Modular models with soft edge technique, double edge vs adjust normals?

McGreed

So I'm working on some modular pieces for the unreal marketplace, and trying to figure out what would be the most useful method to use to get soft edges on the models (non-unique textures), as they both has it's advantages but I would like to get some input about it.

I made an example below, 1 is with edge loops and 2 is with simple bevel and adjusted normals, the results are pretty much the same, apart maybe from the corners.

The edge loops method has it advantage that it's easy to add to models and to remove again, and you can easy make changes to model, however it does increase the polycount a lot, in example 108 triangles.

On the other hand the normals one results in less polygon, however does make it more an effort to get it working, and if changing something, you need to redo some normals as well, so extra work there. But the polycount is only 44. And these two examples are pretty simple.

I'm leaning towards edge loops currently (after writing this down), as it does make it easier to do changes without messing up normals, smoothing or UVs too much, however I'm worried about the increased polycount. Any input/experience about this, what you maybe find is most beneficial for downloaders?

Obscura

The double edge method not only gives more polygons, but also more thin triangles, but its easy to work with. The single chamfer with edited normals gives less polygons, less thin triangles, and better silhouette because you actually round the corner. That being said, it is indeed harder to work with.

McGreed

Yeah, I can't really decide on which would be most useful/practical, might use a combination depending on assets I guess.