2
$\begingroup$

I am new to 3d-modeling, so please bear that in mind and if you find some fundamental flaw in my explanation below, do not hesitate to tell me.

The mesh I want is a cuboid with a lot more faces on one of its sides compared to the others. I inserted a cube, scaled it and applied transformations, subdivided it once and added a few edge loops to help subdivision. The resulting mesh then looked like this:

enter image description here

(Up until this point everything worked as I expected)

In order to get more polys on one of the sides I selected its four inner faces and subdivided them, the result was that adjacent faces "degenerated" into n-gons, which I want to avoid (because the target application does not like them). This is the resulting mesh, where I selected the resulting ngons using select->facesBySide:

enter image description here

This of course makes sense, because these adjascent faces now have more than 4 vertices, but I am not really sure how to proceed from here. I currently see only two solutions to this problem: 1) Either subdivide the whole mesh evenly (which is actually not a solution at all, because the resulting mesh will have the same poly-density on all sides) 2) Connect all the new vertices of the adjascent faces (the ngons) to a "corner vertex" by creating new edges for each, which results in a set of many ugly stretched out triangles.

Both solutions seem undesirable to me. So how would a more experienced poly-modeler approach this?

Thanks in advance!

ps. I realize that this is a beginner's questions which probably has been answered a million times. Unfortunately I did not find anything, but if you know a post or a tutorial explaining these things, it would be great if you could post the link.

$\endgroup$
  • 1
    $\begingroup$ We still need to know why you want to do that to find a solution. $\endgroup$ – Bithur Jul 31 '15 at 19:44
  • $\begingroup$ is it necessary that it stays one object? $\endgroup$ – gladys Jul 31 '15 at 21:45
  • $\begingroup$ @Bithur I want to do that because I wanted to make a virtual "canvas" for use in zBrush: The idea was to paint on one side of it using zBrush's brush engine. Since I pretty much need one poly per pixel for painting in zbrush I didnt want to waste too many polys on the backside of the "canvas" and subdividing the mesh evenly would cut my available "pixel count" to half (more or less) $\endgroup$ – n-i-c-l-a-s Aug 1 '15 at 3:23
  • $\begingroup$ @gladis: no, technically it is not necessary that it stays one object, but it would be nice if it would :) I think other zBrush users just paint on a subdivded plane, but I kind of liked the 3dimensional "feel" of a canvas with thickness. (Also I try to force myself to make some basic meshes in Blender every now and then to learn some proper 3d modeling :) ) $\endgroup$ – n-i-c-l-a-s Aug 1 '15 at 3:30
  • $\begingroup$ However, isn't it pretty common that people want to use higher detail on one side of the mesh and lower detail on the backside? How do people generally do that? I think I missed a very basic lesson in 3d-modeling here :) $\endgroup$ – n-i-c-l-a-s Aug 1 '15 at 3:33
0
$\begingroup$

When you subdivide a face of a mesh, you also subdivide all of the faces which share edges with the face you subdivided. But while a mesh is part of one and only one object, an object may contain more than one mesh. This is the key to the solution to your problem.

One way to do this is to take your original cube, in edit mode, and delete only the face (X > "Only faces") in which you want to increase the vertex density. Then select two opposite edges which helped define the face you deleted, and duplicate them using the duplicate tool (SHFT - D). Use the F key to create a face between the duplicated edges. You now have an object with two mesh sub-objects, a plane, and the other five faces of the cube. For convenience, you can assign these two sets of vertices to different vertex groups, so that the sub meshes can be more easily selected individually, but even so they are part of the same object. Now when you subdivide the new face, you will have no impact on the five faces of the other sub object, and as long as you don't move the four pairs of vertices defining the overall edges of the new face, or that outlined the old face, individually, the two meshes will maintain their relative position.

Another way to accomplish the same thing is to select the face in the original cube that you wanted to subdivide (in edit mode), and instead of deleting it, press the P key selecting the option "selection". You have now made the face you want to subdivide a separate object. Selecting this new separate object, and then the cube (in either order), and pressing the key combination CTRL - J combines the new face with the other five faces of the original cube as a single object, but leaves them as independent sub-object meshes.

