PDA

View Full Version : Morph to assemble, then you need to tweek....? // changing morph paths?



jeric_synergy
10-04-2012, 03:34 PM
So, let's say you've made an endomorph wherein you've spread all the bits of the object out, and you'll morph to get them all back together.

Works PRETTY well, but some objects need their morph path tweeked a bit to avoid intersecting while they move into place.

What's a method to tweek those morph paths mid flight? :stumped:


Thanks!

BeeVee
10-04-2012, 03:44 PM
The only thing you can do with morphs is change the point they are at in their trajectory between 0 and 100% morphed. The way morphed elements move is linear.

B

jeric_synergy
10-04-2012, 04:02 PM
Well, I knew it wouldn't be straightforward. It might require some bones or nodal wizardry. Or something.

Like, eg, combining morphs.

XswampyX
10-04-2012, 04:02 PM
But you can have more than one!

So morph the big bits first and then morph the smaller ones? Or just have the problem items in a separate morph and adjust as necessary.

jeric_synergy
10-04-2012, 04:39 PM
Hmm, how about this: for one-plane motion (like a jigsaw puzzle), the problem bits get 4 subsidiary morphs, N,E,S,W, and one uses those to "steer" the bits thru the problematic frames?

XswampyX
10-04-2012, 04:53 PM
Might work. I was thinking more like this.


http://youtu.be/dpN3UrHM_i0

jeric_synergy
10-04-2012, 08:14 PM
that's from dw's tutorial series, right? I'm mystified how he does that, cuz IIRC .... well, I don't recall correctly, I just remember thinking "wuhhhhh????"

STILL, everything is moving in straight lines: in a 2-d situation, this might cause problem in a morph: I was just wondering about noodging stuff so collisions could be fixed.

RebelHill
10-05-2012, 08:42 AM
Yeah... either multiple morfs for each "group" of parts... or comensation morf that pushes and "offending" part straight along a given axis, added on top of the other morf.

jeric_synergy
10-05-2012, 08:57 AM
I think sometimes figuring out how to just nudge things is harder than the big motions/actions. Maya's layered motions made me drool, but I'm pretty sure we could do the same thing (minus the UI) using DPKit.

RebelHill
10-05-2012, 09:07 AM
not at all... what you're asking for is a basic ADDITION process... you want the item, at some mid-morf point, to move ADDITIONALLY in an extra direction... So... just make a second morf that contains this (small) extra move and add it to the first at the appropriate point.

jeric_synergy
10-05-2012, 11:07 AM
Well, it has turned out (theoretically, haven't tried it practically yet) to be easy, but on first glance even so knowledgeable a user as BeeVee thought "no".

BTW, here's the anim that gave rise to this question: http://www.youtube.com/watch?v=pS8RbY6gRQ0&feature=youtu.be I'm pretty sure some of the more fractally edged counties have some collisions going on, but they're far enough way that as a practical matter it doesn't, errr, matter.

Harumph: my YouTube videos always look so noisy, WUWT?

+: damn volumetrics are so slow.....

XswampyX
10-05-2012, 12:13 PM
Here's one with just 2 morphs. One is just an offset on the y axis (up) and the other one is the random starting positions of the pieces. The weight map denotes the order that the pieces will be fitted. Outer 100 % inner 20%


http://youtu.be/SuxSvSUabvQ

I used weight maps set at 20%, 30%, 40% ...... 100%. Small increments would be smoother.

http://i465.photobucket.com/albums/rr16/xXswampyXx/DualMorphWWMaps_zps13dbf3c3.jpg

jeric_synergy
10-05-2012, 12:21 PM
xswampyx, when I tried to expand that image, it actually got SMALLER (i.e., less readable). :cry:

That's an interesting technique, and I'll have to look at it when I'm smarter, but it addresses a slightly different problem (just to keep the conversation going): My situation was that I wanted everything to move in one plane, like sliding along a tabletop. As the puzzle pieces (actually, county outlines) approach each other, they'd need to scoot around a bit to avoid their edges 'catching'.

Be that as it may, as usual an elegant and thought-provoking response. Thanks, xswampyx!

EDIT1: squinting at that network, I'm wondering, where's the envelope??
EDIT2: a look inside that gradient would help too....

XswampyX
10-05-2012, 05:22 PM
Here's the scene.

The important thing is the weight map. ;)

Have fun!

Edit :- the envelope is the null position. (I just like nulls).

Edit :- Gradient = start on the floor, move up, move down, on the floor. (rested in the random position, float around a bit, rested in the correct position for the puzzle.)

http://i465.photobucket.com/albums/rr16/xXswampyXx/MorphGradient_zps36e918c1.jpg

lertola2
10-05-2012, 07:23 PM
Here's the scene.

The important thing is the weight map. ;)

Have fun!

Edit :- the envelope is the null position. (I just like nulls).

Edit :- Gradient = start on the floor, move up, move down, on the floor. (rested in the random position, float around a bit, rested in the correct position for the puzzle.)


That is interesting. I have been doing some morph manipulation with nodes in a very similar way. The attached scene is an example of one of the setups I have been using. The idea is very similar to yours but uses different controls. Instead of a weight map it uses a uv map. And instead of using a null to control the action I have animated the points on curve node alpha channels.


http://youtu.be/vFoH8Ckd-_M
http://forums.newtek.com/attachment.php?attachmentid=108328&d=1349485530

Here is the scene:

108327

jeric_synergy
10-06-2012, 12:42 AM
Here's the scene.
The important thing is the weight map. ;)
Have fun!

"CSI: NewTek"

Swampy, once again, a huge THANK YOU for all the elegant solutions and explanations you've generously supplied the forum, you're truly an excellent citizen.

And lertola2! Your sentence "... using a null to control the action I have animated the points on curve node alpha channels.", Blew. My. Mind. :thumbsup: