Question about meshes

[ammobake]

I am currently wasting time by modeling a hawker hurricane in 3D in AutoCAD 2016 (see attached).

I found the plans online that included cross sections and I am essentially trying to use RULESURF command to mesh between the closed polylines at each cross section of the aircraft.

 

Sometimes, this results in the mesh becoming twisted up on itself like a tin can.  However, sometimes the results are decent - fore example, these two surfaces are OKish.

HawkerHurricane.bmp

Loft sometimes works between these surfaces but a couple times it has crashed AutoCAD and not sure why.

 

To trace out the cross sections I sometimes used splines and arcs and eventually used pedit to convert everything to polyline and close it up.

SO methinks my level of detail on the polylines might be what is crashing CAD when I use Loft on some of these.

 

Does anyone have any tips on how to get RULESURF to work properly?

Currently, my SURFTAB1 = 96 and SURFTAB2 = 96.

 

I am going to go back to the wings once the fuselage and canopy are modeled. 
I am planning to use a similar modeling technique and use cross sections of the wing every foot or two.  Once one wing is comlete I will mirror it over, basically.

 

Thanks!

 

-ChriS

 

 

 

HawkerHurricane.dwg

[ammobake]

So after fiddling with this I found something cool I thought I would share.

It appears the easiest way to get RULESURF to work properly in this fashion is to use arcs and open polylines for half the aircraft instead of closed polylines going around the entire airframe.  You can mirror the meshes later on to close it all up.

 

I'm not sure why this is the case but I've almost got the entire fuselage meshed out and theres been zero problems.

 

What I ended up doing is cutting the cross sections/profiles in half vertically.  I then started stacking them up at the correct height from a centerpoint at 0,0.

HawkerHurricane2.bmp

 

So I pasted the cross sections from their centerline to the centerline reference point.  From there you can either use move command to the correct Z elevation or change elevation variable in properties.

 

Once I would get 4 or 5 sections stacked up I went back and used RULESURF in between each.

Once I finally get meshes done for the entire length, I can mirror the entire project at the very end and it will be done (for the most part).

 

I experimented with using 3d Solids for this and I guess it can be done.  I extruded the side view of the aircraft and use the plan profile to create a template for subtracting from the side view (if that makes sense).  This way the 3d solid only keeps the extents of the side and top views, respectively and with one step it subtracts everything outside that.

 

3d modeling like this is refinement and patience - slicing and filleting off the sharp, unwanted edges as you go. but the meshes seem to make this a lot easier.

 

You could just use the loft command in between each cross section too.  The resultant entity is a "loft" surface in that case.  But if you use the loft command in between the closed polylines it seems to crash autocad if you have too many polyline vertices lol.

 

When I get super bored I like to find aircraft drawings online like this that include cross sections at various intervals and draw it up in 3D.

ChriS

[lrm]

I see a few major problems with your profiles. 

1. The starting points for each profile are not consistent.  The yellow dot shows where each polyline starts.  YOu want them all to start at the same relative position (e.g., bottom corner or center top, or...). 

2. The polylines do not go in the same direction. G,H,and J are clockwise whereas K and L are counter clockwise. They should.

3. The polylines have way too many vertices. 478 for L, 544 for K, etc. !

I think creating half profiles and mirror after making a solid is a good strategy.

 

[ammobake]

Yeah I think so.  I also simplified the number of vertices on the polylines for the half profiles.  The rulesurf seems to work a lot better with open polylines anyhow which was kind of nice.

I got most of this drawn up except for the minor details on the cockpit and the stabilizer on the rear but I'm gonna worry about that later on.  I've started on an F16C lol.

 

I wonder if it was trying to twist up the mesh because the direction the polylines were drawn in were different .  It seems weird it would affect anything but It would make sense why rulesurf would project straight across for some closed polylines but twist it around for others.

 

Thanks!
 

ChriS

Edited May 20, 2019 by ammobake

[lrm]

Rather than using polylines for the sections I think you should consider making the half section single splines.  Are you comfortable with the manipulation of Fit and CV points of a spline?  If not, you might consider "tracing" some of your polyline profiles, or the pdf drawing to create a spline that is a satisfactory representation of the shape. 

I would start by creating a spline using a dozen or so fit points along the profile. Do not worry about the resulting curve duplication the original. Use close together fit points where there are abrupt changes in curvature.  Use few fit point where the curve is relatively smooth.  The fewer the fit points the less likely you will get undesired ripples.  

