"What are you using MeshMixer for?"

I wanted to convert all the lofts into one body and smooth out the surfaces. On the edit tab there's a command "convert to solid" which I pushed to the limit. Some heavy calculations! After the conversion I smoothed it where needed and possible with "Deform> Smooth" on the "select tab". Works great! I was aiming for this years ago and were blown away by the effectiveness of Mesh Mixer! But converting these meshes back to something worthy seems to be a very big challenge for all the experts in programming. Rhino's user guide have some useful insight about this. TSpilnes as mentioned by OSTexo seems to be an alternative for this sort of work. I tried it for a short while but the results were a bit confusing compared to some videos I watched online. The only way seems to be GMD and MoI for now :)

"it seems you measured the points manually on the tape and plotted them in a CAD program?"

That is correct.

Do you mean that all the surfaces (and molds?) are made by hand?

Yes. The engineer and lead designer behind it, Wolfgang Vormbaum, knows what he is doing! But its impossible or would consume too much time to get these shapes 100% symmetrical by hand. Which is why I need make the most of all I have at my expose to get this right. The next aircraft we will be manufacturing will have this whole fuselage mold CNCd. Smoother surfaces for a lighter body and reducing turbulence. Also cuts down on man hours. In this evolving world anything which saves time and money is worth looking into right?


Just a matter of time before we start testing our prototype. The engine is made by a team located in Durban, South Africa. Adept Air-motive: https://www.youtube.com/watch?v=GVkV_VITXBY

Some old previews of the prototype could be seen over here: https://www.youtube.com/watch?v=dzMqrGDE0Y4

Do the lofts in MoI have the ability to use guide curves?

Elrick,

I actually use SolidWorks, but have been looking at Moi and Rhino for better surfacing tools. But a quick run through the Moi help seems to indicate a network surface in Moi might give you a bit better control if loft does not.

I have done a fair amount of reverse engineering work over the last few years, and now that I understand it, that is what you are doing with this hull.

If you have a concern over how faithful your CAD model is to the hand made form, then you also need a way to measure the difference. I think getting a good scan of the hull will benefit you by providing a reference for deviation analysis, as well as giving you a reference for the modeling process. However, I don't believe there is any way to go directly from the scan to a 3d cad model without a lot of work. Some high end programs like Design X (Geomagic) can simplify the process, but I have not seen anything like your hull produced with Design X. I would love to know how it would handle something like this.

I understand the Next Engine did not work well for you. Perhaps there is another scanner or service that might. Your part is large, so a long range scanner might actually work fairly well and give you reasonable accuracy. Some of the laser scanners that attach to digitizing arms like the Faro only scan about 1-1/2" swath at on time so it would take a VERY long time to do your hull that way. I work with someone who has used a Focus3D (also from Faro) to do body panels. I constructed the nose of a truck using the long range scan from the Focus3D as a modeling reference.

Hello,

Structured light scanning would probably work well for your application. I fuselage itself doesn't look overly complex, however you might want to plan out your model with a variety of techniques that will allow you to more easily change the different surfaces of a model more easily than loft. You typically have less control over surfacing when using loft as opposed to other surfacing strategies. It looks like there are several major components to this model, having a lack of control over surface modification if you need to make a change would be brutal if you lofted that entire model.

Hi Elrick,

> Do the lofts in MoI have the ability to use guide curves?

In MoI that's called the Sweep command, it can make a surface going through profiles and using either one or two guide rail curves.

- Michael

Is this a case where Sweep and Network Surface overlap in functionality, or is it preferable to use Sweep if you don't need more than 2 different guide rails?

For reverse engineering a bigger question is, if there are curves produced from a 3d scan or reference model, which surfacing tools will generate a surface that actually passes through the curve?

Hi,

Tried to keep it short but thanks for sharing all the advice and tips gents!

"I constructed the nose of a truck using the long range scan from the Focus3D as a modeling reference" - Do you have a place where you share pics? The Faro products seems very interesting! How many times did you need to move the arm's position for the measurements? How many days did it take?

"I actually use SolidWorks, but have been looking at Moi and Rhino for BETTER? surfacing tools" You are confusing me when you say this! :) Cos this --- http://www.raptor-aircraft.com/home.html --- made me believe Solidworks gives you everything in one package? Looks like SW do FEA calculations as well. Not too sure about the renderings but I believe Keyshot has this under control! :) We were talking about buying Rhino to handle the surfaces. I believe our next investment will be either Rhino or SW.

"If you have a concern over how faithful your CAD model is to the hand made form, then you also need a way to measure the difference." - At this time I have some freedom to make modifications to the original design. I had to cut the ribs for the hull with the CNC. (skeleton) My cut parts were spot on. This is where I spotted the small symmetrical differences.

