[Plugin] Perpendicular Face Tools (UPDATED 26-03-09)

linea · 12 Mar 2009, 08:06

Thanks Chris.

erikB · 12 Mar 2009, 09:15

Chris,
Thanks for the usefull ruby !
erikB

Chris Fullmer · 13 Mar 2009, 03:00

Your response hasn't gone un-noticed Tig. I thank you immensely for it. I'm still working on implementing it I think there is a small flaw in that I don't point C either. I think I know how to get it, so thats good. In fact, it is essentially what you provided with the "cosine rule". Thanks for that! That is currently what I'm working at. I had one method, but it was rather convoluted and started to fall apart for some reason, so I think it must not have been sound. This new method feels much more mathematically correct. I'll let you know how it goes. Hopefully within a few hours. Thanks sooo much for your help and ideas,

Chris

TIG · 13 Mar 2009, 09:23

To rotate the square you have more than enough info. You don't actually need point-D.

You want to rotate it about the centre-point [let's call it point-O] - so you know that too.
You know point-A and point-C as you have them initially or have just calculated them already.
Now find the angle between the two vectors O->A and O->C.***
Select all of the square's edges (face will go with them?) and make a rotation-transformation about point-O by that angle: using the normal to the square's face as the axis of rotation...

Points A and C, and points B and D will then be coincident...

"new vector"
v = [x,y,z]
vector1 = Geom::Vector3d.new(v)

Where v [x,y,z] is obtained from the sum of the two points v = (pointO - pointA) etc

you might also want to ".normalize" the vector(s)...

Find vector2 (O->C) the same way...

"angle_between vectors"

The angle_between method is used to compute the angle (in radians) between this vector and another vector.

Syntax

angle = vector1.angle_between(vector2)

Arguments

vector2 - a Vector3d object

Return Value

angle - an angle (in radians)

Example

vector1 = Geom::Vector3d.new(x,y,z)
vector2 = Geom::Vector3d.new(xx,yy,zz)

angle = vector1.angle_between(vector2)

###########

A bit more complicated to my mind is doing it with the 'cosine rule',

here it is though - you can easily find the length 'AC'

you have the other two points and can use 'AC = pointA.distance(pointC)' method for that.

You also know the circle's Radius - or can easily get it from the square's side dim.

Angle = Math::acos((RadiusRadius + RadiusRadius - ACAC) / (2Radius))

Note: if Angle > 90 degrees (pi/2) then it's negative...

.

MALAISE · 14 Mar 2009, 15:03

Hi Chris

Small feedback : using customer shape crashes SU ( under Windows XP 3)without any message.

MALAISE

Chris Fullmer · 14 Mar 2009, 22:36

@malaise said:

Hi Chris

Small feedback : using customer shape crashes SU ( under Windows XP 3)without any message.

MALAISE

Oh dear, thats not good. Can anyone else confirm? I don't have xp sp3 anywhere to test it on. Is it possible that it is a conflict with another script? That doesn't seem likely to me since I think I've written my code so that can not happen.

I do have xp sp2 I can test it on. Perhaps if I can duplicate it there, I can narrow down the problem. Thanks for the notice,

Chris

Dave R · 14 Mar 2009, 22:41

Chris, I've done a couple of experiments with it. No crashes yet. I've got XP Pro SP3.

Dave

Ecuadorian · 15 Mar 2009, 07:35

An artist who is also a coder?
That's Chris Fullmer!

Thank you very much for taking the time to code this; with a couple of tweaks, this will be perfect.

MALAISE · 15 Mar 2009, 07:52

OK, let me see which plugin may be in conflict

MALAISE

MALAISE · 15 Mar 2009, 07:54

Hi ALL

Filtering the last downloaded plugins, I've found a conflict with Jim's cd_power_n.rb ( mover2.rb) Has someone experienced that? several times tested.

( SU V7, Windows XP SP3 )

MALAISE

Chris Fullmer · 15 Mar 2009, 08:24

Very Interesting. I have that plugin installed with no conflicts. I'll dig around tomorrow and see what I can find. Sorry for the inconvenience,

Chris

MALAISE · 15 Mar 2009, 09:13

Thanks for playing attention, no matter for inconvenience, don't worry
MALAISE

Jim · 15 Mar 2009, 13:18

@malaise said:

Hi ALL

Filtering the last downloaded plugins, I've found a conflict with Jim's cd_power_n.rb ( mover2.rb) Has someone experienced that? several times tested.

( SU V7, Windows XP SP3 )

MALAISE

Mover2 does not use a file or method named cd_power_n.rb.

Chris Fullmer · 15 Mar 2009, 15:55

I know, its cad_fathers's SCF powertools toolbar plugin that wrapped mover2 into a file called cd_power_n.rb. If you open that file, its just your script. So I'm guessing it has something to do with how its being loaded from his main plugin loader maybe? I don't know. I know I've got my whole script wrapped in a class so I thought I should not have conflicts. But perhaps there is more to it than that.

