I spoke too soon - More Mef Wheels

[NickD]

I seem to have been sucked over the spoke event horizon. Spoked wheels have recently been covered in both the Fiat 806 and Mefistofele threads that are currently running. The conversation touched on the potential for computer modelling tools to depict spoked wheels. I opined on how this might be done. I then thought I ought at least to demonstrate the approach rather than pontificate some impractical theoretical method I ought at least try it. Rather than hijack the thread further, I thought I had better create a mini WIP just on this small aspect. It's probably more of a how to really.

 

The goal is to start from either a drawing and generate a representative (but not necessarily completely accurate) model that could be used as a basis for 3-D printing. The lack of precision is due to i) me being a bit slap-dash; and ii) I wanted to have a minimum thickness that stood a chance of being printed. The output will be a file that contains all the components and consolidated models that can be used in a number of combinations and potentially could be used as the basis for any spoked wheel. The only step I don't plan to do is add the embossed detail, I am not particularly skilled at this and I want to get on with other things.

 

This will be relatively short, I hope but I will submit in installments.

 

I will use a free surface modelling tool called Blender.

 

The steps are: model the wheel and hub; model an architypal spoke; generate sets of spokes; merge all the components into a printable object. I will post the completed model at the end for information and should anyone wish to use it or modify it for personal projects.

 

Wheel and hub models.

 

The wheel I want to represent is that for the FIAT Mefistofele. Sources are: many pictures on the web, though few which cover the full circumference. There is also a copy of the line drawing from the kit instructions and of course the wheel in the kit.The wheels have 70 spokes. On the front wheels, there are 6 sets. The outermost is clockwise (hub to rim) with 14 spokes. then anticlock with 14 spokes. Both sets attach to the outer edge of the rim. Mid-rim there are two sets of 7 spokes that attach to the inner face of the hub. Inboard, the last 2 14-spoke rims that connect to the inner hub. So there are 28 spokes connected to the outer edge of the hub and 42 to the inner edge.

I chose to model the rim and hub two different ways. The first was drawn using the cross-section for the rear wheel using the drawing from the kit instructions. The result is shown below:

 

To generate the model, the image is loaded into a class of object called an "empty" object. That can then be aligned with the cardinal axes, scalled appropriately. If you are new to Blender the interface is a bit obscure but these are fairly straightforward once you get the hang of it.

 

Having aligned and scaled the picture. I added a circle (renamed it - eventually) and entered edit mode. Generating a shape is then a case of align the diameter with the drawing and successively extruding (E), scaling (S) until the required profile is generated. The wheel is modelled as a single object with 3 separate meshes comprising the tyre, rim and hub. This allows various combinations of the 3 components to be used.

 

Once familiar with the interface this type of object takes about 30mins to 1hour to complete, so fairly quick once set up. Counting the spokes took as much again!

[arni]

Way above my head but still interested.

[Jnkm13]

Looks interesting. Hope it turns out great!

 

Regards,

Jeremy

[sharknose156]

Well done on the choice Nick, very interesting idea. "Grande Première" here !!

[Roy vd M.]

Look forward to seeing more of this software wizardry!

[CrazyCrank]

Really curious and eager to see what will come of that....having after my Bug building, a Mef , a Pocher Alfa and a Supercharged Bentley in my stash ...I stay tuned

Edited October 21, 2016 by CrazyCrank

[NickD]

Thanks for the support guys. I sense a general tone that this stuff is Voodoo or Black Magic. At the start it certainly feels like that. However, the tool is free, what it can achieve is only limited by imagination and there is a bunch of really helpful how tos online. So surely it's worth a go. The first part of this WIP is not much different from the skills taught in a the classic bottle tutorial.

 

Reflecting on the post above, and now that I am slightly more confident about uploading pictures, I thought it might be worth a couple of extra pics.

The first shows how the first circle lines up with the tyre.

 Subsequent steps then add concentric rings at points where the geometry changes direction.

 

This procedes until the shape is complete.

 

At this stage it is very facetted. Blender sorts this by offering a very clever feature called a modifier. This is non-destructive. It generates an arbitrarily smooth surface through the points the user has entered. There are modifiers for all sorts of things that can be applied one after the other and switched on or off at will. Utterly mind-blowing how they make that happen. The other astonishing for the crustier amongst us is that my laptop (7 years old) is more than good enough to do this stuff in real time.

