Pros and cons of model design programs

[Tyler Durden]

I have been following these forums for a while and I have been a longtime builder of paper models. Mostly Halinski stuff, GPM, and Fly Model. I have been thinking about getting into the designing side of modeling and after reading the excellent Virtual Airplane tutorial from Wittold Jaworski and the Fundamentals of Paper Model Design by Pixel Oz I still have a very basic question.

Which design software and why?

I am a fairly good amateur with Fusion 360, have used SketchUp in the past, and I have downloaded Blender but have zero experience with it although the graphics and rendering possibilities of Blender seem to make it necessary for the types of models I have in mind.

My interest leans towards very accurate highly detailed and realistically colored and textured/worn models like the ones from Halinski. Thus, the ability to add wear, dirt, oil, grime, etc to the model along with realistic panel lines and rivets is necessary. Since I have built enough of the Halinski models I can appreciate the skill in designing where paper thickness is taken into account and parts always seem to fit precisely if they are cut properly.

Fusion 360 has the ability with sheet metal tools to unroll and account for thickness. I understand quite a few other programs have unroll function but for complex parts like fairings, rounded nose cones, parts with gores, and designing internal support frames I am wondering how useful they are compared to Fusion 360.

Is it worth learning a new program because they have design functionality superior to Fusion 360? What are some pros and cons from those who have actually designed their own models?

[Laurence Finston]

Originally Posted by Tyler Durden

Which design software and why?

That's a very difficult question to answer. The first 3D graphics software I used was AutoCAD, which still seems to be the program of choice for professional applications. The problem with it is that it was and is very expensive. I looked into it recently out of curiousity and it seems that it's marketed on a subscription basis with substantial monthly payments. I didn't research it any further. I would say it's probably only worth it for professionals.

There are lot of programs around for 2D but if you need perspective projections, there is less choice.

I use GNU 3DLDF, but then I'm the author of it and I'm pretty certain no one else does, so I won't go into detail about it here. One of my goals is to use only Free Software, and I use it in combination with MetaPost, TeX and GIMP. I bought a book about Blender and I have plans for using it for surface hiding and shading. That is, I want to pass the output of 3DLDF to Blender in order to use this functionality. At the moment, however, I'm more occupied with learning to use GIMP.

The most important thing any 3D software does is matrix multiplications. If it does that, that's "half the rent", as we say in German.

Fusion 360 has the ability with sheet metal tools to unroll and account for thickness.

Mathematically, the result of "unrolling" a geometric figures is called a "development". It is relatively straightforward to make developments of figures based on circles, such as spheres or cones. This is an example of a model design using a sphere development: https://www.gnu.org/software/3dldf/g...s/clstsph3.pdf
The source code is here: The GNU 3DLDF Astronomy Page

Unfortunately, it immediately becomes difficult after that. I've been wanting to do a model of an ellipsoid development for a long time but this requires "elliptical integrals of the second type" and I've forgotten most of the little I learned about calculus in college. One of these days I will sit down and figure this out.

I doubt that any 3D software package can automatically create developments of complex curved shapes. If they do, they probably "cheat" and reduce the shapes to polygons first. Of course, this is more or less what you do with calculus, except that you use infinitely small "polygons". However, I don't know this for sure.

[Laurence Finston]

I would say accounting for thickness is something the designer has to do. For any woodworker, it's second nature. I assume for metalworkers, too. I don't know how a 3D graphics software would automatically account for it, but maybe some do. I actually wouldn't want the software to do this for me.

[Tyler Durden]

Ok, thats what I needed to know.

I have Fusion 360 so price isn't an issue but it seems that Blender has so much more functionality when it comes to textures and baking it into the model that I will probably just bite the bullet and learn to use it, or make my models in Fusion then export to Blender.

[Laurence Finston]

I just looked up Fusion 360 and found that it's a product of Autodesk, which makes AutoCAD, but with even more features. So you've already got a top-of-the-line commercial product and if you want to build real airplanes instead of just paper models, you're all set. I agree that there would be no advantage in looking around for something else.

Blender seemed like a good choice to me for skins and textures, too.

[LE CRIQUET]

Hello,
I've designed around fifty models to date, and I still use the same methods that I believe (and I'm proving it) are perfectly appropriate for creating paper models.
The models are designed using 2D design software, and I use Metasequoia/Pepakura to develop complex volumes. Colouring with photo software.
In any case, the possibilities offered by free software (or software that can be acquired for a small price) provide ample opportunity to enjoy creating your own models.
The proof is in my website, and I'm happy to help and advise anyone who wants to. It's an exciting adventure, provided you have a basic understanding of design and geometry.

[Tyler Durden]

So the process would be to make a 3D model and then unroll the sections?

I am assuming that the model would be complex geometry, compound curves, etc. that would then be developed into flattened geometry by the unrolling program?

I started building a solid model of an aircraft in Fusion 360 but once I have it I am not sure how to unroll it. From what I have seen, Fusion doesn't actually have the capability to unroll, but depending on what type of file it is saved as, the model can be exported to another program which possibly could.

