Clusters - grouping of similar geometry objects in one cluster

[JohnVK]

Hello,

I was wondering if there is a way to use the cluster feature with groups of plants?

For instance I have a forest object which contains multiple types of plants and grass. If I enable the cluster option (which is great to add more realism) all of the geometry objects are treated individually and are also clustered individually. The problem with this is that, aside from multiple other plants, I have like 6 grass objects which are now not merged together but separated in their own clusters as well, which leads to a somewhat uniform and unrealistic look especially in the grass. It would be great if I could tell forest to treat those 6 grass objects as one and make it distribute them randomly within one cluster. Then the same for other plants within this forest object.

I often use multiple objects for one type of plant/grass/tree with dissimilar looks, to break the uniformity of it all. But when I start using clusters it brings back the uniformity because it clusters only a single geometry object in each cluster. I hope you understand what I mean 

Is there a way to do that?

If there currently is no way to do that, would it be possible to create an option to group objects, so that they get randomly scattered within the same cluster. For instance by adding an option in the geometry roll-out called "cluster group" in which one could enter a grouping number? Obviously this much more complex than simply adding this option, I realize that. But maybe something to think about for future updates?

[iToo]

Why not create small groups of grass as separate objects and scatter them ?

Futhermore, these groups would be generated using several Forest objects, all using the same settings but changing the seed (in the Realistic Grass tutorial you can see the right way to use nested Forests).

[JohnVK]

Hi Carlos,

Thanks for the suggestion 

Indeed I know that's an option. I often use other forest objects as geometry sources. But that would only work well for flat surfaces or mildly curved surfaces. Wouldn't it?

I often use very irregularly shaped surfaces with gaps and holes where the patches would either stick out of the ground or get embedded in the ground, since they will only follow a flat surface or the one they were created on, giving a unnatural look. The separate grass patches used as geometry source would have to still be very small for this to work in most cases where the ground surface has many gaps/holes/ups/downs. Which brings us back to the original problem and the reason why I am using very small grass objects (only a few leaves in them) 

Or can I use UV mode in the patches to make them follow the surface? Would that work? I haven't tested this extensively but in my couple of tests that didn't really work out.


Also, it's very hard to work with exclusion areas if you are using patches because the boundaries would not be as smooth as when using small (for instance) grass objects/patches. For instance, I often use one forest object for larger rocks and then a forest object to do grass & flowers & smaller stones, which uses the large rock forest object as exclusion area. The grass, flowers & stones must follow the boundaries of the large rocks closely for it to look good. That won't work very well if I'm using patches.

Also, fall off density and scale won't work as expected when using other forest object patches as geometry source since it would affect the entire patch, not just the boundary of the surface / area. Right?


Being able to have multiple items within a cluster would solve many issues for me in a very simple manner. I'm very sure others would benefit greatly from this as well! But if it's not possible I will have to think of other ways to solve it 

[iToo]

Ok, i understand. Please, let me to think on it. This feature would be interesting, but involves some degree of complexity in the current Forest workflow. In any case, it should be implemented as a new feature.

Or can I use UV mode in the patches to make them follow the surface? Would that work? I haven't tested this extensively but in my couple of tests that didn't really work out.

No, unfortunately a patch is considered as a single mesh object; it cannot be Z deformed.

[JohnVK]

Ok, i understand. Please, let me to think on it. This feature would be interesting, but involves some degree of complexity in the current Forest workflow. In any case, it should be implemented as a new feature.

It would be awesome if a feature like that could be added! I understand this is a very complex operation. So thanks very much for taking it into consideration!

No, unfortunately a patch is considered as a single mesh object; it cannot be Z deformed.

Yeah, I was afraid that might be the case.

Thanks again Carlos!

[BT4215]

exclude from clustering?

[JohnVK]

Hi,

What do you mean by exclude from clustering?

[RC4697]

Just wanted to chime in on this topic..

This is the very thing I was looking for with clusters. I usually use about 3 different models of every species of trees to get enough variation - for example, a birch tree height 3 meters, 6 meters and 12 meters, and the clustering separates these 3 models to their own clusters which means I have patches where there are only either short or long birches which ends up looking a bit silly.

Maybe something like putting models with the same wire color to the same cluster?

[iToo]

I have some ideas to solve this issue in a clean way. We should test it before, but i think we could have something to show for the next release.

[JohnVK]

Thanks for chiming in, RC!

And Carlos, that would be awesome! Thanks very much!

[iToo]

Hi John, i forgot to post here before:

This problem should be now solved in Forest 4: just create a group with the plants you want to keep together, and assign it as Custom Object. Forest will scatter them as an unique object, but handling each item individually for area and surface checkings.

[JohnVK]

That is absolutely awesome!   Thank you! I am going to check this out asap!