I know this has been asked alot and cant really be answered so I’ll give abit more detail.
Lets say we have a highly detailed scene in a FPS game with a map size of Rainbow 6 maps.
I want to add a random chair, could I go higher count than the second picture ?
As well as what texture resolution should the normal map be ?
1st - 224 vertices
2nd - 350 vertices
3rd - 6414 vertices (Will be used for baking)
I’d bevel the edges on the middle one so the normal will look better. I think they are fine for polycount. I probably wouldn’t go over a 256 texture, it’s just a stool. Anything over 512 would be complete overkill.
For the lowest polygone model, you can trim off another 32 triangles, because you have four triangles in the end of the pegs between the legs, which is a waste. Also, give your pegs sides the same smoothing group, so it will be round. The seat shape has changed compared to the other models, should keep the same shape to avoid issues.
With the second model, you could weld the middle vertex to the side and save a couple of polygons. Depending on how it looks, you might want to get a bevel down the legs on the second model, to avoid issues with normal map.
Thank you guys for the help, the stool now looks better with less vertices.
I thought the count of the vertices would go up but it went down to 272, I’m a tiny bit confused.
Keep it as low as possible to where you’re happy with the results, and remember it’s easier to adjust the resolution of textures than it is to adjust the geometry detail.
Other than that, there’s no specific guidelines because every game is different. If you don’t have a ton of stuff in the scene then you can have higher detail objects.
Looks better, but remember to make the ends of the pegs to have their own smoothgroup, looks like you have one smoothgroup for the whole pegs, which will create issues. Else it looks good, will help with when you bake the
Smoothing groups increase tri count.
When you have the crossbars one smoothing group you made the tris from each face merge. When they are separate groups (hard edges) they actually have 4 verts per face (square), since the normals need split to make those hard edges. When they are one smoothing group the verts that touch are the same and share a normal, the smoothing is balanced between them.
Sometimes less verts is better, but if you want that sharp edge you have to split them.
