Why is "project to surface" widening the profile?

[lichtzeichenanlage]

ello,

as a follow up to this thread I've created an new thread, because I have a new question and I don't know if there is a option to optimize the result. 

Regardless if I'm using "Project to surface" or I'm sweeping the profile and "Project intersect" it to the new sketch, the result is wider than the source. Why is it like it is and is there a way to ovoid this? A simple project is much closer to the width (but isn't usable for me)

If you want to play with the project or want to see the different projections you can do it here.

[davebYYPCU]

In the previous file, your sweep was inclined in one direction, first figure in this diagram, and geometrically accurate to the canvass.

Projections are perpendicular to the source plane.

Project to surface can use a non perpendicular vector.

The second diagram is likely the result from a projection to another sketch (chine) which may not be parallel to the first sketch, but can also be distorted by the camera angle, neither can be detected from the diagram.

[lichtzeichenanlage]

---

@davebYYPCU wrote:

In the previous file, your sweep was inclined in one direction, first figure in this diagram, and geometrically accurate to the canvass.

---

You're right that the sweep looked accurate on the first view. And I shift the profile along an vector along the x axis for both (sweep and project to surface. But if I intersect the profiles by a 90° plane the points from the top sketch and the projected sketch are not vertical to each other. They're of by about 1°. Using the sweep trick or "project to surface" show the same behavior. So my question is still: Why results the projection or the sweep in an wider profile. 

[beresfordromeo]

Are you projecting to a surface using along vector or closest point?

[laughingcreek]

is your "project to surface" set to " closest point" or "along vector".

Unless your projecting to a surface that is perpendicular to the sketch plane, "closet point" projection will always result in a curve that is skewed in some way compared to the original.

[lichtzeichenanlage]

Along a vector

[beresfordromeo]

@lichtzeichenanlage

Thanks for the reply. I can't see any geometry in the A360 link you provided above. Should there be something in there?

[lichtzeichenanlage]

@beresfordromeo: There is something in the project. Sketches, sketches and more sketches 😉

I guess you have to download the project and open locally. 

[laughingcreek]

I just saw your file, missed it the first time.

The projected line you currently have lines up perfectly with the profile when viewed normal to plane "plane deck - 1"

The line from the sweep lines up perfectly when viewed normal to the prow line.  I also did a  "project to surface" (using "plane deck - 1" as the surface to project to) using the prow line as the vector.  It also both line up perfectly to the normal view (and with the current curve from the sweep)

Theses are all results I would expect.

[lichtzeichenanlage]

@laughingcreek: You're pretty much summarizing my thoughts from the last thread. But ... Why is this happening?

Those projected points are create by "Project/Intersect" The top ones are the source, the middle ones are the from the projected (to surface) and the bottom one is from the side view.

[beresfordromeo]

Thanks. Not sure what happened there but I have the file.

Can you please record a quick screencast of how you are projecting. I can only assume you are projecting from a sketch on Plane Deck View to a sketch on Plane Deck - 1.

These two planes are not parallel so the projection result will be dependant on which line you select as the vector.

[beresfordromeo]

Are the planes of the source and the target in this example parallel? If the projection vector is perpendicular to both planes then this will not happen.

[lichtzeichenanlage]

@beresfordromeo: Planes are not parallel and vector is not perpendicular.

[lichtzeichenanlage]

@beresfordromeo: Your assumptions are right. Here is the screencast:

[beresfordromeo]

Ok, well that is cool. 

As @laughingcreek has pointed out this is normal behaviour. So I guess the question now is what do you want to happen?

In the screencast below you will see what happens when I use a different vector. This is entirely normal behaviour but there could be a workaround to achieve what you want to achieve.

[lichtzeichenanlage]

@beresfordromeo:

The green line defines the profile of the upper part of the hull. I've colored the upper side face of the hull red. The topological map shows to different views of the red area. As you can see, the upper side of the side hull is 100% (???) vertical and the top view (green) does only differ in the very right part (front). The yellow part could be ignored, because I can adjust this later.

Edit: Yellow color edit

[laughingcreek]

To achieve what your trying to do, with this approach, your projection vector is going to have to be along the Y axis.  And then you add to the sketch at the nose, to get it to stick out that extra bit.

[beresfordromeo · Accepted solution]

Ok I see now what you are talking about.

There is no clear answer to this problem. Obviously you can do what @laughingcreek has suggested above but there is a real question over the design intent here.

The problem is that the image that you are using is only representative of the actual shape of that hull. That is to say that if the hull profiles were the exact shape of the top profile projected in a straight line to the non parallel plane beneath it, then projecting to a surface using closest point projection is the actual shape of the hull on that plane because that is the whole purpose of closest point projection, to project along a vector of the shortest distance to the plane below. Unlike an along vector projection which is constant throughout, a closest point projection uses multiple vectors  adjusting 'direction' along the surface topography which is why these types of projections 'change shape'.

This is a difficult thing to explain without the use of diagrams and hand puppets but if you want to know more I am happy to try to explain, I just don't want to bore you or anyone else with the details.

In short drawings like the one above often have lines removed so as not to clutter them up.

I guess you can think of it like this. If you look at the top (plan) view of that model then and zoom out far enough that you so that you can see the whole of the boat then the lines along the side of the hull, even on high resolution monitors will appear on the most part to be one line. You would need a very high DPI printer or a very big piece of paper to print it in such a way as to see the distinction, hence why a lot of plans remove lines which are non essential.

If the lower profile follows the upper exactly except at the nose it is a different shape, it has to be to make up for the non parallel nature of the two planes and therefore projecting it is not really possible because it would not be a projection in any meaningful sense of the word, ie you may as well just sketch the second profile.

It is not much of a solution but the good news is you have found a use for closest point projection. I would use it in this model because it is the one which most likely represents the actual design intent of the vessel you are modelling.

I hope this helps please feel free to ask if it doesn't make sense.

[davebYYPCU]

And the sweep procedure was right, with slight extension to the transom end.

cut or Project intersect with the lower plane,

Done. (a few days ago.)

[lichtzeichenanlage]

@davebYYPCU: As mentioned in my opening thread "Project to surface" and "sweep procedure" are producing the same (im my use case wrong) result. You can check this in this image, too:

For both solutions you can find a view port angle where the lines do match. That was my mistake in the other thread. But the question in the other thread was answered, so I closed it. For me this thread is: Not done (right now)  😉