Hi hey gurl! I've been trying to understand something... when it comes to performance, is it vertices or polycount that really matters? I’m trying to get a clearer picture of how they affect gameplay. Hope this isn’t too random tho 😂
Hey nonny, thanks for the great question! I'm not sure if you are looking for a non-creator answer or if you maybe have some experience with meshing, but I try to cover some technical aspects that hopefully will make you understand vertices and polygons better! It's a bit of a longer journey, so grab some snacks and settle in!
Disclaimer: This is my opinion and the definitions are how I came to understand them. I believe what I say to be correct, but I am also human and may have misunderstood something, and making mistakes can happen. Thanks for understanding! TL;DR There will be calculations made on every vertex and every face present in a mesh. So the more you have of either, the worse it is for performance, but if the vertex count is much higher than the poly count, the mesh is especially unoptimized and may impact performance more than it would in an optimized state. Vertex and polygons are actually connected, literally. To understand, we need to start with the basics.
Vertices are points in a 3d space that provide the foundations of a grid-like 3d structure, known as a mesh. Connected through lines known as edges, they form a 3d shape. Three edges form a triangle, or tri for short, also known as a face or polygon. These terms can be used interchangeably, but a tri always has three vertices, whereas faces and polygons can have more than three vertices.
A polygon with 4 vertices (known as a quad) is very commonly used in 3D-modelling, because it can be used to create very neat and well-structured meshes (it is easy to make grids using rectangular shapes). A mesh that is made in a grid-like way may have the same overall shape as one made out of triangles, but the flow of the edges is different, and this plays a role during animation. Having these grid-like lines is usually better for moving and animating characters and avoiding shading issues.
Most meshes for games are thus initially made with 4-sided polygons. These quads are usually further broken down into tris if they aren't already, and three is the least amount of vertices needed to form a coplanar surface.
When the quads of the left sphere are turned into triangles, the grid structure will remain intact:
We can further break down the elements of a polygon. Let's take a simple cube. It has 6 sides, or faces, and each face has 4 vertices that make up the corners. But if we count all the vertices, we will notice that each face shares vertices with other faces. So on a simple cube, we have 6 faces and 8 vertices.
Depending on how a mesh is made though, it may have a lot of edges. You can think of those like cuts in the mesh. If we wanted to cut up the cube, we would separate each face from each other. Now we still have the same number of faces (6), but the vertices are suddenly much higher, because now, every face has its own set of 4 vertices, meaning that we have a total of 24 vertices.
In video games, polygons are usually rendered as triangles. So for our cube, that would mean that every face would be diagonally cut in half, giving us a total number of 12 faces/polygons/tris.
If we take it to the next level and separate all of these triangles from each other, the vertex count would be 36! To recount, on a mesh that is not split, it's only 8. Having these cuts (or split edges, as they are commonly called) in strategic places in the mesh where vertices are split can be important for the final look of the mesh in the Sims games.
You may have already encountered some meshes that have dark spots or shading in some parts, often at the hems. That can happen if the mesh is not split correctly in these areas. It has to do with how the normals work (they are important to how light gets reflected off the mesh).
To not go too off-topic, I won't get into detail about that here.
Ideally, a mesh is only split where it needs to be so it will look right. If there are unnecessary cuts, then the vertex count (and with it, the polygon count) goes up when it doesn't really have to be. If a mesh has a lot of these areas where a lot of faces have their own vertices, then the vertex count is much higher than on a mesh with fewer split edges. This can be a sign that a mesh is not well-optimized. So if you see a mesh that has much more vertices than polys, you may want to question it in regards to performance. CC converters will probably know that Sims 4 meshes come with a lot of these cuts that are needed for Sims 4 because of the way the CC is made (for example, pants have such a split edge cut to allow boots to look tucked-in), but those are superfluous for TS3. So I encourage every Sims 3 CC creator to remove those to save on both vertices and polygons when they are converting! The thing is, in the game, the mesh will be processed and there will be calculations made on every vertex and every face present. Each vertex is indexed, and triangles need to be drawn from the information. Other processes will need to also use the vertices for their calculations. There is more going on during rendering than simply drawing the polygons. The shaders have to calculate their effects, the textures need to be rendered and applied accordingly, filtering such as anti-aliasing will be applied, etc. So the more geometry you have in a mesh, the worse it gets in terms of performance. More geometry means that the polygons are smaller and more numerous, which will in turn affect the vertex count too.
As CC creators, we cannot address all the factors that impact performance, as we can only influence a small part, but we can try to adhere to the game's standards to minimize the performance impact as best as we can. If we wish our meshes to be more detailed than what the game's standards are, then we impact the performance more.
Instead of just cranking up detail like crazy, it is good to see if we really need it. And ask questions such as how often will the player view the item closely? If the player is usually far away from the item or it is a very small item, making it super detailed will not really do much, since the item looks the same as a low-poly item from a certain distance. We can also make compromises and maybe make an item 50% more detailed instead of 200% more detailed, providing a more detailed item while minimizing the performance impact.
This is a bit off-topic, but since I'm here already, we may refrain from using a bigger texture for a very plain piece without any notable intricate detail (like a basic T-shirt, for example), since it already looks fine with a smaller texture, but it may pay off for a highly detailed texture which looks visibly better with a bigger texture, warranting the performance impact (like a lace blouse with an intricate pattern, stitching, and small buttons).
On that note, using the texture space efficiently can also do wonders to increase detail. Use all the real estate you have on that UV map (that is a 2d representation of your 3d mesh and used to apply the texture to the mesh). Have you tried upscaling the texture yet? Okay okay, I'll stop now and get...
Back on topic: The number of polygons also matters, not just the vertex count. When a mesh has lots of smaller polygons, a lot of them are necessary to form the entire mesh. Bigger polys will not provide as much fine details, but will have less impact on hardware because fewer calculations to be made. A good mesh uses only as much detail as needed while using as few polygons as possible to still maintain functionality.
Sims clothing meshes need to animate and move well, so they need to be able to flex at the joints, and the shape must also be able to change to accommodate body sizes, thus the mesh needs the minimum amount of geometry to achieve this and still look decent, which is a sweet spot the developers have to find to craft meshes that perform well while looking reasonably good.
A good habit for creators looking to make meshes for the Sims games is to look at similar meshes in the game and study them (so for TS3, check the original TS3 game meshes, for TS4, check the original TS4 game meshes). Check how many polys they use and how good the topology (structure) of the mesh is.
For example, how many rows of edges do they usually have, and what sizes are the polys usually? How many vertices are in one round of edge loops, for example for the arms? How are the different mesh parts connected to each other, where are split edges used? Then imitate that.
The professionals who made these original meshes knew what they were doing, and they optimized them to be gameplay-friendly, so if in doubt, copy what they do.
Game meshes are usually especially optimized and it is very much game-dependent how they are structured, so a mesh from a different game can look totally different. CC Creators looking to optimize performance in their meshes should try to match the original game's mesh in terms of structure and polycounts. This will also ensure they animate well!
If you made it this far: well done, thanks for bearing with me! I hope you feel more knowledgeable and ready to apply what you've learned to your own meshes or when selecting CC for your favorite Sims game.