rows of trees + mask

Started by Dune, May 01, 2010, 02:16:10 am

Dune

May 01, 2010, 02:16:10 am
This is part of a large undertaking; making a complete winter city in the Middle Ages. Church, streets and boats are in the making. My problem is to get the trees to line up. Fortunately, many rows are perpendicular, so a straight mask (dots where the trees are supposed to be) and sub-settings in the population of 0.1 will do the trick I think. But some lines of trees are angled at, say 15 degrees. Like in the front row. You can see they line up, but the line 'jumps'. My idea is to make more than one population, and/or more masks. If anyone has a better idea, I'd be much obliged.

Don't mind the house, I sank it too far.

---Dune

gregsandor

May 01, 2010, 04:22:17 am #1
It looks fine.  What's the question?  I use dotted masks all the time for trees, just setting the frequency to something that ensures decent coverage.  Make a square mask, set it to your area of interest size, and dot where you want trees.

Henry Blewer

May 01, 2010, 09:07:10 am #2
It may be easier to model the mask like you would an object. Render this for the color mask. This could ensure that the tree/foliage matches the building and streets just right.
http://flickr.com/photos/njeneb/
Forget Tuesday; It's just Monday spelled with a T

dandelO

May 01, 2010, 10:38:50 am #3
Ulco, the shader array node has editable rotational values, could that be used to place a surface layer mask to blend by? ...

Dune

May 02, 2010, 03:45:04 am #4
Thanks, guys. I know the things you mention. I wasn't too clear I guess. I was just thinking aloud how to adjust the size of the dots in the mask to the distance between the trees in the populator. If you have fairly large dots, and fairly large distance between trees (and the seed is after all more random with the subdivision at near 1, instead of 01), the chance of a tree ending up exactly central in the dot is not 100%. If you make smaller dots in the mask, you need to decrease the distance, or a tree will not 'hit' the dot. I could of course make several populations, set the subdivision to 0.1 and thus create one or two rows, then adjust the angle of the population to adjust to the angle of pathways. But I prefer to make one mask and one population. So I guess I have to experiment with the smallest size dots possible (for exact placement) and small distance between the trees in the populator, and hopefully not miss dots or get two trees at one spot.
If you get my point

Henry Blewer

May 02, 2010, 08:50:10 am #5
I think you hit the spot. err dot.

I used a mask in Blender for some old project a few years ago. I ended up 'adjusting' the objects for use in Terragen 2. It was a pain and I got tired of trying to figure out where the b boxes really were. Long ago, before John told me about increasing the atmosphere samples to get rid of the grainy sky.
http://flickr.com/photos/njeneb/
Forget Tuesday; It's just Monday spelled with a T

Dune

May 03, 2010, 02:28:12 am #6
@Dandel0: A simple shape can be used for that as well, I figured yesterday. But even just a population rectangle can be angled. With a subdivision of 0.1 a straight line of trees would appear. But it would be most interesting to have a way to get exactly one instance in the center of each dot of white in a mask. But perhaps that is not even programmable?
I will work this problem out, and fill in the missing trees with another seed and crop render.

---Dune

otakar

May 03, 2010, 02:35:23 pm #7
I am following this thread with great interest as I was trying to populate a ring with one row of objects. Could not do it with equal spacing using just one population and a procedural mask. Also interested why you are using a variation value of 0.1 instead of just 0. Is this a trick?

Dune

May 04, 2010, 01:34:08 am #8 Last Edit: May 04, 2010, 01:36:30 am by Dune
@otakar: I'm full of tricks   No, it was just that I didn't want the trees to exactly line up, but have a slight variation. I came across a problem once, that might be your lucky chance. If you move the initial object away from the 0/0/0 where it automatically sits for the population, the populations tends to arrange itself in a ring. Perhaps you work that out.