Just checking on the snap feature

[detbear]

Hey Everyone.

 

I tried to find my old thread about this but could not locate it.

 

I inquired previously as to if there is a way to snap a bone's base to a cp. Has that been figured out? Or is it

still not possible?

 

Sorry about the re-posted subject.

 

Cheers,

 

William

[robcat2075]

The only semi-automatic solution i recall is to use the Group constraint with "Use Offsets" OFF to translate a bone to the center of a one-CP Group.

[detbear]

Thanks Robcat,

 

Again..sorry for the re-post.

 

So that means that each cp would need to be turned into a group prior to any bone snapping......Correct?

 

Would be awesome to right click on a bone and have a "Snap to cp" option on the list. Once you click on that option,

then you "left click" on the desired cp and it snaps the bone to the cp. Only the translation....not the rotate or scale.

 

 

William

[robcat2075]

Tell me again why you need this?

[detbear]

I have a dense mesh and I need to put a bone on each cp.

 

It is very tedious to do by hand.

[robcat2075]

Tell me again why you need to do that.

 

How many CPs is it?

[detbear]

Hundreds

[Fuchur]

Just a quick question: Why do you need to put a bone to each CP?

Would it not be easier to just use CP-Animation for that or do you want to create dynamic constraints on each bone or what is the reason behind that?

 

See you

*Fuchur*

 

PS: I think this should be rather easy to implement with a script-language which can manipulate text-files.

[detbear]

Because I want to expand the potential of each cp.

 

Very hard to explain. I want the layered option to first drive a cp with a bone.

Then all the other stuff you mention can be layered on top of that process.

 

This potentially can expand the flexability of a cp's in a model in a "re-targeting way"

[robcat2075]

Can you show a small case, like on a simple lathed cylinder, that demonstrates what you will do with this?

 

I still don't' see it.

[detbear]

Hey.

 

Well.... I guess if there's no way to snap bones to cps....there's no point.

 

If I can, I'll try to get something basic together. But I don't have much time for it at the moment.

[robcat2075]

It would not be impossible for a coding person to create a wizard that would run through a selected set of CPS and create a new bone at the same position as every CP.

 

Before the coding person would go to the trouble of doing that you'd need to explain a compelling reason as to WHY this is useful and SHOW it being useful in sample case that you did by hand.

[detbear]

OK..... I cannot express how completely primitive this is, but it is my weak attempt

to explain One of TWO neat uses of being able to attach bones to cps.

 

This is ONE of the expandable uses:

 

A way to make a base mesh have "Re-targeting" abilities. Today I messed around with

the "Group" constraint and found it to be quite successful. But you still have to place bones

on each cp and group each bone to

 

Here Goes!

 

Let's say I have a base human mesh Head that I'm satisfied with. I rig it and get it all working

with all the gidgets and gizmos so that it has a really great animating system/ weighted/ etc.

 

Then I deside I want a second human character.But I want this one to have slight changes as we humans do.

So I take the Head mesh with the same cps and re shape it into a new guy/ gal.

 

I don't want to re-rig the thing, I just want to re-target animation from the first rig.

 

I believe it's possible to get real close using a method similar to what I've attached here:

[detbear]

Here is an action Test Of the examples

[detbear]

I threw things together pretty fast for those samples, but with time and patience, a weighted and rigged model

would perform much better.

 

I envision the potential possibility of a set of different faces with the same mesh that could receive retargeted animation

from motion capture or hand keyed work.

 

 

THIS DOES not require being able to snap bones to cp's, but that would help in the process.

 

The second help that a "snap bone to cp" would bring has to do more with matching cp's on imported models into A:M.

Most .obj's have heavy meshes as you all know. But working with them and morphing their shapes is not easy.

[detbear]

I'll just do it by hand I guess

[robcat2075]

I hadn't had a chance to digest it yet. Can you put the pictures back?

[detbear]

It's O.K. Big Guy. I decided not to bog everyone down in something like that until I can get a more clear

example up and running.

 

I've got some face plates from morphs out of another AP. They have to be imported as separate models in A:M.

Even though the cp's are the same(but moved to make the shape)....they have different names and there is no way to

easily move the base model into the new shape.

 

If you could snap the cp's of the base model to the cp position of the morphed model it woul make the process much more accurate

and way less time consuming.

[robcat2075]

If you could snap the cp's of the base model to the cp position of the morphed model it woul make the process much more accurate

and way less time consuming.

 

 

This sounds like a completely separate thing, right?

 

 

Here's how I think it could be done...

 

-create a plugin that will move a CP in Mesh A to the nearest CP in Mesh B when both are in the same model or pose window.

 

the immediate problem with that is that perhaps Mesh B is so morphed that the "matching" CPs are no longer the nearest to each other. so...

