investment in this project has been on the coding end- with more investment in content, that's not a problem. I just haven't had time. This is going out for testing in a few minutes, expect a PM.

A few last shots to show people what this can do, at this stage.

Right-click to enlarge:


Imagine having to place all of that, by hand- that's several thousand objects there

[rattle]

a map where the lowest ground is at height > 90, for example, will only see deciduous forests).

Are you going with a fixed minimum/maximum height or do you calculate the lowest and highest point over water?

[Argh]

Are you going with a fixed minimum/maximum height or do you calculate the lowest and highest point over water?

I originally used a fixed height, but found it impractical- Spring maps do not have consistent height values, due to many of them being "inflated" after the initial .SMD was made, and some are scaled very differently than others.

So, instead, I use a percentage approach, using the largest and smallest values for height. This also results in some inconsistencies, which I am not sure as to the cause, but as the screenshots clearly show, it works well enough most of the time.

[Forboding Angel]

I came


Argh, lurker, whoever else is working on this.... YOU GUYS ARE FUCKING AMAZING!


Don't suppose I could get a shot of Delta Prime with this could I?

[Argh]

Ok, Forb, I couldn't find a map called DeltaPrime, so I've shot a bunch of shots of your stuff (including a really surreal one of that "glass" map, lol)... here ya go.

[Forboding Angel]

Dude, thank you.

Wow, that is amazing. Simply amazing.