[Laurence Finston]

Originally Posted by Tyler Durden

So the process would be to make a 3D model and then unroll the sections?

Yes.

Originally Posted by Tyler Durden

I am assuming that the model would be complex geometry, compound curves, etc. that would then be developed into flattened geometry by the unrolling program?

I don't believe that this can possibly work this way, because it would require higher mathematics. What could work is to convert surfaces to polygons, which is what rendering software does anyway, and then rotate them individually until they lie in a plane. This would be straightforward to program, but could take a long time, depending on the number of polygons.

This is a short video I made in 2005 that illustrates the process in reverse: https://www.gnu.org/software/3dldf/graphics/birth_1.mpg

In the general case, surfaces based on Bézier curves or NURBs cannot be developed analytically, i.e., by using a formula. This would only be possible if they represented surfaces that were, like spheres, cones, etc.

To develop a surface, you need to be able to calculate arc lengths. You take the length of an arc and draw it as a straight line in the plane. For circles, for any angle, you can easily calculate the arc length. For ellipses, this requires integration. For parametric curves like Bézier curves or NURBs, I've never heard of any method of doing this.

If you have an actual physical model or the object itself, you can approximate it by taking a piece of string, laying it over a section and measuring the length, but the result will probably not be very precise.

Originally Posted by Tyler Durden

I started building a solid model of an aircraft in Fusion 360 but once I have it I am not sure how to unroll it. From what I have seen, Fusion doesn't actually have the capability to unroll, but depending on what type of file it is saved as, the model can be exported to another program which possibly could.

I don't know anything about programs for making developments, but in general, doing this would be more of a research topic than an exercise. I don't think you're going to find anything better than Fusion 360.

[Laurence Finston]

For anyone who never learned this or whose math is a little rusty, this is the idea:

A parametric curve is one whose points are found using a formula that depends on a single variable, often called t. The points on the curve are found by varying the value of t. Often, it varies from 0 to 1 with the point found for t = 0 being the first point (p_0) and the point for t = 1 being the last (p_1). The point for t = .5 (p_.5) may be more or less halfway between p_0 and p_1 by distance, but there is no guarantee for this. The distances between p_x and p_(x+1) on the one hand and p_(x+1) and p_(x+2) on the other, where x is some number between 0 and 1, need not be the same and probably won't be.

If C is the curve, L is the length of C, M is the magnitude of the increments we use for finding points on C and N is the number of increments, we can try to find L in this way:

A first approximation would be simply the length of the straight line between p_0 and p_1. The next would be the sum of the straight lines between p_0 and p_.5 and p_.5 and p_1 (M = .5, N = 2). Then, for example, M = .3 and N = 3 to divide C into thirds. The approximation would be the sum of p_0 to p_.3, p_.3 to p_.6 and p.6 to p_1.

L is equal to the limit of the sum of the distances between the points as N approaches infinity and M approaches 0. That's how integration works. The best one can do with a parametric curve is to use a high enough value for N and a low enough one for M to get a reasonable approximation. This is a so-called "numerical solution". With computers, there are only ever numerical solutions since they haven't invented a computer that can perform integration (or many other things) yet.

[mbauer]

Originally Posted by Laurence Finston

Yes.



I don't believe that this can possibly work this way, because it would require higher mathematics. What could work is to convert surfaces to polygons, which is what rendering software does anyway, and then rotate them individually until they lie in a plane. This would be straightforward to program, but could take a long time, depending on the number of polygons.

This is a short video I made in 2005 that illustrates the process in reverse: https://www.gnu.org/software/3dldf/graphics/birth_1.mpg

In the general case, surfaces based on Bézier curves or NURBs cannot be developed analytically, i.e., by using a formula. This would only be possible if they represented surfaces that were, like spheres, cones, etc.

To develop a surface, you need to be able to calculate arc lengths. You take the length of an arc and draw it as a straight line in the plane. For circles, for any angle, you can easily calculate the arc length. For ellipses, this requires integration. For parametric curves like Bézier curves or NURBs, I've never heard of any method of doing this.

If you have an actual physical model or the object itself, you can approximate it by taking a piece of string, laying it over a section and measuring the length, but the result will probably not be very precise.



I don't know anything about programs for making developments, but in general, doing this would be more of a research topic than an exercise. I don't think you're going to find anything better than Fusion 360.

Wondering if Fusion 360 shows you what the ARC length is if you select it and look at the property box for it?

I use AutoCAD and Inventor. Inventor has several features I'm trying to learn to be able to do my models in 3D. I also have Fusion 360 as part of a package deal that I bought back in 2019. I did not see an unfold option in it.

Still doing my drawings using 2D in AutoCAD 2006, switching to 2019 and feature based 3D modeling is a pain!

Yes, I refused to upgrade and kept my Windows 7 machine up and running. It never goes on-line now, bought three exact backup machines when Microsoft quit supporting 7. If I have any issues, I use the two spares to backup the original, simple to switch the machines out and keep drawing. Hoping they (3-machines) last for the next 20-years.

Mike