#Bend and Smoothing?
14 messages in this thread
John:
You're right. Sometimes a 2d morph is the only reasonable solution. I have a
fairly inexpensive packgage by Gryphon that is really quite impressive.
Somehow, though, I wanted to conquer my "eagle" morph even if I had to pull out
what little hair I have to accomplish it. I could get the eagle to essentially
flap the wings using geometry and a morph but couldn't get the smoothing on the
bent object.
I figure anyone who can create a map for a "soccer ball" could whip out the
answer I wanted to hear. <G> (That one really was a major cool project].
Jerry
>> I figure anyone who can create a map for a "soccer ball" could whip out the
answer I wanted to hear… <<
Then I kinda curious as to why you didn't post your question to him in the
first place… I'm not sure the analogy is the same as the bends in a soccer
ball aren't as severe as what you're trying to achieve but… then again, who
knows… by increasing the number of morph targets, you can lessen the angle
between morphs and by changing the material morph smooth degree value you
should be able to achieve the desired result…
Good luck,
John
John:
My most humble appology. For 2 years now as I check forum messages I had you
associated with JKJ who did the "most excellent" soccer ball. My project had no
relationship with the soccer ball whatever, I only mentioned it as a
complement, that anyone who could do that soccer ball project could easily
figure out my problem. From now on I'll assume that there are more than 1
person in this world named "John."
BTW. I tried more morph targets but as the bend became progressively greater, I
began to get the stray faces as well as the smoothing problem.
Jerry
P.S. This is the Jerry B. from San Diego, friend of the "Big A." <g>
Jerry,
Compliment most graciously accepted… JKJ is a great guy on top of everything
else… and yes you are quite correct, there are quite a few Johns here… I've
tried all kinds of monikers but in the end I'm John, so that's were it is. 🙂
And BTW no apologies necessary… except maybe to JKJ but I'm sure he's got a
good sense of humor about it… though I imagine he does appreciate getting due
credit for the soccer ball… <g> I really should down load it, now that I know
who the creator was…
>> I tried more morph targets but as the bend became progressively greater I
began to get the stray faces…. <<
Well go ahead and upload it if you like, no promises, but I'll see what I can
do or try to provide an alternate solution…
(Glad we got that straightened out. <BG>)
Actually if apologies are in order, I'm sure I owe you one for a meeting along
time ago, I never made… maybe someday I'll get a chance to make it up…
regards,
John
John:
Actually, I should have known there was more than 1 John, now that you
mentioned it. My DAD's name is John. <g>
I'll send the problem child to you. Don't worry about getting back in a hurry,
cause I know you have lots of projects of your own to conquer. Hope to meet you
at Siggraph.
Jerry
Jerry,
I'll check it out as time permits, at the latest next weekend, see what can be
done with it…
Later,
John
O.k. Jerry,
Comeing back at ya… 180 degree curvature… now that you've seen how to do
it… see if you can figure out what you need to do to smooth it… <g>
John
John:
O.k….O.K. ….You "almost" got it. Your bird certainly bent better than mine.
If you look close though there are some faces that actually moved clean though
the object to the backside.
I found that if I tesselated your most cool flat bird, by selecting the long
skinny faces for tessalation, that the bend was "near" perfecto!!
I further, selected the front flat face before the bend, which made it easy to
see the offending faces when bent. Now I could,… smooth, faces, selected…
to get the smooth front with yet still sharp edges in the outline.
I have one last question. In your "bird" that you split in two which gave
additional veritcies and faces in the center, I lofted it, moved element to get
the left and right to match, selected the union verticies, and welded for one
object. It look near identical to your object but still didn't bend as well.
WHAT IS YOUR SECRET? <BG>
Jerry
Jerry,
>> If you look close though there are some faces that actually moved clean
through the object to the backside… <<
Those triangles that are coming through the backside are dealing with the same
problem that you encountered in the first place… no place to bend… and
having given you the solution to the first part I thought it would be apparent
that you would need to do the same with the lower half, especially if you tried
smoothing it…
But as it turned out you went through the lofter and tried tessalating it…
which is not the answer… Tesselation has a very limited use and in this case
has no bearing on what you're trying to do…
If you quarter the bird again, and you may even have to seperate some tail
feathers to get it right, that will solve not only the "bend" problem but also
the "smoothing" aspect of it… 🙂
Polygonal Bird Warping I think rates at least a 400 level course… don't you?
