Download

Landscape painting vs imported landscape (static meshes) performance differences ?

i couldn’t really find an answer to this

what would be better performance wise ?

use the landscape painting tools or create a few large 3d models in an external 3d software and bring it in as a handful of static meshes ?

You will get a better performance when you use the integrated landscape tool. -> good lod system :slight_smile:

that was fast

thanks!

This makes sense from a performance standpoint. Also, if you run into issues where you are in need of a very specific landscape you can still make a height map in another package and UE4 can do the rest (from what I’ve read… haven’t done it myself).

That’s true. You are abel to import heightmaps directly with the landscape tool → you can create heightmaps with world machine, photoshop, in your 3d program,… :slight_smile:

i knew about being able to import height maps from the documentation but i haven’t tried it yet, it will probably be part of my workflow though - either way thanks for trying to help :slight_smile:

The landscape demo was made with that technique -> https://www.youtube.com/watch?v=yVTAQqo2eMo (very informative video) :slight_smile:

thanks will check it out

A landscape system is obviously superior for an open world, but if your maps are small like Counter Strike or TF2, then static mesh will work absolutely fine, and you’ll have greater control over optimisation.

Terrain that you will never reach, like skybox terrain, is best done as a low poly mesh with baking from high poly. If I use a tool like pro-optimise in 3D Studio Max, I can take a 100k volcano, and bring it down to below 10k, and it will appear to look exactly the same because the light-mapping and normals will be baked into a texture using the 100k version. No one will ever get close enough to know the difference because its outside the playing area. You can also make the detail stream-able, like with a landscape system, by using tessellation.

You can use LODs with static meshes, and even tessellation instead of LOD, so there are ways to optimise static mesh terrain when you do use it. You can use the vertex colours in your modelling software to paint where different texture blends will be. You can leave it simple and just do black and white, one channel, and divide the mesh up per material (grass/dirt, grass/sand, dirt/rock, etc). Or if you don’t want to divide up that much, you can use all 4 channels RGBA. It’s up to you :stuck_out_tongue:

i have been thinking about this, the landscape tool does seem a little bit overkill for what i had in mind and your suggested skybox terrain does seem better suited since most of the terrain would serve as an unreachable background but my map would be bigger than the average counter strike or tf2 map (i think as i did not play tf2 and cs has been a lifetime ago) and i do not know where the cutoff size wise for static meshes > landscape tool would be when it comes to performance

hope i am not remembering it wrong but in UT3 the onyx coast map was fairly large and i was thinking maybe at most about 1,5x that one but not bigger

http://www.team-azerty.com/images/photos/lan/24102014/ut3-war-onyx-coast.gif

yes LODs will be crucial to me from what i remember after reading the documentary for a quick overview of what the engine offers, definitely will have to look into it again

You could even do a mix of both. Static mesh for skybox, landscape for some parts of the playable area, LOD/tessellated static mesh for others. Whatever works best for you :slight_smile: