there might be a simpler way...there's a curve fitting program called zunzun which has a python api that I can't figure out yet... https://code.google.com/p/pyeq2/
On Feb 11, 11:34 am, notanymike <notanym...@gmail.com> wrote: > Well, I'm not an algorithms person, actually. So my ideas on the > matter can't be all that reliable, I think. I really just wing it when > it comes to Maya's Python API, I haven't actually learned Python yet, > or any other programming language, for that matter. That being said, > I'm not really aiming for something that makes a perfect skeleton on > it's own. However I'm fed up with auto-rigging systems that start out > with "placers" that are way off the scale and proportion of the model > and sometimes even not in the right pose. I would even be more > satisfied with having a curve-extraction script which I can create > joints off of as long as I had a decent head start with joint > placement. For now the only way I could think of doing it is take the > largest spheres among a group of spheres that are close within a > tolerance, and use the API to get the center of that group by creating > a MFnMesh and adding polygon with all those positions, getting that > polygon's center, then creating a joint, locator, or curve point from > that center. The problem with that is the tolerance needs to be > adjusted when the volume in the mesh shrinks (from the torso, to the > arms, to the fingers, for instance). I have no idea how to do that > part.... > > On Feb 11, 10:07 am, Justin Israel <justinisr...@gmail.com> wrote: > > > > > > > > > What would be the exact criteria for making it produce output that is best > > suited for your joint extraction? > > The sorting im doing does greatly improve performance by getting the data > > radius sorted once very quickly. As for your needs im sure its just a > > matter of how to adjust the comparisons. Right now the algorithm is geared > > towards filling the space with as few spheres as possible. Can you maybe > > explain your pseudo process for what you want to achieve? > > Yea I get kind of distracted easily by these kinds of things :-) > > Its good practice for me anyways to try and optimize maya code. Im not even > > sure why the original developer used pymel. Its way to slow to use in heavy > > loops. Really the goal is to do as little as possible on the python side > > and let the C api handle what it can. Small things like setting properties > > once dont really gain much though obviously. > > > On Feb 11, 2012, at 8:25 AM, notanymike <notanym...@gmail.com> wrote: > > > > Hi, I think the radius won't be set unless in the actual nParticles- > > > createNPartciles menu, the box under "points" is checked, I believe. I > > > had a similar problem as well until I did that. In terms of my slight > > > fancy to have this work as a skeleton-extraction script, do you think > > > sorting method you introduced would make it easier to pick the spheres > > > whose positions would be best for joint placement? > > > > I'm glad that I'm not the only one who got hooked on this little > > > script. Congratulations on what you've gotten so far, I didn't know > > > MFnParticleSystem supported NParticles! > > > > On Feb 10, 7:48 pm, Justin Israel <justinisr...@gmail.com> wrote: > > >> Hey, > > > >> While I actually am not really familiar with the original problem that > > >> this > > >> trick solves, after seeing your code I got somewhat hooked today trying > > >> to > > >> optimize it further. I thought I would post what I've to be able to have > > >> some discussion about it. There were a couple things I was confused > > >> about... But first, here is my pass at further modifying your code: > > > >>https://gist.github.com/1795840 > > > >> Some changes: > > > >> 1. Removed all the remaining pymel > > >> 2. During the first loop that computes the radius of each point, I > > >> create an MPoint as well and save all of the data in an object. > > >> 3. I sort the object by radius size instead of doing that extremely > > >> repetitive loop to find the biggest sphere each time. Instead I > > >> figured we > > >> could just start with a sorted list and work through it. Am I wrong > > >> about > > >> this approach? Did they need to remain in the original order for some > > >> reason? I didn't think they did and all we care about is having the > > >> biggest > > >> spheres first. > > >> 4. For the main intensive loop, since we no longer have to do the size > > >> test, we just go forward over our sorted list, popping off the next > > >> biggest, and then doing the collision test on the remainder. > > > >> I noticed that (at least in my own tests) the original version, yours, > > >> and > > >> mine all did not properly set the radiusPP values on each particle. They > > >> all end up being 0.2. I can go and change them individually in the scene > > >> via the component editor, but I have yet to figure out how to get them to > > >> actually set in the script. So that remains unsolved. > > > >> I think my version should be faster the more steps you use and the more > > >> complex the geometry. When I tested 60 steps on a very large sphere, my > > >> results were showing a ~30% increase in speed from yours. > > > >> Very interested in messing with this :-) > > > >> On Fri, Feb 10, 2012 at 12:26 PM, notanymike <notanym...@gmail.com> > > >> wrote: > > >>> Update: > > > >>> It turns out the code was written by Djelloul Bekri, and Dorian had > > >>> just explained how it works on his blog... > > > >>> On Feb 10, 9:05 am, notanymike <notanym...@gmail.com> wrote: > > >>>> I've converted the following code from being all pymel to mostly using > > >>>> the Maya API, mainly for fun: > > > >>>>http://www.fevrierdorian.com/blog/post/2011/03/14/Remplir-un-mesh-de-... > > > >>>> My code is available here: > > > >>>>http://pastebin.com/3KbhXagZ > > > >>>> It's a bit faster...I also attempted to translate the french parts of > > >>>> the code into English, though I don't speak French, so it's just an > > >>>> educated guess.... > > > >>>> I was hoping to create a "skeletonizer" script, but maybe someone will > > >>>> find what I've got here useful as is... > > > >>> -- > > >>> view archives:http://groups.google.com/group/python_inside_maya > > >>> change your subscription settings: > > >>>http://groups.google.com/group/python_inside_maya/subscribe > > > > -- > > > view archives:http://groups.google.com/group/python_inside_maya > > > change your subscription > > > settings:http://groups.google.com/group/python_inside_maya/subscribe -- view archives: http://groups.google.com/group/python_inside_maya change your subscription settings: http://groups.google.com/group/python_inside_maya/subscribe