In both cases, you can subdivide the individual face as many times as you want, without affecting the other sub-object, but all six faces in the two mesh sub objects are part of the same object, and when the object is manipulated, the six faces are manipulated in the same manner.

$\endgroup$
  • $\begingroup$ Thank you very much for your reply. This was very helpful! Weirdly, I would never have thought of separating the object into different meshes, I need more practice I guess. This totally makes sense of course, so thanks very much for the tip :) Is this common pratice when working with complex meshes? I guess one might get problems if it is necessary to animate the object, but for static meshes this seems to be a very nice clean solution. $\endgroup$ – n-i-c-l-a-s Aug 1 '15 at 14:52
  • $\begingroup$ This works very well for me, I just have to be careful when I subdivide the separated object in zBrush in order to avoid holes at the transition between the two meshes; masking out the backside of the canvas and only subdividing the front plane works well though and it looks lot "cleaner" than my previous approach using triangle fans Thanks very much again, this was very helpful! $\endgroup$ – n-i-c-l-a-s Aug 1 '15 at 14:52
  • $\begingroup$ @n-i-c-l-a-s, complex items are often constructed of multiple objects. Consider a child's tricycle, and assume it is to be animated to the wheels turn, and the handle bar and front wheel turns, as well.. Starting from the back, the rear wheels will be one object, the frame and seat a second, the handle bars and front wheel assembly a third, and the front wheel and pedals a fourth. Now, in the case of the frame, the seat and the frame may be part off the same object, but for convenience, they may constructed of separate mesh sub-objects. $\endgroup$ – brasshat Aug 1 '15 at 15:57
  • $\begingroup$ Keeping the sub-meshes in proper orientation to one anther is made simpler by using the parenting facilities built into Blender. $\endgroup$ – brasshat Aug 1 '15 at 15:57
2
$\begingroup$

I am glad you asked that question. I see many meshes from beginners with triangles or Ngons (polygons with more than 4 vertices) when it's not necessary. In most cases it's easy to change mesh density using only quads--once you know how. The image below shows some examples to get you started. Good luck.

$\endgroup$
1
$\begingroup$

You can't subdivide a plane without turning the adjacent faces into ngons or triangle fans. If you want the mesh fully connected you must create a transition between the two resolutions.

Within the one object you can have several disjoint meshes with highly varying amounts of geometry (granularity of tessellation). You can detach the plane from the surrounding geometry by selecting and and pressing Y and then subdivide it (that way the surrounding geometry isn't modified because it isn't attached in a topological sense)

enter image description here

I've scaled down the subdivided plane in the last example to show that it is detached from the surrounding geometry. It depends heavily on your use-case which of these options is most suitable.

$\endgroup$
  • $\begingroup$ Thanks very much for your reply! This helped a lot! In particular the part "You can't subdivide a plane without turning the adjacent faces into ngons or triangle fans. If you want the mesh fully connected you must create a transition between the two resolutions" cleared up a lot of confusion for me. As I wrote above, I couldn't think of a solution other than using ngons or triangle fans and it is very helpful to know that there is none :) (except for separating the object into mutiple meshes) $\endgroup$ – n-i-c-l-a-s Aug 1 '15 at 14:53
  • $\begingroup$ I think in my case, the triangle fans wouldnt be too much of a problem, but I thought this was bad practice, because people seem to be very fond of "all-quad"-geometry.When I tried this first the triangle fans kind of screwed me, because I didnt think about it sufficiently and stupidly had all the triangle fans on the front of my "canvas" (instead of the sides) which led to some very ugly artifacts later on(after sculpting and painting), which were not visible in zBrush, but became very apparent when I rendered the lowpoly mesh with a normal map in other applications. Thanks very much again! $\endgroup$ – n-i-c-l-a-s Aug 1 '15 at 14:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.