Porcupine Plots

Hi all....
one silly question for you.... but extremely hard for me to get by :

What the HELL are porcupine plots?? How do you understand them?? all i see is a 1000 blue lines shooting off from wherever they want!

please help me understand

 

Hull surface model or flow field?

 

well... i am a first time user of Autoship, and i am only at the joining curves and making surfaces phase... so i think the answer there would be "model field".

but, as future thinking for when i will encounter them, would you be so kind in explaining the flow field porcupine plots also?!

Thank you very, very, very much

 

getting curious, could it be the fairness tool or curvature plot of intersecting lines as in freeship?

 

What you are talking about are the "normal" vectors for the surface; i.e. the Z axis of a X-Y plane tangent to the surface at that pont (or it could be the normal to the surface of the mesh face is the surface has been descritized, all the same thing).

Here is how to look at it

Draw a RIGHT HAND X-Y corordinate system (X to the right, Y up) on a sheet of paper.
Place it axis system up on the table.
Lay the side of your RIGHT hand down on the X axis so your little finger is aligned with the X axis and your thumb is sticking straight up.
Curl your fingertips in the direction of the Y axis.
Your thumb now points in the direction of the normal of the surface.
Now pick the paper up and curl it, rotate it, swap sides, etc, and do it again. Each time your thumb will point in a direction that is perpendicular to the surface at that point.
Why is this important.
Draw another set of axies on a 2nd sheet of paper.
Tape the two togeter at the ends so that the X-Y axies are on opposite side.
Now find the normals...notice that they are on opposite sides of the paper.

This is what the porcupine plots are for. In order to get a correct mesh for some of the advanced features the normals of surfaces "stiched" together must all point in the "proper" direction. It is possible to stich together surfaves with opposite normals, but then things can go squirrly and mesh surfaces can look like Kline bottles.

In a broad way, Yipster is correct. Depending on the program, normals can be made to represent features (such as instantious rate of curvature, similiar to the polyline node tangent handles) that can be used to check fairing and even change curvature at a point, but mostly they are a visulization tool of what the surface looks like and to ensure all the surfaces are stiched together properly.

In flow fields, a porcupine plot shows the direction and magnitude of the flow at the tail of the arrow.

The ability to visualize in 3-D is necessary for a good hydrodynamicist. Fancy plots and lots of colors cannot be a substitute though some think they are. I has been my experience that it is an intrensic ability that people either possess or not. I have found that it cannot be taught.

 

THANK YOU VERY MUCH...

in your opinion, what can you tell me about these.... they are 3 different section of control points of my starboard side.

also.... could you PLEASE tell me what "Mean Curvature" and "Gaussian Curvature" show.

cheers!!

 

your answer made complete sense after i posted my reply.
truly thank you.
so, come to think about it now... those pics will do you NO good, as you need a 3d view, right!?

but what is the meaning of the difference in length of the lines?

well, just in case i get lucky, i'll leave them on!

 

The porcupine lines in Autoship represent how tight the curve is at each point along its length. The main purpose is to highlight discontinuities (unfairness). The outer ends of the porcupine plot should also form a fair curve if the curve is fair, and will make sudden transitions if it's not. It's used not only for lines but also for the control lines that define surfaces, and is one of the most useful and oft used features of Autoship.

You can adjust the length of the porcupine lines with a little up/down thingie beside the button that you turned them on with. Otherwise you have them shooting off all over the screen and they make no sense.

The most common mistake when starting to use Autoship is to use too many control points to make a surface. At uni some of my classmates used up to thirty rows and columns to define a hull, and in the end they had me delete most of them to fair it. Start with a 5 x 4 or a 5 x 5 surface and maintain fairness right through the process. You can tack extra rows or columns on later to refine the shape.

The gaussian etc. curvature renders are just a visual aid to locating unfairness in a surface. You'll find that it tends to turn up around knuckles and suchlike and isn't always apparent unless you use the curvature renders to check.

 

Aramas is correct, what you show are instant curvature lines of a line, not a porcupine plot of a surface. The line lengths are proportional to the rate of curve (i.e instant center) at the origin of the curve radius line (i.e on a straight section they would be infinite and on a circle they would all be the same length and growing larger as the circle grew larger. However the transitions from positive to negative curvature look funky...but that may be a plot programing artifact.

 

Actually it's the other way around in Autoship. The tighter the curve the longer the porcupine line.

 

thx a lot guys!

 

That's funny, because thats not what the posted plots show.

And why show 1/the instant center when just calculating the instant center is more computationally efficient.

 

I think the posted plots show an isometric view or something, using curves embedded on the surface. They're more or less never used since editing them moves the other control points as well and results in a godsawful mess. The porcupine plot function needs to be turned way down in the posted screens before it would make much sense, and it's completely useless in isometric and camera views anyway. They're a complete dogs breakfast as they're shown. I didn't actually look at them before. I've just been using Autoship/yacht since 1991, right back to their original DOS Autoyacht, so I have a vague idea of how it all works.

It makes perfect sense to me. Perhaps it's someone's brain that could be more 'computationally efficient'? Mine certainly could.

Here's a few screen captures from Autoship. The first shows a curve with the porcupine plots turned on and set to a sensible length, although the website compression seems to have lost bits of the plot - just click on it again after you open it and they'll smooth out. The other two are just renders of a model that was faired entirely with porcupine plots, and perhaps they might imply that Autoship's porcupine plots are quite computationally efficient enough for their intended purpose

 

Our office gave up on Autoship for lines back in '92. CAM errors were too great, we could never get the shape we wanted, especially solid intersections and ploting, and performance was poor compared to other options we had. The mechanicals and trons still use Acad, and sometimes we do something for them, but mostly we work in Solidworks and Rhino.

 

So why does this image show frequent lines initially..... and less lines (on the some curvature) less frequently further on? what does it mean??