Sense or nonsense of Enviroment Content Packs

Hey guys,

I just downloaded and installed an interior environment package for UE4 from the marketstore. I don´t want to harm the guy who created this package, that is why I will not name the package. First of all, it is very well done and the demo levels look fantastic, so full marks for the designer. But are modular packages sensible in terms of performance? I don´t think so. The package consists of very small assets you can place in the level and combine with other assets to create an unique look in the editor.

As allready mentioned, it does looks stunning, but even the small demo map has hundreds of tiny assets. Sure, the designer wants to reach a lot of potential buyers and therefore creates one that offers the most flexibility and I can´t blame him for that, but I can´t imagine that this kind of level design is very optimized.

Well, I don´t really regret the purchase of this package, but I don´t exactly know what to do with it :-/ .

What do you guys think about packages which are highly customizable but have small bits and pieces?

Best regards,

You can use culling (drawing) distances and with blender its possible to make more LODS. Takes around 1min to get LODs for wanted mesh. E.g. with tiny pieces you can have 1000m draw distance. Those dont then affect performance after getting 1000m away. I have around 20000 meshes, few million foliage meshes and around 1,5km*1,5km map divided to 16 streaming levels. Having no overall FPS hit. If there isnt locally much meshes or ligthing the FPS is very high. Just takes time to optimize how you Place the objects/ligthing to get the best performance. With proper LODS you can use more dynamic lighting over the meshes.


I second SaOk : with all the methods available to optimize level, the fact that assets are modular with many small parts, is usually an advantage. For example, you have all lighting optimizations, level streaming, clever level design with occluding geometry well placed, a good usage of instancing and Lods.

Generally speaking, the trick is more about draw calls count, in the end. Huge amount a static “unmovable” meshes won’t be the main performance hit. But many individual textures can generate many draw call.
On a package point of view, I think a good point for a package is to provide textures as atlases as much as possible, ie the same textures shared by as much meshes as possible. Poly count reduction is more a concern in mobile-pack, but the number of meshes you put in a level is not really the critical point.

More generally, I am a strong believer in modular asset. I am totally convinced by this paper for example.

I agree with SaOk and FreemanCIT: modularity is the way to go.

A pack you bought could be problematic to use if the models included dont have any LOD. Should be something to watch if you buy a complete pack. Does your pack include LODs? Perhaps you could contact the creator to ask if he could create LODs?

Smart Actor Merge usage combined with Precomputed Visibility that 40fps Demo level might well run at 120fps. That has been my experience with outdoor modular kits anyhow.

Thanks guys for your opinions, keep them coming :slight_smile: .

Well, LODs will certainly help in large environments with a lot of space, but not in dense interior scenes. And as far as I know UE4 doesn´t use instanced static meshes, so I will end up with a lot of draw calls (even if they share the same material) . Instanced static meshes can be created inside of blueprints, but not just by dragging the assets into the scene. Please correct me, if I´m wrong here. Foliage is a different case, that´s why it works well even with lots of little assets. As noqnoq says, I think merging assets is a good way to increase performance, but this involves other steps (exporting the assets to FBX, merging them, dealing with smoothing group and material problems and importing them again.).

I wish UE4 had a grouping / merging feature inside the editor. This would save a lot of time, increase performance and get rid of these baked shading errors, because lightmass could handle these groups on the same thread.

@ Martin Bean: No, the package I bought doens´t have LODs.

you can name the pack, you should
and even leave the review for it on store page, as warning for other people

Maybe THIS plugin from @Rama could help you? I haven’t tried it yet by myself, but it looks like it could work in that case.

Hey ForGobbler, have you tried actor merging? You may have to enable it in the preferences if it isn’t automatically checked. Check herefor info how.

Thanks for your interesting posts, Slavq and SE_JonF, I will have a look at these :slight_smile: .

@ CriErr : Well, there is nothing I can warn you from, because the package is acually well done, as far as I can tell after a few days looking at it. Materials and assets are fine and offer a lot of tweaking possibilites. It´s just that everthing is too modular for my liking :wink: . I don´t want to ruin the guys business by posting something that sounds negative and is not really his fault.