What causes these bands in wide-area grass?

Started by yesmine, March 18, 2014, 11:15:21 pm

Previous topic - Next topic


I've been testing grass clumps covering a wide terrain and, until now, I had never noticed a strange 'banding' pattern when distant terrain is viewed at lower angles. It also seems to occur in foreground coverage, though not as obvious. See the images (and where I've noted in red), it looks almost like hedgerows, and at first I wondered if there were actual ridges in the terrain, but that isn't the case. It shows up in different views at right angles (sample 1 faces west, sample 2 faces south), so it seems to be happening in the grass population at large. In the foreground there is a curve to it as well..it doesn't continue to be a straight line pattern..and there's a definite change in the coloration of the grass clumps. So, just maybe it's the coloration change that's creating the illusion of ridges. If so, why is the grass color changing in such a consistent pattern?
I don't know if this is just the way it is; if some other vegetation was breaking up the terrain it might not even be noticeable. But if there's a way to fix this, I'd appreciate the info.


Are you using population color variation?

- Oshyan



(Oshyan): "Are you using population color variation?"
...No, not that I'm aware of (unless it defaults in a node somewhere).

(Yossam): "Or a strata and outcrops shader?"
No (haven't even gotten to try one of those yet..)

Basically, this is a DEM for the terrain, and I've been testing with a couple of grass patches (One 'neuspadrin' sample and one 'Mr. Lamppost' sample.) It's simply a patch selected from the library and brought in as a population. Since this terrain usually also has bushes on it, I may simply have not noticed this before now. But when I disabled the bushes and happened to be testing a ground cover sample, and then just one grass alone, I suddenly noticed the banding pattern in the far background.

I've been testing just now and found something about the 'Object spacing in a,b' that may well be where the issue arises. I noticed that with the earlier samples (above) the a,b values were set to 1,1.  I usually set them to different numbers to avoid uniformity, and found just now that setting a,b to 1,2 or even 1,1.5...anything with some variation over 1, seems to make this much less obvious. But when I set a=1 and b=.7...or anything 1 or less, the grass seems to suddenly turn to terraces (see the image below).
  It appears to me that it is only the change in the coloration of the grass that makes it appear to be ridges. The grass seems to shift from darker to lighter (or vice versa) front to back, in rows across the terrain. It's back where the change becomes very distinct that the illusion shows as dips in the terrain.

I think that's what I'm seeing. Nevertheless, as long as a,b doesn't go below 1, I think this is a non-issue, or at least can be easily avoided (though it might be a useful trick to know in some circumstances.)


It appears to me there's a certain banding in the terrain, so some areas of the grass look denser, which is logical.


Thanks Dune, and your comment brings up something I may need to know. I've noticed that surface layers, and perhaps other layers, can and often do intersect an underlying terrain, and that can cause the grass to appear more dense here and there, or absent completely in some places. So I'm watching for that now. But the underlying terrain I am using here doesn't seem to have any uniform banding in it, at least I see no general patterns that appear to correlate to the wide 'hedge row' look I had been getting.

Still, you have a lot more experience at this than I do so, beyond intersection issues, etc., can something about a terrain itself effect the density or brightness of (in this case) grass or other things sitting on it?
Thanks again.


My understanding of the a,b variation is this :  1 is essentially full spacing variation. 0 means no spacing variation.

If you want absolutely regular placement you would use 0,0 and you would get the clumps spaced evenly at exactly the distances you specified in the object spacing, giving a perfect uniform tiled effect . If you tried a test at 0,0 values for variation it should show up any irregularities in the terrain since the spacing becomes dead even.


Can you post the tgd, so we can have a look?