Landscape Grass Type has incorrect Rotation when density is low

Development World Creation
, , , ,
1

Grass Types can align to Landscape, which is especially useful in hilly terrain or near cliffs. It works perfectly when the density is high (aka lots of objects spawning), as seen here

However, when the Density is low, alignment no longer works correctly. Notice it with the cylinders close to the cliff i marked in red here.

This limits Landscape Grass to be used with only high density configurations. Would love to know if there’s a workaround or if a fix would be possible. Have submited a bug report a few months ago.

A 5.0 project for reproducing is attached.

Cheers,
Jan from RealBiomes

Landscape_Grass_Bug 5.0.zip (1.4 MB)

1 Like
2

Yes, bizarre. I’d log it as a bug maybe.

I notice it doesn’t happen if you turn off ‘align to grid’…

1 Like
3

Indeed, it can fix this bug in some occasions.

However, it will introduce a new bug in other occasions, where it doesn’t align the Mesh to the slope at all in some areas.

4

Also have the same problem, really wish Epic would finally address this issue as its causing so many unecessary environment bugs and forcing us to employ awkward solutions…

1 Like
5

Bumping this to reference two feature requests for LGT/Landscape Grass Types that would be great for those building large natural worlds:

6

bump

7

Bumping this, because there’s been work on Landscape Grass Types for the 5.6 update. It will somehow be integrated into PCG.

8

I believe the original issue is your fault - or more precisly, your Asset’s fault.

The angle is driven by the grid placement plus the Pivot point of the mesh.

Take a mesh that 1m x 1m. Place the pivot at its center.
Paint the mesh as foliage and align to grid.
You should always have accurate matchup of normals.

Put the pivot anywhere else - you likely get issues.

The other option is that the normals for the landacape are just bugged.
You can verify this by way of export to mesh and checking the normals value via any DCC.

In the case of your second image, it really looks more plausible that the landacape has wrong normal values…

It could also be that the grid system for placement is calculating the wrong normals btw - but debugging that is just painful.

I guess make a flat mesh that prints the normal value of the pivot point , then paint it. And you’d find out precise values from the engine…