The next step is to change the editing mode to Control Vertices and see how you can improve the fit by moving the CVs. Use splinedit e a to add additional CVs if they are needed.  If you need a sharp corner in the spline then drag two CVs so that they are coincident.  I am in the habit of creating profiles in a CCW direction.  In 3ds Max this ensures that surface normals are pointing outward and reduces confusion later in the modeling process.  

The attached file includes your polyline profile (green) for station J  and a spline  (yellow) I quickly created.  It needs some tweaking but you can get a feel for the CV layout.  It is important to note that the first two and last two CVs of a spline control both the radius of curvature and the slope of the spline at the ends.  If you do not want a crease at the bottom of the fuselage for instance, make sure that the last two CVs are in line horizontally.  Your polyline has 546 vertices while the spline has 28 CVs. I think you will find you will get a better quality solid model using splines and not polylines.

section-J.dwg

[ammobake]

Yeah that's cool man.  In some ways the splines make it easier to trace out the profile too with all those crazy curves and stuff.  I think the reason mine ended up so intensely packed with vertices was that I converted all my arcs and splines to polylines using precision 30 (for some weird reason) when I really didn't need to.

 

-ChriS

[ammobake]

I think I've figured this all out now.  I applied the same methodology to an F-16C.  I was able to find a very interesting drawing online from the contractor that had cross sections at every foot or two along the airframe.  The hardest part has been the little nooks and crannies.  The cockpit was also interesting to model.

 

Thought I would share

 

ChriS

F16_-_1.bmp

[ammobake]

Final result for the F-16C (not perfect but still neat).

 

ChriSF16.bmp

[ammobake]

I wanted to share the drawing here but it's 10.4 MB lol.

One thing I've learned about doing meshes and lofts is that you don't have to get too crazy with number of vertices for polylines and splines.  After the meshes are in, you can smooth to the desired smoothing factor later on and it will clean it up nicely.

 

So, for example, you don't need to use rulesurf command with like 300 surface tabulations.  (surftab1, surftab2 commands).  All you really need is like maybe 60 ish and use smoothing factors for the rest.  The more tabulations you have per mesh, the more it bogs down the memory too.

 

For lofting, the command would occasionally crash my computer.  BUT this was my own fault and was because of the number of vertices I was using originally.

 

Also you always want to use join command to form your cross sections.  If you use pedit and join from there, it converts all arcs and splines to series of straight lines at the desired number.  But the lofts and meshes like smooth edges to close everything up.  So join command is ideal because it maintains the smoothness of the curves but without adding 3 times the vertices.

 

What was interesting was the end-result mesh shape.  The final shape would vary depending on where I would click the two polylines to connect.  Very strange.

 

In some cases I was having trouble with the shape not being to my liking.

In those cases I would explode the cross section, join it back up again, and in many cases it would draw the mesh perfectly fine.

 

-ChriS

 

 

Edited June 10, 2019 by ammobake

[f700es]

14 hours ago, ammobake said:

Final result for the F-16C (not perfect but still neat).

 

ChriSF16.bmp

Dude, upload something like a png file or even a jpg. Stay away from bmp files

[ammobake]

Sorry bout that here you go lol.

 

ChriS

 

[ammobake]

I did attempt this with an F117.  In that project I attempted to create one 3d solid and "hone" the 3d solid down to it's final product (at least the fuselage portion).

It was much more challenging than I expected, despite the straight line edges of stealth aircraft.

 

I was able to get the nose and most of the fuselage completed but couldn't quite get the tail to work out.  Would share but it's on a server of a previous employer lol.

 

The most challenging part about the F117 was creating the primitives needed to manipulate the 3d solid in the proper way (if that makes sense).  Much of the geometry was for reference only and got way too cluttered.  Orbiting around in 3d with stuff overlapping gets confusing.

 

On the F117 the drawings I had weren't as detailed as the F16C.

I had front, rear, side profiles that had been posted online many years prior and that's pretty much it.  But it looked gorgeous.

 

The simplicity of straight line edges I guess - Reminiscent of the 1980's in general (Although, historically, the first 2/3 scale technology demonstrator of the plane that would eventually become the F117 was built at Area51 back in the late 1960's).

 

Kind of puts things in perspective as to what we likely have these days.

 

-ChriS

Edited December 11, 2020 by ammobake

[Ace737]

Thanks Ammobake, I'm doing exactly the same. Using an old RC-plan of a C172, to model this to a 1:1 scale cockpit mockup, which I can then CNC and "glue" together. Very useful to follow this thread, knowing I'm on the right track.

 

Anco