 

I repeated the above exercise using dimensions taken from an Italeri Mef wheel. I generated a pile of circles 

Not a very exciting picture - just goes to show that you don't need to have a lot of vertices to get a good result. from this the geometry is generated in a very similar way to the first wheel giving this...

The image is fuzzy but that is an artifact my image processing rather than Blender. Compared to the first wheel its is clearly, as has been noted elsewhere a lot thicker than the car actually has.

 

Direct comparison is easy. The wheels are concentric and scaled to the same OD. So switching off the modifier and highlighting the drawing-derived wheel gave...

One of the things the tool does make easy is this sort of comparison. One note of caution though: the alignment is only approximate. It is not obvious whether to use the tyre wall or hub as reference.That said the overall depth from tyre face to hub outer face is very similar.

 

Next up - spokes...

 

[sharknose156]

amazing, no need to ask you must be with NASA ! ( or a Vodoo priest ) 

 

 i am curious to see how the spokes would align.

 

Pardon my stupid question please, on the above picture why is the alignment only approximate ? is it because it is at this early stage ? the tyre looks simply deflated just as if you just bought it from the shop and about to fit it on the wheel and fill it with air. 

 

Also, how will / can the printer reproduce the exact thickness of every spoke ?  amazing.

 

Thank you for this fascinating WIP

 

Edited October 21, 2016 by sharknose156

[Roy vd M.]

Beautiful work and good to know that it's actually very simple 

 

I don't assume the spokes will be printed??

[NickD]

This will just be a short post (famous last words) to answer some of the comments above.

 

Spoke alignment and 3D printing - the main reason for building this model in the first place was to see how much of a faff it would be to model the spokes and what a wheel with spokes that are thick enough to be printed would look like - would the spokes be so overscale to look ridiculous. The jury is still out. (in fact I have not shown it to the jury because that's you guys). Back to modelling the spokes. While Blender is fantastic, it is not a CAD package so there are certain things it does not do well. Or it does them but they are difficult. One thing I always struggle with is keeping cylinders circular when you start to bend them. This exercise was an opportunity to try some things with a shape is fundamentally quite simple. Having drawn a perfect spoke I did not want to have to place 70 of them manually so it was also an opportunity to learn how to handle arrays of objects - more later.

 

Alignment - Very good question. There is a number factors affecting the answer. Firstly Blender is very precise even in 32 bit form. That does not mean that what the humble modeller produces will be accurate. As the screen shot shows, even with a line drawing , line thickness is a signficant issue. Also all drawings are wrong and all photos are massively distorted. Assuming one is happy with the resolution and alignment of the reference material, Alignment is in the hands of the modeller. The comparison between the measured and modelled wheel dependes on having a common frame of reference. Axle centreline is one. but the other is not obvious because the model wheel does not truely model the hub geometry and the tyres are very different. I chose to align tyre back face but tyre centrline or hub front face might have been better. As usual it needs to reflect the modeller's aims. In this case I am more interested in the lessons that the spokes and printing will teach me. I have not tried to hard to get the tyre geometry perfect.

 

Deflated or not - there are two possible answers here. To generate the screenshot I switched off the modifier that controlled the mesh resolution, leaving just the basic mesh which is crude and flat. With the modifier applied, it behaves like a rubber skin that flows within but does not necessarily exactly correspond with the mesh. The second answer is that I might not have the base mesh in the right place. Minor tweeks could easily make bulge bigger or smaller. I might come back to that later.

 

Again thanks for your continued interest and support

 

Nick

[sharknose156]

Thank you for these clarifications.

 

Crossing fingers really strong that the spokes will not come out over-scaled as you say above. 

[NickD]

OK now to spokes.

 

As I noted above there are 6 sets of spokes. Modelling each spoke indivdiually would take for ever. Fortunately Blender has a really neat modifier call an array modifier. It allows the modeller to generate multiple copies of one mesh in real time. It's use is a bit complicated. There are some good tutorials but you need to pay attention which appears not to be my forte! The key is not the array modifier on the object itself but something called an "empty" that it is tied too. Bear with me. I think the rationale is that because you are creating an array of objects you need a single item that defines the position, rotation and scaling of the ensemble. I took me a while to understand this, and, as a result generated some really wacky objects. Eventually I got it and it became much easier.

