# Face Normals Get Flipped When Collapsing Edges

DELFTship forum Hull modeling Face Normals Get Flipped When Collapsing Edges

• Author
Posts
• #39352
Claudio Boezio
Participant

At times when I collapse edges, surrounding faces get the normals flipped. Sometimes it’s just the surrounding ones, sometimes half of the faces in the model (hundreds) get flipped. Before I collapse the edges all normals are correct and pointing outside. Is this normal and intended behaviour for DELFTship or am I doing something wrong?

Thank you for any hints.

• #39353
Icare
Participant

When you change something in a perfect model (punching a hole, collapsing polygons’ edges…), the software can be puzzled between inner and outer surfaces.
When you click on the “CHECK MODEL” button, it will reconsider the whole model, find the leaking points and flip the surfaces when they look odd in its “eyes”. When the volume is closed (when there’s a deck and no leaking points),the operation works well. But when you have too many leaking points, too many “inside out” surfaces, an open volume (no deck), the software can give you strange results.
Generally, you just have to click on the “CHECK MODEL” button again, and then you’ll get the expected result.

You can get many inside out flipped surfaces when you draw polygons one by one by selecting their corners. In the manual it’s said you have to select the points clockwise or counter-clockwise (depending of the camera position: inboard or outboard). Just remember this when selecting the points! ðŸ˜‰

Mind you: I drew a Klein’s bottle and the software could understand how to flip the surfaces! B)

The Klein’s bottle is a mathematical curiosity: a non-orientable surface. With such a bottle, there’s only one surface which is inner and outer at once. (Wiki link: https://en.wikipedia.org/wiki/Klein_bottle). It’s a little like the Moebius’ strip (Wiki link: https://en.wikipedia.org/wiki/M%C3%B6bius_strip) applied to the volumes…

• #39358
Claudio Boezio
Participant

Thank you Icare and sorry for the late response.

My model is completely enclosed (decked). If the normals point all to the water and I execute the command, the normals will get messed up. I’m not sure whether it has something to do with my model being a stabilized monohull and there are some limitations for that in the software. What puzzles me though, is that when I collapse edges to simplify/edit the model, the flipping of normals is not limited to surrounding surfaces. That wouldn’t be a big deal. Instead great portions of the model get flipped and reversing hundreds of them back again is very time consuming. It’s like if the command would be executed automatically, which in my case is something I don’t want becuase it messes up the normals. I’m just wondering if this is intended behaviour.

About your point: When I insert new faces, I usually can’t always predict in which direction I have to select the corner points in order to get the desired face normal. But that’s no big deal, because I check and correct the new faces right after creating them and they are usually not so many. The sudden change of many already existing and well-oriented faces when collapsing an edge is what worries me.

I made it a habit to check the model by setting the draft in the project settings so high that the whole model gets submerged. By assigning the underwater volume a unique color, I can visually check whether all surface normals are correct, even those above the waterline. This tells me if the calculated displacement volume and other hydrostatic properties are trustworthy. Unfortuntely, not always when a surface has the wrong orientation, the displacement will amount to zero. It can indeed happen that the calculations will just be off slightly, (e.g. being a few cubic meters short) depending on size and location of the ill-oriented face(s). Without means to counter-check the displacement volume, one might be relying on wrong hydrostatic properties without noticing it, due to an only marginal difference. That’s why after I do major edits, I check the model like this just to be sure that all normals are pointing to the water/outwards.

Interesting about the bottle, I forgot about that one. I’m sure that in most cases, DS will orient the faces correctly.

• #39359
Marven
Keymaster

In most cases when the software orients the normals incorrectly this is caused by incorrect geometry. For example there could be edges in your model with more than 2 faces connected to it (in which case the edges turn either purple or green), or even a double face (2 faces using the same points).
If this is not the case please send us your model so we can have a look at it, as perhaps we might need to improve the algorithm which checks for the orientation of face normals.

• #39368
Claudio Boezio
Participant

Thank you marven for your answer! As a matter of fact, my model indeed has an area where there are four faces attached to the same edge, see attached sketch of transverse section. Although I understand why that might be an “incorrect” geometry for DELFTship, shipbuilding-wise this is a perfectly legal configuration. Do you think this could be the cause? Thanks.