Need help with a surfboard loft

[karolisd]

Hi Fusion community,

I need some help with my surfboard design. I sketched some cross sections of a surfboard and then lofted those sections to create a solid body. I used a single point as the last loft section (nose) of the board. Here is the screenshot:

In Fusion it looked almost perfect, but when I cut this board with a CNC, i noticed the nose is very thin and sharp and has a slight corner on the deck. 

I want the nose thicker and more rounded. To address this issue, I thought I would create a small profile and use it as the last loft section instead of a single point. After that I should be able to use fillet to round it if needed. Sounds easy, right? Well......

Now my board has some strange corners on the bottom. It seems that the shape of those corners depends on the shape of the nose sketch, but no matter how I draw the nose, I can't get rid of the bottom corners. 

OK, so next thing I tried to do is go back to "loft to point" approach, but round the profile line around the point. Unfortunately, no go. Fusion complains about non-smooth neighbourhood of a point:

I would appreciate any ideas how to thicken the nose of the board.

[TrippyLighting]

You cannot create all of those shapes using a single approach that works for every shape/topology.

If you need varying curvature on the bottom you can make another profile curve at the front of the profile and loft.

A sweep was just the simplest solution creating a simpler surface. If you need more control, you need to add more inputs.

However, the more control you add, the more   you have to work. that is another reason that trying to create one universal model that can reflect anything is not a good idea.

[karolisd]

---

@TrippyLighting wrote:

You cannot create all of those shapes using a single approach that works for every shape/topology.

If you need varying curvature on the bottom you can make another profile curve at the front of the profile and loft.

A sweep was just the simplest solution creating a simpler surface. If you need more control, you need to add more inputs.

However, the more control you add, the more   you have to work. that is another reason that trying to create one universal model that can reflect anything is not a good idea.

---

I understand that very well. There is a zillion of features in surfboards that I am not even trying to model now, such as bottom channels, wings etc etc. The concave is a very basic feature that most of the boards have.  If it does have a concave, then the concave radius is never constant across the board, that's why the sweep solution didn't fit well.

Regarding the new model you uploaded. What can I say, it looks like it works. I don't know why lofting both sides of the board works better than lofting  one side, but the end result seems to be way better.

[karolisd]

I haven't integrated this approach into my main design yet, but after playing with it for quite a while I can confirm that the lofted shape appears to be accurate. The lofted bottom is sitting well below the intersection line along the length of the board according to my provided concave shape.

@TrippyLighting can you explain why this example works, while the previous one didn't? Also, if i apply the same sketch technique in the top cross sections, would it fix the shrinking loft problem there as well?

[TrippyLighting]

---

@karolisd wrote:

@TrippyLighting can you explain why this example works, while the previous one didn't? Also, if i apply the same sketch technique in the top cross sections, would it fix the shrinking loft problem there as well?

---

I don't have any insights into the geometric modeling kernel, but my guess is that not having to evaluate for a tangency constraint toward a helper surface makes it a lot easier for Fusion to create a surface.

Yes, I believe that would fix hat as well. I'll have to experiment with tis when I get home from work later tonight.

For me the real trick is to get the fade from sharp to full round correct.

[jeff_strater]

@karolisd and @TrippyLighting - I am out of the office this week, but I have been lightly watching this thread.  I'm responsible for the modeling kernel, and I hope to be able to dig into this thread next week and try to understand all the twists and turns that you have been through, to either explain some of the behavior or create bugs if there are problems or bad behaviors with the loft kernel code itself.  If anyone wants to try to summarize the main points of confusion that still remain, that might help us understand where to start.  Thank you.

[karolisd]

I applied this approach in my project. Long story short: 90% of the bottom shape problem is fixed. I wrote a longer post with screenshots, but my session here timed out while I was doing it so the post is gone.

I experimented with this for several hours and basically found 2 things that caused the big weird bump on the bottom:

 - Bottom rail curvature: Nose is curved to a higher degree than tail, so the loft gets distorted more by the bottom rail.

 - The width of loft profiles. It is possible to spread out this distortion just by making the concave profiles wider. The sides are going to get trimmed off anyway.

Final result (yellow = previous, pink = current)

The bottom is still slightly curved upwards, but it is possible to compensate for this by increasing the concave at the nose.

I will try to see if using a similar technique could improve the upper loft as well

[karolisd]

@jeff_strater It is difficult to say whether the problems we are discussing are bugs or not. These could be just features of the loft that I wasn't aware of.

On the other hand, there are things that I personally think are either bugs or major inconveniences:

1. Whenever I edit a loft and add a profile to it, Fusion starts treating it as a new body as if the previous loft was deleted and a new one created. This means that all subsequent operations are broken. On the timeline it looks like this:

If you are not an expert, it can take 5-10 minutes to find all broken references and fix them. I've done it probably 200 times today alone. I'm sure you can imagine my feelings about this time investment 🙂 And yes, with practice it no longer takes 10 minutes, but since it's the same loft as before, maybe it's possible to prevent it from happening?

2. "Rails can not be tangent to profiles". It is sometimes shown in situation when rails are clearly intersecting all profiles through intersection points. I still haven't figured out how to fix this error, my only approach is to hit "Undo" as many times as it takes me to the moment where the error wasn't shown. From there I can usually repeat all the changes and the error won't come up. 

[TrippyLighting]

The attached model is somewhat true to size as its based loosely on the initial model. It serves to show a workflow that creates a smooth transition form a sharp edge on the bottom to a full round edge along the length of the surf board.

 

[karolisd]

That is one complex model. My Fusion couldn't stand the pressure and crashed immediately 😄

I will look into it more carefully tomorrow, too tired today. Been torturing Fusion for a week now (or the other way around). 

[karolisd]

This example is very.... well, different. And more complex. From a practical point of view, at least 2 more inputs would be required for the top and bottom surfaces (1 each), but those can be added easily. The first impression is that the top surface is too "rigid", but I could be wrong. I have to try to get the required shape and see where that gets me.

It does take care of the sharp-round rail and the loft shrinking issue as well. The price is added complexity and a lot of places where things can break. 

I am very close to my goal with the previous workflow so I would like to finish that first and try to cut it.

[karolisd]

I played with the model a little bit more and there is one interesting thing I noticed about the loft: adding one more control point to each of the profiles increases the loft shrinking effect a lot. Here is the original model I started with:

You can see a tiny bump around the nose, but the rest of the board seems to be smooth. I then opened each sketch from the loft profiles and added one more control point at the side of each profile to make the board edge more "rounded". I projected the original profile shape to another sketch to have a good reference and copied the top surface of the spline as well as I could making it almost identical. 

To illustrate what I did, here is the original sketch:

And here is the new sketch:

If you put them one on top of the other, the shape of the top surface is almost identical and the only tiny difference is a more rounded corner at the bottom right. I did the same to all of the profile sketches.

And the result is that the new loft "shrinks" a lot more than the previous one.

I thought this would be an interesting thing to share.

The solution is probably to keep splitting the loft into smaller lofts. Either use separate top/bottom surfaces like @TrippyLighting did in his example, or use another way to separate the rail loft from the top surface loft. I'll keep playing.

[TrippyLighting]

The decaying surface quality of lofts with too many loft profiles can be observed with the bare eye, not even using any of the analysis tools. It will also be clearly observable in a machined physical artifact.