For each ring, the array needs to be set to the required number of spokes. The empty is then rotated by 360/number of spokes. Having set that up for an object that can contain geometry, in edit mode, one can start to create the spoke.

 

I made the spoke as thin as I dared - I think I have probably overdone it. I fear it should be 1mm minimum and this shows 0.5mm. We will see later - much later. It should be fixable without too much effort.

 

I thickened the spoke at the rim end to represent the nipple and rim bulge. It really simplifies the rim model and comes together when the rim is added.

 

The other end is the challenge. Bending tubes virtually is like real life. They usually wind up crimped and flattened. To avoid the problem Blender provide a modifier that called "curve" which can be used as a former for the mesh causing it to, well, curve. Simples - well almost.

 

That's it - keeping things planar is a bit of challenge sometimes but the geometry is really very simple.

 

 

[vontrips]

Please don't take this as a negative comment and I love your enthusiasm for the subject, but until I see some beautifully printed 0.5mm spokes, for me it's a no go!

 

I spent over 20 years working with CGI 3D modelling...that's why I now only model with my hands!

 

Keep it up and looking forward to the results.

[sharknose156]

am more optimistic than VT!   ( hec my father is 90 and he spends hours on his mac reading and booking trips on planes to far away destinations he really ends up going to..)

 

in any case fascinating stuff Nick.  These spokes look like the real deal.

 

Am really enjoying it and learning whatever i manage to understand. Your explaining is very clear. Thank you.

But the key seems to effectively enter in the logic of this Mr. "Blender" to understand how "The empty is then rotated by 360/number of spokes"... i like your choice of words... very sci-fi. 

 

lets get this wheels with 70 spokes ?

 

[vontrips]

Sam, I've spent too much of my life watching 'supercomputers' crash during rendering 'state of the art' - blah, blah, blah, motion graphics...usually at 3:00am (when even the seediest Soho eateries are shut), not to be a little cynical!

 

I'm also totally unimpressed with the majority of car design on the roads these days...I blame this largely in people that know the design software, but have no sense of design!

 

Plus, we're aiming for 0.3mm diameter spokes for 1/12th modelling. I have yet to see any 3D printer capable of rendering in such high resolution?

Edited October 22, 2016 by vontrips

[Roy vd M.]

I think we can all agree that metal spokes will look nicer than printed vinyl or plastic ones, without a doubt... at least until some super 3D-printer is available. Maybe in 10 years from now. 

 

That said, this is still a nice quest to find out whether it is possible to 3D-print a spoked wheel that looks good. Maybe not perfect, but good. I think many modelers would be extremely happy with such result. Don't forget that this forum features some of the most talented car modelers around this globe, and let's not name them this time so they won't have to deny it. These talented car modelers in the large scales raise the bar for the rest of us and it would almost seem that weaving your own wheel full of spokes were standard practice. In my view, it certainly is not. It's top notch modeling, performed by maybe 5% of all large scale car modelers. For the remaining 95% Nick's attempts could be very valuable. 

 

Maybe I see it wrong, but this is how I view this thread. 

[NickD]

Hi VT thanks for commenting. You raise several interesting and useful points.

 

0.3 spokes - not a chance. As I said above .5 is pushing it and as you will predict, if I have not completely wound you up, half mill would look quite meaty. Are they good enough.

 

Why bother - Couple of reasons: 1) I am sure some of our ancestors swore by flint rather than bronze because held a better edge and didn't have the health and safety issues yet the world still moved on. I also have many years experience of what works and what doesn't. Some seemed like a very good idea at the time and really wasn't. 3-D printing is at that stage for me. I don't like the materials much (usually because I can't afford the good ones). This was a proof of concept to see what did and didn't work. As Buzz would say "failure is an option" - probably likely even. 2) There seemed to be some interest in what could or couldn't be done with a tool like Blender. Blender will model at 0.3mm quite happily. When (probably if) printing catches up (dot matrix vs laser vs inkjet springs to mind - there are some smart people out there), The individual in a shed or garratt will absolutely be able to do this sort of thing to high levels of accuracy. The key enabler is whether it is easy or difficult to use the tool. Earlier comments suggest this is more of a concern. With this thread I was trying show that something as complex as a spoked wheel can be modelled quite quickly once the basics and one or two readily learned "helpers" have been mastered. Building the model I describe here took way less time than writing this thread. If this helps others gauge what can or can't be done or prompts a line of innovation that would not otherwise have happened I have succeeded in giving a little bit back.