Chris

TIG · 16 Mar 2009, 09:37

I'm not getting bug-splats with custom-face, BUT it does inconveniently make a group of the selected face and not explode it at the end. It's easily fixed by adding a line of code after line #299, to explode the temporary group...

@group.explode

How's the orienting face rotation on end of line going ?
.

MALAISE · 16 Mar 2009, 13:51

I've tried this improvement, the group is still present, not exploded.

MALAISE

Chris Fullmer · 16 Mar 2009, 15:54

Hey Tig, I probably could explode the group without any conflicts. I was thinking I didn't want the added face to interefere with the model, but its probably not an issue. Or maybe I explode it by default and have a ctrl key modifier that leaves it grouped. I've been wanting to figure out modifier keys.

As for the rotation thing, I still have not got it worked out I'm having a hard time pinpointing point C from my diagram above. I got close and it sort of worked, but it fell apart if the face was too slanted or something and I got so frustrated I have not even looked to figure out why. I'm sure its a million times easier than I'm making it. Since I can't find "C" yet, I intersected a flat plane with the face plane to get an intersection vector. That vector is the vector that I want my top(or bottom) edge to exist on. So I then did an angle_between the 2 vectors (the found one and the existing edge one) to get the angle that I want to rotate the face. That seemed to work in testing, but did not work well once I put it in the script. Sorry to drag this out. I'll probably put some more time into it today since I got a few other scripts updated last night.

@Malaise, I think maybe that added the ungroup function to the custom face tool only, and maybe you tested it on one of the other tools. It would need to be added for each tool. I'll look into it, and maybe even get it implemented sometime this century

Thanks everyone, thanks TIG!

Chris

TIG · 17 Mar 2009, 09:26

It's the original face that needs to be un-grouped. I agree that the face put on the end of the line is best left grouped to allow you to move its origin or rotate it about the line etc - you can explode it before doing a followme etc...

Here's my thinking on the rotation conundrum...

Let's do the square for now - seems to me that first you make a group containing a 'flat' square with its centre set at the origin and then move it to the select vertex at the end of the edge, transform rotating it so its face is perpendicular to the picked-vertex's edge... So far so good. Now we check the geometry of the square - it only has four vertices, so find the two with the lowest Z-value***. These are the vertices of the bottom-most edge. Now find the mid-point between them - we use this to work out the vector from the 'centre' (picked-vertex's location) to this mid-point and the angle between that and a 'vertical' vector in the same plane as the face: that 'angle' is the angle that we'll use later on to rotate everything so that it is 'orthogonal' to the XYZ axes.

Let's call the centre 'c' and two lowest points 'p1' and 'p2' - these are known - and the mid-point we'll call 'm' - that we need to get...
To find a point mid-way between two others use...

 m = Geom;;Point3d.linear_combination(0.5,p1,0.5,p2)

We can then get the vector from 'c' to 'm', let's call it 'cm' - simply subtract the point values...

 cm = c - m

Now to find the 'vertical vector' (let's call it 'vv') - first find a point ('t') 1 unit away from 'c' above the surface of the face by the edge's [positive] 'vector' and then project it down till it hits the face at 'v' - the line between 'c' and 'v' is the 'vertical vector' against which we'll rotate the face...

 t = c.offset(vector,1)
 ang = (t-c).angle_between(t-[0,0,-1]) # i.e. the tilt on face normal
 hyp = 1/Math;;cos(ang) # i.e. the drop from t to face plane
 h = [t.x,t.y,(t.z)-hyp] # i.e. the point t projected onto the face's plane
 vv = c-h # the vertical vector

Now get it's angle to the 'mid-point vector'

 angle = vv.angle_between(cm)

You then need to work on the face of the square and transform/rotate the whole thing around the picked-vertex's edge by the 'angle'... Done.

***Note: IF the Z of all of the square's vertices is the same [simple test to look at the four] then it's 'flat' so then we'll find the lowest two Ys and rotate it to be orthogonal about the Z axis instead. Also note that in this 'flat' case the 'vertical' vector needs to be (0,-1,0) to match the Y...

For the custom face do something similar but group the face, copy it, explode the original back to where it was and then use the copy's bounding box centre as its 'origin', move and transform it to the vertex, use the new location's bounds to get an enclosing rectangle to get minimum Zs etc, all as above etc etc...

I haven't tried this code, but it should at least form the basis of something that''ll work...

.

Chris Fullmer · 17 Mar 2009, 18:44

Thanks for walking me through this TIG. Hopefully I'll get a chance to try to get this implemented tonight or tomorrow. Lots of work-work and school-work today

Chris

cosycat · 17 Mar 2009, 21:19

looking at this, I wondered how difficult it would be to create a script that 'blended' between two diffrent faces at either end of a line.