 

-make the plugin restrict its search to only CPs in a Group the user has defined. You might make a group for "Upper Lip A" in the A mesh and "Upper Lip B" in the B mesh. The plugin would only look for nearest neighbors for group A CPs only among the CPs in Group B.

 

You might have to make these groups very specific before the right targets were being consistently found but that would be easier than manually matching every CP individually

[detbear]

Yes. That would do it for sure. Even if certain cps didn't get "snapped"...it would be easier to

correct some by hand.... rather than the entire thing.

 

WOW...that would be a great plugin because it would essentially transfer morph targets by a back door

method.

 

How difficult would a plugin like that be to create I wonder?

 

Since I'm not a programmer, I have no idea how difficult that would be. Wonder if anyone would do something like that for

pay? Although...I don't know if I can put it in the budget.

 

I actually did one morph by hand and it looks soooo Good until I've been racking my brain as to how this

may be accomplished easier.

[robcat2075]

WOW...that would be a great plugin because it would essentially transfer morph targets by a back door

method.

 

It's a notion I've had in the back of my head for a while. Ideally we would be able to do what you were trying before, to paste a mesh copied from a model window into a pose window but there's no way for A:M to know how to relate the two different meshes.

 

This is the only path i can think of in the absence of an importer that could fully interpret other programs' rigging and morph target stuff.

 

For this plugin to to be useful, we are presuming the user has access to some polygon program where they can manipulate the original character model they are wanting to remake in A:M and export various versions of it like "smiling", "frowning"... etc.

[detbear]

Yes.....it would be a plugin for:

 

1. Advanced users....most of whom have knowledge of character usage in other apps.

 

2. Those who would like to be utilizing the un-matched features in Animation Master with external characters

that they are going to be working with regardless.

 

Personally, I think it would be a draw for A:M......But that's just me. I would certainly like to have this function with

my characters from other Apps.

 

Wish I could make this happen. I certainly understand the limits of the programmers for Hash. Nevertheless, it is a huge

part of my wishlist.

[robcat2075]

Hey,Kevin... have you been using "Snap to Surface" when making these pose shapes?

[detbear]

Hey,

 

Yes. That was one of my first solutions in trying to do things by hand. I've been working on it little by little after hours on my laptop

though which only has V15 and no "snap to surface" feature.

 

Still.....in order to get the cps into an extremely close position it requires "nudging" most cps with the "arrrow" keys most of the time.

[robcat2075]

Could you post an example of an original and morphed mesh that I could experiment with?

[detbear]

Robcat,

 

sent you an e-mail with a sample attachment.

[detbear]

I managed to get 2 morphs done by hand. It took a bunch of time though.

[detbear]

Robcat,

 

Did you get my e-mail(s)? Just checking so I can send it again if not.

 

William

[robcat2075]

All I got were movies. Send me a model with an original mesh and a morphed version.

[detbear]

I sent it a second time earlier.

 

If you didn't get it, I'm not sure why?

[robcat2075]

Nothing yet.

 

Is it large? You may need to Dropbox it.

[zandoriastudios]

What if you were using a face rig like ZignTrack on your characters? Since it is, or can be, bone based animation, the facial animation ( or poses) from one character will work on another with the same rig. Just like other actions in A:M.

[largento]

If I'm understanding this, he doesn't want to have to re-weight the CPs. Each CP would have it's own bone.

 

To give an example of what Will's talking about, there's this thread where Mark S. placed his face rig into my Wannabe Pirates models and ran the same zign track action on each of them.

 

You can see that differences in the meshes did play a factor. Part of that was probably due to the neutral expression of each model and how big their proportions were.

[robcat2075]

My understanding is that zigntrack animates bones but it doesn't do the weighting for you... or does it?

[detbear]

Actually these are two separate things in this thread.

 

However....Largento is correct with the facial mocap issue. If there is a bone controlling the points, then a base face can drive the cps.

Weighting is then not being created as it normally should. But rather.....as in Largento's Pirate....a separate character and it's rig is driving

the shapes. The cp's need to be the same in the base model as in the target model.

 

AS FOR the other issue in this thread, When you import the same mesh from another App twice. The first import is a neutral pose.....The second import

has the same cps(because it is the same model in another App) but is posed with a blink, or a phoneme shape, or an eye brow raise......

The issue there is that there is no current way within A:M to blend these two shapes in a pose so that the neutral obj transforms into the second posed obj.

 

You can do this by hand in a pose by bringing both models into the same model and the matching the cps in a pose....but that takes forever.

 

****The reason I originally started this thread was to try and find out if there is some way to snap cp's to cp's or bones to cp's in order to speed up that process.

[robcat2075]

Send me the file in a private message. Those can have attachments.

[detbear]

OK,

 

Robcat....I sent over the files via the forum mail. Did you get it?

[robcat2075]

Got 'em.