Reducing faces advice?

[Fenix Eldritch]

I'm reworking a building facade to be more efficient. Originally the windowed wall was not connected to the columns and ran behind it. Now I'm removing the extra verts and having the wall's edge use the column's edge. As I fill in the wall, I am trying to use as few faces as possible and have a dilemma.

The picture below shows the two ways I've filled the wall between the window and column. The right side uses more faces, but creates a fully contiguous (is that the word?) geometry. The left side uses fewer faces (yes I know I haven't triangulated the quads yet, but it's still less than the other side). I'm actually doing the same thing with the sections between the windows too... But unlike the right side, I don't connect each and every vertex.



So my question is this: Is there any reason why I shouldn't use the method on the left (and center) for keeping the face count down? Is there any drawback to not having the geometry completely connected at every vertex? (is there a phrase for this?)

[ChinRey]

---

Fenix Eldritch wrote:

Is there any drawback to not having the geometry completely connected at every vertex? (is there a phrase for this?)

---

Usually that is not a problem. Lining up disonnected faces like you suggest requires a bit more precision in vertice positioining both on the 3D model and on the UV map but as long as you're aware of it, you shuld be fine

[Vulpinus]

I have had issues with doing that; lining up two non-connected faces in a single object.

When uploaded to SL, from just the right (or wrong) angle I got feint, shimmery lines (single-pixel width no matter how close I zoomed) at the boundary between the faces. The faces were flat like yours and aligned as accurately as Blender allows. It happened to me every time I tried to do it this way and was noticeable enough that I gave up on the idea.

I ended up reconnecting everything and living with the increased face count. Alternatively, depending on the topology, embedding one face slightly into the other works but if you want the faces planar, that's no good of course.

[ChinRey]

---

Vulpinus wrote:

I ended up reconnecting everything and living with the increased face count.

---

That's actually what I do most of the time too. Yes, I'm notorious for my obsession with low LI and high LOD but there are limits to how much extra work even I am willing to do just to save a few polys.

[arton Rotaru]

Here is an old thread about the topic. https://community.secondlife.com/t5/Mesh/What-is-the-best-topology/m-p/1644591/highlight/true#M16443

Peronally I also avoid having edges overlapping each other. I'd rather construct the mesh in a way where I can intersect the edges into faces. Or make them watertight when it makes sense. Sometimes watertight geo is preferable to avoid "dead" texture space.

[Chic Aeon]

I really can't see well enough what is going on in your screenshot, but if -- as someone suggested it has to do with vertices that are unconnected but more or less lined up you can MERGE those Vertices. Occassionally I have builds that because I was getting a bit too fancy need some vertices joined.

 

I do agree that trying to upload and use a mesh with vertices NOT joined when needed is a really bad idea. Fluttering and tiny little stitching lines are often in evidence.

 

To MERGE   select your vertices, choose the W key and then the MERGE function. (Pay close attention to your selection clicks of the vertices and which merge method you choose our you will have edges that aren't straight. )

 

Good luck. ww

[Fenix Eldritch]

Thanks for the responses, everyone! The linked discussion was pretty much exactly what I was looking for..

So the consensus seems to be favoring a fully connected topology with slightly more faces instead of fewer faces with overlapping edges.

But just to be safe, I want to make sure it is understood that I don't have any duplicate vertices in my picture. There are no extra verts that could be removed or merged. So in that light, does it still hold true that creating faces with a few additional overlapping edges can lead to generating MORE verts in SL's internal model format as indicated by the linked thread?

[arton Rotaru]

---

Fenix Eldritch wrote:

...does it still hold true that creating faces with a few additional overlapping edges can lead to generating MORE verts in SL's internal model format as indicated by the linked thread?

---

Unlikely. I would think the opposite would be true, at least for hard shaded edges. There could be less vertex normals which would lead to less duplicated verts in the non-watertight geometry. But as always, it depends on the model of course. And with smooth shading it might be a totally different story. Then there are still the UVs, and material borders as well, which can have an impact on the vert count.

[Codex Alpha]

---

Fenix Eldritch wrote:

So my question is this: Is there any reason why I shouldn't use the method on the left (and center) for keeping the face count down? Is there any drawback to not having the geometry completely connected at every vertex? (is there a phrase for this?)

---

I am responding as an intermediate user of Blender, and certainly not an expert (yet). Others can jump in and correct or add to my statements as they need.

First off, it is good practice to have your model faces divided into quads, even if Second Life pretty much is going to triangulate your model when you upload it.

Second, keep in mind that if you use/mark seams in Blender on the object, they might as well not be connected together, as apparently Blender treats it like a mesh split and doubles the vertices in essence.

Third, because of inevitable seams and UV unwrapping, having it one piece may not help you anyway, and you're doing extra work and possibly warping your surfaces with n-gons (faces with more than 4 vertices).

So one could debate if creating an object in one mesh - or 'shell" (Maya), or 'island' (Blender) - matters, since UVing the piece is technically going to break it up and separate areas of the mesh anyway

Apparently, every newbie believes they must make their object out of one continuous/contiguous mesh, then forget their seams and UV unwraps are going to cut and split it anyway

If I were to design that same feature that you were,I would design the curved areas and perhaps vertical middle bar together, but create the horizontal cross-bar separately, and the entire thing separate from the wall itself.

I would keep each piece in quads, and triangulate if needed (especially cleaning the arch ways). I would CTL-J each piece together instead, and let the vertices float where they may for each component.

I do this, because when it becomes time to create lower LODs, each piece of the window can be decimated (removed lines and vertices) much more cleanly and efficiently, almost down to a billboard, without losing too much shape. If you have everything together in one continuous 'shell' or 'island' or 'mesh', decimating it down may be much harder.



For example, when making a simple wood box, I did not make it out of one shell, but rather built it similar to what a carpenter would do.. board by board. I could bevel each piece as I needed, and when lower LOD came, i could  decimate them so much easier and they held their shape.

I hope I explained myself well enough.