Subdivision Modelling an Aircraft - Awkward Triangular Area

[Aured (Kris)]

I’m not sure if this is the right place to post this, so apologies if it isn’t.

For my first attempt at subdivision modelling, I’ve chosen an aircraft fuselage and I’m running into problems with my edge flow. The fuselage can be divided into three parts: A rounded top, a flat side and a rounded bottom. The flat side ends near the front end of the fuselage in a triangular shape, and both the top and the bottom join together from there. Each segment joins together with a sharp edge.

Trying to figure out the edge flow where the three conjoin is beyond me. If I leave the triangle as it is and inset the faces of the sides, I get an artefact. My latest attempt was to try turning the triangle into a pinched quad and continuing the entire side around the front of the craft, but this looks wrong in terms of accuracy to the aircraft I’m modeling.

I’ve edited the picture below to show the basic setup of the craft. The red circle highlights the problem area. It’s still left as it was when I tried to turn it into a quad, but the pinched end is supposed to be a triangle. Can someone please demonstrate how this could be done? Thank you in advance.

[Aured (Kris)]

In my second attempt, I created a small N-gon where the triangle ends. It conjoins the three parts wonderfully - but it leads to the part sharply jutting out. Attempts to rectify this have failed.

[DNorman]

I am not sure of the exact shape you want, but I would try to avoid a triangle in this case. You could extend the creased edge to go all the way around the triangle and bring the edges of the triangle closer together forming thin quads.

[Aured (Kris)]

Already tried that bud, appreciate the response though. I definitely need the three to come together at that point. I’ve managed to topologise it in such a way as they blend sufficiently - but the part where they meet ends up jutting out, and I have no idea why. Your model is bang-on by the way

[DNorman]

The point of the Ngon (or a triangle) is jutting out because beyond the point where the creased edges meet you have nothing to hold the geometry and the subsurface is curving your edges inwards. You have to either extend the edge crease around the tip or use support loops (also extended around the tip).

I prefer to use support loops as edge crease gives more shading glitches and with support loops you have more control of the sharpness of the edges. To get the support loops I beveled the edge with the triangle with 2 segments.
Notice that the triangle is still there but as it is inside the flat face it does not give glitches. It is best to use Ngons and triangles only inside flat faces, not on their defining edges.
Hope it helps.

[Aured (Kris)]

Your topology is fantastic - thats basically the way I have mine set up right now, and it does get rid of the artifacts. But the problem is now the imaged area. Before, it curved around uniformly as shown in red. Now it juts out as in your image.
The only solution to this I can think of is to add more supporting loops. But I’m out of ideas completely at this point.

[DNorman]

Haha, my topology is not usually fantastic! That is why I like these questions as they help me think.
Yes to get a general curve on the sides you need to add extra supporting loops.
Here is a cleaner way to do the triangle (this time all quad), it helps to curve the extra loops as all the loops are vertically straight.
I did teak the 2 vertices marked in red on the top image by moving them out very slightly on the x axis to compensate the point of the triangle pulling into the top and bottom curve.

Edit: this time I added the support loops for the sharp edges manually with loop cut.

[BlenderP1 (ProjectOne)]

I encountered a similar problem. In my case the solution was pretty simple - merging verts by distance and then recalculating normals.