Cineversity

I’ve been trying to figure this out for awhile and I can’t…I’m thinking it shouldn’t be hard.

I’ve modeled a flange object with some holes making sure I’m using all quads that will work fine when subdividing.  It works fine but when I extrude a pipe and drop it into a subd the extrusion looks faceted, while the original flange part of the object appears as I was expecting.  The extrusion looks okay with the subds off, but I was hoping to keep it all as one object and would like to know why the faceting appears when subds are on…and what I can be done to make it appear correctly.

Any help would be appreciated.

Thank you.

Scott

Hi Scott,

Would you mind to share an example scene file? I will have a look into it. The Phong Tag (even if it sounds counter intuitive) would be the first part I would look into, then the checking the polygons itself, etc.

Edit: it is hard to tell from a small and compressed image what the reason is in your case. If you like, just delete 7/8th of the model, before you attach it here. /edit.

Cheers

All the best

Sassi

Here’s the scene.  Thanks for taking a look Dr. Sassi…

Hi Scott,

Thanks for the scene file, it makes it so much easier to tell what is the reason for that.

My first impression is, that it is too dense from the start. You are close to have a model that feels already SubDivided, but it isn’t.

I know that the holes are the starting point for the density of the mesh. There are ways to get the density down, which is work of course but helps to lower the work-load all the way down to the rendering, etc.

The target is always an efficient mesh, not only for render-time, the more points your model “owns”, the harder it is to work on changes. This object has 4608 Polygons, and the Subdivision pushes it on “level 3” to 299.520 polygons. This is certainly huge for a single and simple object like this, which could be seen as an subjective input of mine.

Quads are certainly a good idea, but the ideal is a square, not a super long rectangular. Which could be a “lead” to the problem, but I guess there is something more prominent in the object: Check the image that I have attached. Each second edge along the Z axis follows NOT the circle that it certainly is supposed to follow. If that is only every second edge than the Phong Tag might blur that away (hence my first check on these). With the SubDivision, this change of angles each second time will be amplified, will say the flat part is more pronounced. The angle between two neighbors is calculated along the Phong settings to round the surface—from one normal pin to the next, as an progressive interpolation, soft and smooth. If the SubDivision introduces more edges, these edges will have their own “Phong/Normal” calculations. In other words, the area that was smooth before has now shrunk to an 1/8th of its former space (3to the power of 2 equals 8—the result of Subdivision of 3). In this way the “faceting” of the original surface will be stronger visible. The problem you ask to solve.

In this Multi-gon shape instead of a circular shape I see the main problem. Secondary would be the continuation of the high density of the mesh and the long rectangular polygons. Any detail might work with less polygons.

I placed the object to a XYZ 000 position, and tried to rotate the object, so I would have it symmetrically. As we have twelve holes the angle between two holes is 30º, so I set up the R.B to 15º, but it didn’t aligned, I had to go to 11.3º, certainly not a “biggy”, but a sign that something is off. If you go to the holes and switch to the Line/Wire view you might see even more. Checking the mesh, resulted in 96 non-planar polygons (Edit>Project Info), which is not a problem normally, but I do not see a reason for that, they could be all planar in this model. They are all inside of the hole, so not a concern at all, but an indicator to me, as you asked to inspect this object to find answers.

I do not critique nor criticize here your model or anything thereof, I share my findings. All in all, your ideas how things have to be is not far from being successful, just a tiny little bit more care perhaps to make your work great. You are very close, go further!

As a tip, I show you a little procedural—how to reduce the density while you move on.

All the best

Sassi

P.S.: I used a reference Circle Spline which I had set up to 1º for the next intermediate point, to get the best “reference”.

Thanks so much for inspecting the scene Dr. Sassi.  I gladly welcome any critiques and criticisms!  Up until now, I’ve just been a forum ‘lurker’…almost afraid to ask what most would deem a silly or obvious question.  I’ve come to the conclusion that I just shouldn’t care and ask away anyway.

I suspected there was something wrong from the start and my initial mesh density had me questioning the way I was approaching this. I noticed the points not perfectly aligning after extruding and could not figure out why that happened. I’m going to give it another try with a different approach and see if I can get better results.

Thanks again.

Scott

Hey Scott,

Thanks for feedback. An old saying is—there are no stupid question (only stupid answers perhaps.)

My idea is that anything that is not clear—will stop any further progress along the way. It will always sit there and enjoy to be the “black-hole”. I’m glad you ask, and I guess others will enjoy to have that question answered. So, thanks to you.

I have attached as well a screenshot how to lower the polygon density over time. Sometimes we need lots of polygons for a single area, but it should inherited to the rest of the model. You might find other solutions to lower the amount. Just do a little sketch and test how it works with Sub-Divisions. Note that I have only quads in it, as long they are flat they wound harm the surface, even if they are not symmetrically. You know already the idea of the “support polygons” to establish a nice “bevel” in a Sub-Division set up, this will be needed around those “reductions”. Note that I went from 27 polygons on the left to one. With subdivision set to three, this would read as 216 to 8 polygons as an result only. A lot of reduction, and that pays of in rendering. Of course it needs to fit with the requirements of the surface in the first place.

Have a great weekend, and if there is a question, fire it up!

Sassi

P.S.: A little scene with three (more or less) extreme polygon distributions. Each has its strength and needs, if used in the right context, they will perform well. It always depends on the needs. Each starts and ends on the left/right side with 31 polygons. ;o)

As I normally do the things I suggest here, I went one time trough the model, and of course used the “final” version, to apply my polygon reduction idea from above.

I have left the steps as copies in the scene, so if one likes to re-engineer the steps—have a look.

I had split it to work only on a 1/12th, and that base element has only 156 polygons. I wouldn’t go much further, even if a lower number seems achievable. I guess is a gut feeling when a balance (quality vs time vs points) is reached.

:o)