Mirror point changes

[JamesMK]

Maybe Studio Fabian's "Mirror Point Position"? http://www.studio-fabian.de/


.

[Kuroyume0161]

The only downside with this, as brilliant a solution as is, is it appears to require symmetry of the points on either side. This is not guaranteed - and where it gets a bit tricky. Otherwise, I'd whip something up quickly enough.

[bunter]

I would love to see a plug that does exactly this; maybe it could even use selection tags to choose which points get mirrored. I find the symmetry object pretty limiting/frustrating.

[Cactus Dan]

Howdy,

CD Morph has a CD Symmetry Tag which allows you to move points and have the mirrored points move with them:
http://www.cactus3d.com/SymmetryTag.mov

... of course you have to set the symmetry before moving points around.

The current release won't allow you to add or delete points without losing the symmetry assignment, but the next version will. There will also be a few helper commands included. ;)

Adios,
Cactus Dan

[Kuroyume0161]

Dan, will this work if the points on either side are not fully symmetrical (that is, if some points aren't in identically symmetical positions and maybe counts)?

Thanks

[Cactus Dan]

Howdy,

The CD Symmetry tag will simply ignore the points that are out of symmetry:
http://www.cactus3d.com/NotSymmetrical.mov

Adios,
Cactus Dan

[pjz99]

Keep in mind Dan is talking about his plugin, which I expect is also outside your constraints if MOCCA is. You realize you're hosting a different flavor of this on your site already? Spanki/Cage's morph transfer utility:
http://www.kuroyumes-developmentzone.com/~Cage/scripts/scripts.html

[Cactus Dan]

Howdy,
Keep in mind Dan is talking about his plugin, which I expect is also outside your constraints if MOCCA is.
Actually, Kuroyume0161 is a registered user of my plugins. ;)

But, the CD Symmetry tag is an obscure tag within the CD Morph plugin, and I'm sure many people who have the plugin, have not really used that tag very much.

Adios,
Cactus Dan

[Kuroyume0161]

Keep in mind Dan is talking about his plugin, which I expect is also outside your constraints if MOCCA is. You realize you're hosting a different flavor of this on your site already? Spanki/Cage's morph transfer utility:
http://www.kuroyumes-developmentzone.com/~Cage/scripts/scripts.html (http://www.kuroyumes-developmentzone.com/%7ECage/scripts/scripts.html)

Yes, I realize this already. I followed the development of this script over at Renderosity's Poser Python Scripting forum. There are some downsides to their approach (mentioned in the thread there). It relies upon the vertex normals to determine which way to check and possibly move the vertices which can lead to anomalous results.

This is a key issue with this type of thing - trying to transfer a morph from one geometry to another or morph one geometry into another, for instance - how you move the points is very important when one-to-one vertex correspondences do not or may not exist. One possible solution might be to find all of the correspondences algorithmically and then maybe have the user make the rest. But this could lead to a lot work for the user if there are many lacking correspondences. After the beta-test and update is released.

[Per-Anders]

For Morph transfer then why will VAMP not do the trick? VAMP has a number of algorithms for transfer ranging from closest point methods to closest surface and raycasting in various directions, there really aren't many more optimal automated methods for locating a mirrored point along a single axis without either going into mesh walking which of course assumes that the topology itself is symmetrical (rarely the case), even using a non planar mirror results in more overlap and fold issues than this typically, or starting out with a symetrical mesh and precalculating the mirrored points before deformation occurs (VAMP backprojects when dealing with Morphs in order to try and do this). There should be a few interesting papers out there on the subject, I believe it was a hot topic in the early 00's.

I'd say that if none of the existing methods are acceptable, then as you're a developer already the best course of action would be to make a plugin yourself to do exactly what you want, people can always do with better tools!

[Kuroyume0161]

I can't speculate. It was someone else who first made the proposal to me about said support due to insufficient quality and success with VAMP.

This is exactly what the fishing expedition is about. :) If solutions already exist then why should I divert time and energy replicating them. If not, then I can start researching the methodologies.

Same reason why I made an inquiry to Maxon a couple months ago concerning COLLADA support before starting any process on my own. And, as everyone sees, it was a very worthwhile inquiry that helped avoid useless diversion.

Thanks,
Robert

[pjz99]

I'll own up that it was me that initially made this suggestion (at least I think it was) and the scenario that I tested with involved a figure that was modeled in a pose not-so-great for this kind of thing (legs tight together, so the mesh went nuts at the crotch). For other applications I don't have experience with how well VAMP will convert a morph target, although I expect with a figure modeled more like DaVinci's Vitruvian Man, at least where the legs are concerned, it would do a better job.

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.