# Fill - Curved Surfaces - Interesting Inconsistency, Need help!

Hello Community,

I have an interesting problem. I have shrunk the end of an open cylinder and created quadrants to bring the open end to a point.

For the bottom 2 quadrants I selected the two lengths and all the ‘curve’ edges of the cylinder and Hit -> F to fill it. For these bottom 2 quadrants the fill created a single face with a surface that appears nicely curved.

For the top 2 quadrants I repeat the steps and instead of a neat curved surface, instead I get a semi-circle and a triangle at 90degrees to each other. What happened the first time and how was it different the second?

Pic 1 is looking from outside, just inside the tip ( ignore the rig in the background )
Bottom lefdt and bottom right sectors have curved surfaces. Top right is empty and top left has a non-curved surface (you can see the intersecting line).

Pic 2 is looking out from the inside at the inside of the tip. You can clearly see the chord. If I move the camera down to the sector below you see a curved surface with no chord and the ‘face’ dot indicator hanging ‘in mid air’.

Pic 3 Shows nicely the difference between the bottom of the cone and the top. Bottom being neat and curved, top… well, not. You can see the semicircle that appeared, which did not appear on the bottom quadrants.

I repeated the same steps for all quadrants but yet the results are very different despite the cylinder being uniform.

Here’s a couple more pics to illustrate. Pic 1 is the selection, just before I hit F.

Pic 2 clearly shows the result - a triangle with semicircle, which is simply not the case of the lower 2 quadrants which both appear as complete curved surfaces.

Hope you can help!

Kind Regards,

John

Blender is trying to draw a polygon that can’t exist in the real world, ie a non planar polygon. As you look at the representation from different angles it looks different. The word ‘Appears’ is the problem here. If an edge is not being drawn as a line there is no edge there.

Hi DruBan, thanks for the reply. Whereas I completely agree with what you are saying, I have looked at my top and bottom quadrants at all different angles and they are actually drawn differently. The ‘curved’ surfaces appear to have drawn a line from the tip to the tangent of the curve, from all angles, but the top quadrants do not have this line and instead show a clear intersecting chord. I think the first picture is the clearest to see this. Again, viewing from different angles does not ‘resolve’ the problem. Blender has clearly handled the curved surface on 2 faces (surfaces) and clearly not on the other 2, from all angles. The 3 pics in the first post demonstrate this. If I flipped it up 180 degrees I’d expect to see the same thing from the bottom quadrants but this does not occur.

Using your keyword, the bottom quadrants appear curved and correct with no intersecting chord on the inside, from all angles. The top 2 appear to be a triangle aligned with a chord, from all angles.

Let me iterate this differently. It is impossible, on the bottom 2 quadrants, to view them as a triangle and semicircle from any angle. They are always correctly curved surfaces.

I’m Baffled.

Here’s another illustration of the point.

The above pics are taken from Wireframe Editmode. One can clearly see the face of the curved surface. This interconnecting line from tip to tangent is simply missing in all the faces which do not appear correctly. It does not matter on viewing angle, the same two are always right and the same, other two are always wrong.

Maybe the question is not, why the broken surfaces, but is actually; Why the perfectly curved surfaces (from all angles)?

To quote DruBan:

“If an edge is not being drawn as a line there is no edge there.”

So, The quadrants are Actually different. Why?

The problem is that you’re using ngons in a non planar situation.
And in Blender, that’s not going to work well.

You can try to “help” the ngon in your situation by selecting 2 vertices (the top one and one of the bottom) and press J to add manually an edge where the blender ngon is not able to “understand” it should fake it.

Hi Sanctuary,

Does this suggest that, the first two quadrants did not create an ngon situation, and only after the third quadrant was created Blender started to have difficulty? This would explain why the apparent difference.

I guess I’m not actually going to leave it this way, I am just really curious as to what is going on in Blender

Already said with so many words but maybe this helps to understand it, with visuals.

Yes, because ngons are in fact a hack that hide the triangles under it, it has lots of limitations, i agree it shouldn’t be like that but unfortunately for us that’s very likely one of the ngon limitations playing there.

One of the reasons you should avoid to use ngons in a non planar situation, or at least use them only temporarly.

Because as an example , i have this half cone

that i convert into a non planar ngon

And if i look around , the ugly happens like in your model :

the only way i can fix that while keeping ngons as i mentionned previously is to “help” blender with added edges to guide the ngon shading

You could try to report that problem to the bug tracker if you think it’s really slowing you down in your modelling, in case it’s not one of the limitations but a bug, maybe one of the devs will see better what’s going on.

Thank you, JA12 and Sanctuary for your visual examples.

In the case of the half cone, in my case the first half cone appears without any ugliness from any angle at all and only the second half cone shows the ugliness.

I am inclined to think that this is due to Images Draw Method setting being set to GLSL, if its even related. I could be wrong.

I understand there is some fakery going on due to it being non-planar. I suppose I expected the ugliness to show in the half cone as that is what I normally see. I am baffled as to why Blender ‘chose’ to display a perfect curved surface, from all angles!

In any case, thanks for the explanation. I shall have a play with the settings when I have the time and report my findings.

Thanks!

Regards