"Perhaps there is another scanner or service that might." - A sales representative of 3Ds gave us a quote for a complete scan of the fuselage. He did some demonstrations with the Arctec and Sense 3d scanners. It might be an option for the future cos we wont be able to afford something like the Arctec (thats if we are really looking for reversed engineering tools). The quote is equal to 3 months my salary. So its better to continue with the data we already have for now. I did some turntable scans with NextEngine. The results were great! Its very handy for parts to be scanned by the table! But the accuracy is lost with single scans. Its quite an effort to assemble the single scans. And after the fusion the comparison were off by more or less 3% over a span of +- a meter. Didnt take me long to realize single scans wont work.

"having a lack of control over surface modification if you need to make a change would be brutal if you lofted that entire model". I fully agree with you on this! Making modifications in GMD is brutal! Because a 2mm difference in the guide curves makes a hell lot of a difference in the shape. Thats why I hope MoI will help us achieve our goals. We are talking about making a 4manned version and I would like to see if I could maneuver the lofts like Andrei did in his videos. His method seems good and the results of his rendered boat looks flawless. That way it wont be a hazzle to modify surfaces. There might be one problem and I hope Im not aiming of target here. My average control points through these profiles were about 22, which sounds like a lot? Will I need to make blue prints of the fuselage profiles and project them in MoI? Sounds like a lot of work but totally worth it if works.

Regards

 

 

 

 

Some related pics

Elrick,

The truck was scanned with the Focus3D from 6 locations (6 scans). Actually, we did not need most of what was scanned, just the area around the grille and bumper. Each scan typically takes 5 - 10 mins but this can vary depending on accuracy (sampling), point spacing, and color. If you don't need colored points (the scanner can take a set of photos to color the points) 5 mins or less per scan would be typical, particularly if you limit the scanning angle. The scanner is placed on a tripod and it rotates automatically around the vertical axis.

SolidWorks is good for many things, but constructing freeform or organic surfaces can be difficult and time consuming. Additionally, it does not have the tools to make the quality surfaces that are often needed for cosmetic work ie. Cars.

SolidWorks has various levels and add on packages, and it can get quite expensive. I just have the Standard version.

Hi Elrick,

> My average control points through these profiles were about 22, which sounds like a lot?

Well it depends a lot on the particular curve.

Having 22 control points for a very short flat line segment is indeed sort of a lot. But for a longer bendy curve it's not particularly a lot at all...

That's not an amount that I'd be particularly worried about, I'd be more generally concerned about the overall quality of the shaping of the curves.

- Michael

@chrisd,

> Is this a case where Sweep and Network Surface overlap in functionality, or is it preferable to use Sweep
> if you don't need more than 2 different guide rails?

There is some overlap between Sweep and Network in some cases like this, but they also do go about constructing the surface in slightly different ways which will have some impact on how the surface is shaped in the middle areas.

See here for some previous discussion:
http://moi3d.com/forum/index.php?webtag=MOI&msg=3081.23
http://moi3d.com/forum/index.php?webtag=MOI&msg=5296.3


> For reverse engineering a bigger question is, if there are curves produced from a 3d
> scan or reference model, which surfacing tools will generate a surface that actually
> passes through the curve?

By default all surfacing tools will make surfaces that pass through the input curves to a tolerance of 0.001 units (for objects with a normal overall bounding box size). The only way you'll get less precise surfaces is if you change some of the options like set profiles mode to "# points" mode and use a low point count for the reconstructed profile count.

- Michael

Thanks Michael! Im getting the hang of these control points. My difference between the profiles are usually not more than 0.2 mm/ 0.008". Its easy to inspect in GMD. My middle finger gets tired from zooming all the time. Getting eager to start :)

Chris, Your scan and alignments looks very neat! Seems like you also placed the newly made part over the scan for comparison? Were all these shots from GM Studio?

Elrick

The deviation analysis was from Geomagic (Wrap). The other images were from SolidWorks.

Hello,

Dynamic sketching is very useful in VSR as well as the retopo tools in TSplines. If the surface gets too complex in Tsplines I use the surface approximation, control point modeling and deviation analysis in VSR as a compliment to simplify the surfaces while keeping the deviation where it needs to be. Those plugins in conjunction with Rhino and MoI are great for reverse engineering and organic surfacing, and I use them regularly for that purpose, along with scan data. Only downside of those plugins is that both companies were snapped up by Autodesk so their future with Rhino could be in jeopardy.

Greetings,

Got MoI for myself now. At work I operate with 1600x1200 resolution (17") and here at home HD (27"). Very big difference! Wont touch the 17" again! 4K will have MoI very shortly :)

I started fiddling around with the lighting options to get the most out of the surfaces. It gets a bit confusing. Is some settings supposed to exaggerate bad curvature or are they really that bad? Which settings are most reliable? I REALLY like the ability to change lighting settings!