<g>
>> WHAT IS YOUR SECRET? <<
Its not a secret really, just that most people don't use it as they get
accustom to using the lofter….
Instead of bringing the shape into the lofter, I brought it directly into 3D
build where I welded the vertices of the two halves. Then with selected faces
on, I used the FACE/ Extrude command to give the plane depth… no tesselation
required… Its how the polygons are facing relative to the angle of the bend
that matters. The more sections you have on which to bend the smoother the bend
will be… (Which goes back to David's point about sections…) The cut down
the middle gives the bird a set of vertices on which to bend… the lower half
of the bird (the tail feathers) do not have their respective vertices joined to
anything so they don't actually push through the back but rather remain fixed
or rigid relative to the rest of the model… which is why they need to be
divided as well…
I think that makes sense but if not just split the tail feathers so that there
is a segement on which to bend and the rest should fall into place for you…
Either way its good practice right? <g>
BTW I look forward to meeting you at Siggraph as well. 🙂
John
John:
What an absolute novel idea!!! How do you come up with stuff like that???
Thanks much for the detailed explanation. Ideas like that can be adapted to a
multiude of problems.
I figure if I ask enough questions and try to solve enough problems, even if I
never become an animatior I'm going to know the answers. <BG>
Looking forward to shaking your hand at Siggraph.
Jerry
Thanks Jerry,
>> What an absolute novel idea!!! <<
I think if you were to have to construct a model at the vertice level you would
choose a similiar approach, and that's my frame of reference… Maybe that is
novel, I really don't know. Most good modelers keep the fundementals of
building meshes in mind. The problem is that as we get more advanced tools the
tendency is to rely on them too heavily without an appreciation for the basic
concepts behind model construction. There are pros and cons to that but if I'm
here to help, then as part of that responsibility I need to stress the
fundementals… the other approach is no less valid but it has its own set of
complications which adhereing to basic principles can eliminate… if your
boolean doesn't work in a morph for instance… where do you go next…
So out of curiousity how did you plan to morph the boolean version to the flat
version… and will you get the same result? It would seem that the edges on
the boolean version wouldn't line up the same but then that would depend on how
you shaped your block…
>> I look forward to shaking your hand… <<
No special reason I trust, other than a "howdy do". <G>
John
John:
<<So out of curiousity how did you plan to morph the boolean version to the
flat version… and will you get the same result? It would seem that the edges
on the boolean version wouldn't line up the same but then that would depend on
how you shaped your block…>>
Basically, the project I sent you had the morph in the keyframer and seemed to
work very nicely. The boolean became the original and I simply copied it in the
editor, used bend to 180 degrees. Then element rotate and move untill it was
aligned with the original then morphed. This gave the alignment I wanted. [I
got that trick from someone on the forum as well]. I wanted the edges to be
sharp and only the front face to be smooth. I believe I selected the front
faces before the bend and then smoothed the faces as a selection.
Did you look at the keyframer to see the results? Even though there was only
one morph target I felt it did the job I wanted. Also, you could easily bend
along the opposite axis. The object ended up with about 3000 faces but I didn't
feel was excessive. I could probably have used a smaller grid.
I'm not sure what you mean about not being able to use the boolean for an
instance object? I didn't need to make one in this case but I don't know why it
wouldn't work. Of course there's still a FEW things I don't know…. : ]
Jerry
I can't believe it!!!!!!. I finally have the answer. Check out your E-mail.
Solution to getting consistent faces with perfect bend…. Boolean UNION gave
perfect results. I blew it with a boolean subtraction which left the odd narrow
unbendable faces. Finally retried the UNION which gave a very smmmmmmoooth
bend.
Thanks for all your assistance. This will definitely be one I'll remember. <BG>
Jerry
>> Solution to getting consistent faces with perfect bend… <<
Whatever works for you… I think its the principle behind it that has been
missing all along and that is you can't bend a triangle… so you have to
create "bend" points… the rest is overkill in terms of efficient modeling and
effort…
I will check out my e-mail as I'm curious how many faces you ended up with.
Later,
John