 

What's good enough - As Roy suggests, some of us are much more easily satisfied than others. I am not saying anyone is right or wrong. These are just models. If we were serious we would be building 1:1 using the original tools and materials (some of the WWI replicas are amazing). Everything else is a balance, a compromise and often a trick of the eye. In the year or so I have been following threads here, I have been motivated by 2 things: learning new stuff (a lot of things) and sharing the journey other people such as yourself are on, that I will never attempt. It is this latter drive, that triggered me to have a go at this model, I wanted to see how far off the manual methods 3D-printing got us.

 

Finally thanks for the ongoing support from the Roy and Sharknose. I get the hurry-up. The next post will have 70 spokes. promise!

[NickD]

Having got one disc of 14 spokes, I just duplicated the object and modified spoke length, number and rotation until all 6 intersected the hub and rim. Going back to alignment and VTs comments on accuracy, I could have done a better job with regularity of spacing and sorting out the way the hub end of the spokes intersect with the hub (picture below). Because the spokes are overscale, and in a desire to have the spokes distinct rather than intersecting, the spoke ends are too far apart laterally.

 

Still here are a few photos. First is the wheel with all the spokes in. To demonstrate the flexibility of the approach I used the drawing tyre and rim and the model hub.

 

 

A couple of things to note: the spokes in the top picture are not radially distributed uniformly enough and should be tweaked. In the x-ray picture the tyre is had the smoothing switched off so the spokes are visible so it is too chunky.

[vontrips]

It's still a very good tutorial Nick!

[NickD]

VT - Thanks - Praise indeed.

[NickD]

JUst wanted to show a close up of the spokes. It highlights the sorts of things I would need to consider if I reoptimised the spokes.

 

[sharknose156]

the Mef wheel is built in four parts and the spokes do not intersect but align very well. Oliver also showed how each one of the four parts can be upgraded in spokes so as to reach the 70 spokes per wheel. Would that be a solution for Blender ? to build the wheel in four parts to avoid intersection of spokes and overlaps ? just wondering aloud.

[Jnkm13]

Hi Nick,

 

This will be 3D printed? I have tried Shapeways before, and there are noticeable striations / layers which are quite difficult to sand off. It would be near impossible to sand each 0.5mm spoke if the printed layers are noticeable. Perhaps a high spec printer can print them nicely.

 

Another point to consider is after layers of surfacer and paint, the spokes might be considerably thicker.

 

Please do not see the points above as negative feedback. I would love for this project to take place as the Mephistofele is a great kit that would look even better with these wheels.

 

Regards,

Jeremy

[NickD]

Thanks for the comment chaps,

 

Separate rings would work. It would be a trade between lateral separation and whether spokes intersecting is noticable.

 

3D printing - those seem lke reasonable comments to me. I've had those concerns myself. First step though is to combine all the above geometry into a single object and see what Shapeways make of it. At that point I think there are a number of options. Print the wheel as a single item (with various limitations described above); print rim tyre and hub separately ( not a great stepforward but it might simplify the drilling, improve rim and hub geometry), print spokes in one material and rim in a different one, photoetch the spokes (might be OK visually if very fine), plus probably numerous other ideas that have not occured to me or the rest of us yet.

 

Keep the ideas flowing.

Regards

 

Nick

[Roy vd M.]

Forget about photo etching 1/12th scale spokes, unless you are looking for square spokes. 

 

As a modeler I would be most interested in working with a combination of:

 

- 3D-printed tyre + rim (in one piece) with marks where the spoke holes should be;

- 3D-printed hub with marks where the spoke holes should be;

- 3D-printed jig fitting both of the above;

- Metal wire for making the spokes; and

- Wiring diagram and instructions.

 

Now if someone offered a package like that for 1/12-kits and 1/8-kits I bet they could earn some money. Something like Friuls, only for cars rather than tanks.

 

By the way, the 3D-printed specimen could be a master; copies could be made in resin. 

Edited October 24, 2016 by Roy vd M.