Also began tests shelling these lofts and the results weren't what I expected. Takes a bit longer (not that I mind, as long as it works!) and really bad artifacts pops up after the command is executed. Tried the body as a whole and separated surfaces. It appears to work flawless with some surfaces and others not. Probably due to bad geometry? Have A feeling that the last profile on the loft (very small one) might be causing this one on the nose..? Any suggestions


Thanks
Elrick,

Hi Elrick, the "metallic lighting" option will tend to make curvature more apparant, it basically makes the lighting work in a reflective style.

re: Shelling - unfortunately the geometry library that MoI uses does not have a particularly robust shelling mechanism and so shelling can have problems in many situations. It will be particularly difficult to shell or offset surfaces that have tightly bent areas in them where you try to use a thickness that is greater than the radius of curvature in those areas. That will cause the offset result to have badly formed surfaces with self intersections in them.

The geometry library in other CAD programs can tend to do a much better job of shelling than MoI, you might try exporting your model to ViaCAD which is inexpensive (like $99) and see if it can help you out with some of the more problematic shells.

- Michael

Hi Michael, Thanks for your time!

So the metallic lighting option would help best for surface inspection?

"It will be particularly difficult to shell or offset surfaces that have tightly bent areas in them where you try to use a thickness that is greater than the radius of curvature in those areas."

This explains a lot! Are there any software capable of measuring minimum curvature in lofts or splines? I know where this occurs in the lofted part and I might be able to get rid of this problem by moving the control points in the 3d splines. Might speed up the process if I could quickly obtain minimum curvature. If only it were possible to keep the curvature in a spline larger than the required shell this problem might be dodged then?

I have to say that MoI have a higher success rate in shelling than GMD! But GMD have an alternative way to shell. The one in MoI are similar to the "Thicken" command in GMD's surfacing tab. Offsetting the surface perpendicular to the selected one (which I like to think of as the organic way), and then the "shell" command where you select faces of a solid and the shell is, what I like to think as, done mechanically. This way is better to me since you dont have overlapping geometry when you do a shell on individual lofts and assemble or unite them later.

This shelling is very important to me. I had to make copies of the lofts and scale them down and boolean subtract them from the original to "shell" this body. This only made the lofts hollow. Then had to trim away excessive planar geometry. In this situation it might have been really great if you could scale down a part respective to the XYZ axis. Are there any such way in MoI? Manipulating the bounding box of a whole body..?

Thank you very much.

Elrick

Hi Elrick,

> So the metallic lighting option would help best for surface inspection?

Yes, I think so.


> Are there any software capable of measuring minimum curvature in lofts or splines?

Rhino has a CurvatureAnalysis command that can measure min or max radius of curvature and display it with a false color display on the surface.


> If only it were possible to keep the curvature in a spline larger than the required shell this problem might be dodged then?

Yeah it would help to avoid this one category of problem anyway with offset surfaces becoming self intersecting.


> In this situation it might have been really great if you could scale down a part
> respective to the XYZ axis. Are there any such way in MoI? Manipulating the
> bounding box of a whole body..?

Well, there is the Transform > Scale command - that will scale an object uniformly along the XYZ axes around the origin point that you pick. But I'm sorry I'm not really quite sure if that's the kind of scaling you're referring to or not, you may need to explain what you're looking for in some more detail.

- Michael

Hi Michael,

"But I'm sorry I'm not really quite sure if that's the kind of scaling you're referring to or not, you may need to explain what you're looking for in some more detail."

What I meant is to have the ability to Scale a part in all three dimensions rather than uniformly. When a rectangle prism is scaled the offset between the, original and scaled, surfaces wont be equal. I will share examples when I have some time. Not entirely sure but I think mesh mixer have this function where you could adjust the l/w/h of the bounding box. That way, when you do a boolean subtract, you could have a much more accurate "skin" rather than shelling gone bad. The inside of this fuselage aren't a concern really. But I need to cut on as much unnecessary geometry as possible.

Thanks
Elrick

Hi Elrick, there are a few different ways you can do a "non-uniform" scale with different scale factors for X, Y, and Z.

The easiest is to click on the size line in the properties panel here:




That will bring down an "Edit size" panel where you'll be able to enter in new values for the bounding box size:



By default it will scale uniformly but if you uncheck the "Maintain proportions" checkbox that will then allow you to change x y z independent from each other for a nonuniform scale:


Another way is to go to the Top/Front/Right 2D views and use the edit frame to do a one directional scale, some info on that here:
http://moi3d.com/3.0/docs/moi_command_reference11.htm#editframe
That's when you grab the sizing handle and drag it vertically or horizontally, you will see a tracking line and when that's active you'll be doing a one directional scale. Repeat that in other directions.


The other way is to use the Transform > Scale > Scale1D command, which allows you to scale things in one direction. You can use this repeatedly on all of the X Y and Z directions to do a non uniform scale in all those directions.

Hope this is what you were looking for.

- Michael

Thanks Michael! Quite some options! I tried your first method mentioned. Quite funny how you can shrink this whole body down to just n few millimeters. Something not quite sound this way. I'll make sure before I post.

Will try the other methods tomorrow! This is totally what I was looking for!

Much appreciated,

Elrick