More details or Fewer Vertices

Hello, I’m new here. I’ve been working with Blender for some time creating simple content packages for some game engines.

I always try to optimize the vertex count of my objects as much as possible and maintaining an optimized mesh, I have always read in all types of forums that the first rule to follow is optimization and vertex count.

But how true is this today? I know that each particular case would have its own rules to follow, my question is to what extent can I leave aside the visual quality of my objects in exchange for a low vertex count, with today’s technology would this still be a problem?

I’m currently creating some buildings and on each edge of the walls there is a detail simulating overlapping stones.

The image below has 3 situations:

1 - The object on the right is the original object, a simple square with some extrusions and no rounded or detailed edges.

2 - The middle object is the same object but with the bevel modifier to give it a more beautiful appearance.

3 - The object on the left is the more detailed version with damage simulation on all edges.

The first version is disposable, it is just the standard object that I will use as a basis for the final version. I would really like a point of view from someone with more experience than me to guide me in this decision.

Many projects that I created and launched didn’t have so many details because I always followed this motto of low vertex count at all costs, I really don’t want that anymore, I know that I also can’t extrapolate and create objects with hundreds of thousands of unnecessary vertices , but as you can see in the attached image, in some cases it is necessary to have more details even if it costs more vertices.

I appreciate any help.

PS: I don’t create any animated objects, characters, animals or anything organic, my focus is only on inorganic modeling without animation.

If you attach it as a wire frame, it will help you refer to it.
Reducing the edges in the blender uses the Decimate modifier. But I’m not sure if this will be useful. :thinking:

1 Like

Welcome :tada:

You might have overlooked something… context. What are your objects; where are they visible and to what extend. Do you really need all this different L-shapes ? Or are maybe only two or three with somekind of corner breaking silhouette needed and more individual appearences are maybe possible by different texure normal maps…

Also : for still image art, a low poly game game, high poly, stylized, realistic…

3 Likes

Since you are creating these for Game Engines, what you already have is LOD ( Level of Detail) data… Where the base cub is the Far Distance View, and the Left is the Up-close View… so because of this you still need more vertices to support the higher level and less for the others…

You still need to have optimization of your meshes and work in Quads, even though it usually is converted to Triangular mesh in the Game Engine. The other thing to consider is further optimization by Baking the Details down to Normal Maps to lessen the dependency of more Verticies…

It WAS a thing back in the day where the lowest poly-count as well as the lowest resolution of images was a MUST. Today that is not nearly an absolute with better hardware and display combined with unlimited storage. Just look at the Size and resolution of some games, with a package size of over 100 GB with 4k textures…

Just a game asset, with 18,500 faces…Lod data textures at 512 and 1024 sizes, animation capable as well as physics at a frame rate of 60fps ( depending on hardware)

So it is a combination of all these factors to consider as you create your assets…

Happy Blending!..

2 Likes

Oh yes i remember when the whole Lara Croft model once was getting bigger than the first Tomb Raider game… now only the vertices data will be bigger… or maybe even the pony tail ( damn in the new ones she do not even have her famous pony tail ):wink:

2 Likes

Thanks for the answers. Unfortunately the game engine I’m currently using doesn’t support all types of PBR materials, so I have to create some details in the mesh itself if I want more detailed objects.

For a project with a short screen distance, the details will make all the difference, I currently have audiences from various segments, I will still keep some projects in low-poly, but I have many ideas for new projects that I will definitely need to focus more on the details instead of absolute economy.

1 Like