Organic Modelling Part.II -- Subdivision Surfaces Features

The interface, modeling, 3d editing tools, import/export, feature requests, etc

Moderators: jesterKing, stiv

philovivero
Posts: 0
Joined: Tue Jul 06, 2004 10:57 pm

Remove Doubles doesn't work. Still want feature!

Post by philovivero »

Download the .blend.gz, unzip it, and try it. Remove the doubles. Go ahead. It won't work.

I don't know why it won't work, honestly. I thought maybe it's because I'd end up with six-gons or five-gons, but no matter how I play the algorithms over in my head (for removing doubles) I think it would be trivial to end up with tris and quads.

Also, there is some extremely intricate geometry with vertexes crossing the symmetry point in a number of places (this was to allow for folds and creases like actually happen on the human body) -- in those places the removed vertexes cause the normals to be impossible to calculate correctly, and the mesh becomes wrong. (Again, try it -- remove the doubles, go out of/into/out of edit mode and in object mode look at the back and chest area).

Remove doubles won't save me any time here, and certainly wouldn't save me much time if I was trying to sew a hand onto an arm or a head onto a neck.

Others in other threads have asked for a "fillet" option which is explained as being when two meshes get close to each other (or two sets of vertexes,whatever) -- a new set of vertexes is calculated and joined between them such that the implicit surface is as close to the original shape as possible. That would be a KILLER feature.

--
philovivero

slikdigit
Posts: 133
Joined: Wed Oct 16, 2002 3:52 am
Location: Northampton, MA (US)

Post by slikdigit »

hey downloaded the blend - I'll have some pointers later, but just to tell you you can increase the limit in the buttons window to something higher than .0001 and rem doubles will work- this is the threshold of closeness at which it works, alternatively you could scale or merge the vertices together.
awesome modelling bytheway.

slikdigit
Posts: 133
Joined: Wed Oct 16, 2002 3:52 am
Location: Northampton, MA (US)

Post by slikdigit »

ok, been looking at your mesh a little closer, and come to the conclusion its too messy at the center to use rem doubles.
reason is some of the verts you want to merge are further apart from verts you shouldn't.
Advice:
use select-> non manifold to select the center region, hide everything else, then hide the mouth and eyes (these are non manifold) then work on the profile section by section. The complicated parts, i.e. mouth, crotch and crack you might want to merge verts one by one.( I mean, two by two)
mirror modelling takes care of that problem since the middle verts line up exactly, at most requiring a uniform scale in the x direction.

philovivero
Posts: 0
Joined: Tue Jul 06, 2004 10:57 pm

It appears I lack clue more than features.

Post by philovivero »

Okay, if I'd known about that "select non-manifold" before, then I would have been quieter. Thanks for the pointer.

Second, I don't know why I didn't see the "remove doubles limit" slider since it's... well, RIGHT NEXT TO THE REMOVE DOUBLES button. I feel so silly.

Finally, here's the latest screenshot:

Image

And the .blend.gz that creates that (with Rigging included):

http://faemalia.net/JumpingGirl.blend.gz

My current feature request (aka what I don't know how to do) is to have bones that are tracked to other bones at ratios: In this case, her kneecap is a bone that tracks to the shin, but I'd like it to rotate 90 degrees when the shin rotates 180 degrees. This might be achieved with weights on the vertex groups, but I don't see how that could be a viable solution long-term.

Am I clueless again?

--
philovivero

oin
Posts: 0
Joined: Fri Apr 02, 2004 6:34 pm

Post by oin »

probably I am saying a nonsense, as I am total noob.


But....

would it work adding a "track to" constrain, and just move to the left the slider of "influence" , just as much as you see more or less fit to the way you want it to react?


If was a nonsense of mine, excuse me...

oin
Posts: 0
Joined: Fri Apr 02, 2004 6:34 pm

Post by oin »

look, I'm a lazy noob, but somewhat curious.

I tried it right now with copy rotation constrain.

in bone mode.

It worked for me if the bone was broken from the bone chain, ie, entering edit mode and adding a bone from whereever I did set the cursor.

I set a bone just in the vertical direction of the bone that will be the origin of the thing.

went to bone mode (ctrl tab)
Added a copy rotation constrain to the new created bone. I did click on the influence number to set numerically the influence instead of slider. I set there 0.5 excatly.

Well, as i rotated the parent bone to horizontal position, "looks" like the "child" one got into 45º in same way. couldnt tell if exactly as I dunno how to see th erotation numbers in bone mode. and anyway, I did not set the bones pixel-perectly vertical, anyway, but more or less :)

philovivero
Posts: 0
Joined: Tue Jul 06, 2004 10:57 pm

Bug?

Post by philovivero »

went to bone mode (ctrl tab)
Added a copy rotation constrain to the new created bone. I did click on the influence number to set numerically the influence instead of slider. I set there 0.5 excatly.
I must have a bug in my version. Whenever I try to type a Rotation Constraint Target, it clears the box of whatever I've entered. Regardless, it never follows the object I've specified.

I chose KneeCap.R and constrained rotation with Shin.R (any Influence number).

The only thing that seemed to cause the KneeCap.R bone to move was if I renamed it, then it would move to some random location (?).

If I deparented the Shin.R from KneeCap.R, it would be even worse. The parenting relationship at least seemed to put the kneecap in the right vicinity.

I talked to my Maya buddy tonight and she said this effect is achieved through vertex weighting, and expressed surprise that I was trying to achieve this effect by adding another bone. I guess I'll start looking into vertex weighting.

--
philovivero

philovivero
Posts: 0
Joined: Tue Jul 06, 2004 10:57 pm

Features vs. Errors

Post by philovivero »

Okay, it's been made pretty clear that of the original 4 things I asked, only one is unanswered. That is, trying to maintain the implicit surface shape when modifying the control mesh.

When I do "face loop cut" or when I do "knife cut" there should be at least an option to try best to maintain the subsurf shape after the new vertexes are added. The very first screenshot I posted with the problem knife cutting the back of the head is definitely still a problem.

Second, about the armature rotation tracking, I figured out how to do it. You must first type in the overall object name ("Armature" by default) then the component part in the box that appears.

So in my .blend, the Kneecap.R should be rotation-constrained to the Shin.R as part of the OB:Armature.

It still doesn't quite do what I want... but I think I'm going down the right track. I'll post a more specific question if I can figure out a problem (rather than just not understanding how to use it).

--
philovivero

oin
Posts: 0
Joined: Fri Apr 02, 2004 6:34 pm

Post by oin »

sorry, philovivero, I was having problems to tell you it was that the other day...now that you've discovered your self(I had to read that in the doc long time ago ;) )...have you tried to put 50% of influence in the slider? just shift click on the slider number...

it worked for me... a non connected bone was rotating and the targeted bone for the constrain, was rotating influenced more or less half of the º of the "father" (is not father-son, is just an isolated bone having a copy rotation constrain in relation with other bone, and setting half of the influence in the constrain...)...

Post Reply