modeling and rigging (some nudety) wip
polycounter lvl 18
Hi all , i havnt posted for a little while, ive been working on a new rig that has facial muscles as well as driven keys etc, im trying to get it so there is almost no tweeking , but the nastyest areas to get looking right are the shoulders and thy areas ...the face bits arnt rigged and the texture is not really even a serious attempt , ive gota add loads of detail to that ... (been looking at daz`s new tut) wot i really want to know is if anyone has any good shoulder and pelvis systems , and what they are ... im working in maya and i want to steer away from using too many driven keys ... and i dont realy want to use deformation cages at all ... just bones and ik ... i wish i could show the hypergragh node layout as this would show wot ive got going on ...
the modeling for the head and body is almost done ... about 35k at the mo i think ... i havnt modeled the hands and feet yet really ... the main trouble like i said is the areas where folds are ...
sorry about the generaly bad way ive presented this ...
i just wanted to quickly get some feedback , while im finnishing the rig ...
the modeling for the head and body is almost done ... about 35k at the mo i think ... i havnt modeled the hands and feet yet really ... the main trouble like i said is the areas where folds are ...
sorry about the generaly bad way ive presented this ...
i just wanted to quickly get some feedback , while im finnishing the rig ...
Replies
She has got a cute face though
The biggest deformation problem is definitly at the hips, and results from (I think) joints that are too low.
Have a look at the center or rotation on your own upper leg.
[edit] I am not a big fan of double joints like you used in the knees, since they don't represent the real bone joints. Do they work with an IK setup?
Surely these aren't so much 'rig' problems as skinning problems? I'm not convinced you'll ever be able to get around that creasing in the hips there unless you use blendshapes. Are you skinning or wrap deforming?
Youve seen this stuff Im sure:
http://www.optidigit.com/stevens/rigtut.html
http://staff.ci.qut.edu.au/~barkerc/Final%20PAN%20website/charactersetup.htm
I dunno, it seems like youre looking to achieve a really realistic setup, but without set driven keys I think thats gonna be tough. Are you trying to build a generic rig that you can use again, or a one time perfect setup for her?
the actual model proportions are taken from a real life model ... thats why they are so skinny .. as for the hips , the hip bones are ok , they fit the model , and the hip joints are in the correct place in relation .. but i get some really bad streatching on the ass , and some really bad folding on the front ... i tried to fix this with 2 bones set to move when the leg box translates in y ... it sortof works , but i cant help thinking ive missed somthing ...
i dabled with some crazily complex ik systems to get the shoulder to work and in te end i recon the more bones you have , the worse the deformation ...
is my mesh too dense where im trying to fold ? ill post a wire frame ..
funny , that second link i was looking at just yesterday thinking that shoulder setup they have looks quite swish ... i do want to steer away from blend shapes.
notice how in the first one they never show the leg bent upward
i think this is sorta a rig and skin problem ...
maestro , thnx for the suport , im sure daz didnt mean anything bad, i respect his work . hes darn good.
anyone got any ideas , ive just considered , maybe a spine ik bone structure attachech to the movement of the fatty areas of the back side end ... i would draw a diagram , but i cant be assed, (get it)
i gota say , i never meant this to be contreversial in any way , apart from maybe my modeling style
Issue? I saw just making a observation and just stating my opinion and experince. Any thing you took from that may have indeed been out of context. Sorry.
The red line is where I guess your current axis of rotation is.
The green is where the axis of rotation should be, and the blue shows where the fold for your skinning ( how the vertexes are assigned to the bones) should be.
Now before the acusations start, the redline is just where I guess your axis of rotation is, judging from how the model folds at that particular place (and from looking at the second picture you posted). But it might as well be a strange way of skinning it (like Daz suggested?).
I did some research on the knee- two joints issue and while there is some evidence for your two joints theory (the knee is a pretty complex joint), I (and quite a few leg-prothesis makers
But you might be right, a two bone setup will probably give even more acurate results, but I doubt that it is worth the extra work... I will experiment with it some time.
Ok on to the other stuff
I don't think it is the small boobs that make her look a bit young (not really 13 though), it is more the hairstyle and a maybe a bit to large head compared to the body (just a tiny bit). Oh btw the neck might be a bit to long also, not sure though.
Well and the wireframe shots show that there is something generally wrong about the hip and butt shape.
The butt-cheeks (is that how it is called in English?) are too...hmmm...'pointy'(?)(exspecially for a extremly skinny girl that that) And the part of the hipbone that sticks out is at the wrong place, and also over pronounced (maybe not since she is so skinny).
The area where your leg is the thinnest (a ring right under the butt-cheek) looks really strange too. Exspecially on the side of the leg, where a normal leg is not bend inwards like that. I might be overinterpreting this (!), but together with the placement of the hip-joint it shows me that you might have a general misconception about how the hip works and looks.
And last but not least: There are some horrible (ok I might be overreacting) mesh-flow/egdeloop termination problems, exspecially where the face and the rest of the body meets.
Now remember:
"If someone is criticized, he must be doing something right, because only the guy who has the ball gets attacked" -loosy quoted from Michael 'Air' Jordan
[Edit]
[ QUOTE ]
Im confused Thermidor. If there has been a meshsmooth how come there are triangless in the geometry?
[/ QUOTE ]
That confuses me too
Personally, I think a lot of problems can be solved by good skinning, but that's debatable. How are you doing the skin weighting by the way? Painting or using the component editor?
And one last thing, I like using the two joint knee setup too. It's helps make the knee less mushy looking when it the leg bends.
Daz : this is a early wire of it before smoothing ... theres tringles all over the place , i guess it just divides them up too ...
Fatassasin : yeah , my girlfriend bought me that book for my birthday in august, been trying out lots of stuff ever since then ... i havnt got the hang of the component editor way ... so im just doing some small editing on the paint weights tool .. want to try and get a reasonable initial bind...
most of the modeling was done with reference photos and a really good anatomy book in front of me , but there are some areas where ive not finnished modeling yet .. thats why the neck has no modeling ... i just stuck it on after modeling it seperately , just so i could test the way it deforms on the rig ... thnx for the help guys , keep it coming ...
I finally figured out that you're using a somewhat unorthodox smooth algorithm that doesn't turn everything into quads. Youre using 'linear' instead of 'exponential' right? Any reason for that? I would really try and stick with the latetr If I were you, and furthermore try to use less tris in the first place. I think you'll find everything else will come a bit easier after that. ( and light will bounce off the model in a cleaner way. ) It's alot easier to delineate particular edge flow lines with quads, its less 'stuff' to deal with visually and easier to lay our UV's etc.
If you are doing this, you should definitly NOT do this. Smoothing is basicly done as the last step before rendering (and in between sometimes to check the result, but always reversed afterwards).
The subD cage should have all the detail that is in the final mesh (excluding bumpmaps, displacementmaps and or the Zbrush stuff), smoothing is just done to make it smoother
P.S: Daz the soothing algorithem mostly used is called 'catmull-clark' if I remember correctly, and it does turn everything into quads.
P.P.S.: the visible part of the hipbone needs to be in front of the hip, not on the side. The fold/ thinnest part of the leg should be place diagonally right below it, not almost horizontal like you did it.
about the hip , its mainly visable from the front as the sides for the pelvis ... the front of the pelvis is sticking out abit too much i think tho ...
this shows the joint layout as it is , and im pretty sure its more or less right ... maybe i should have the ik on the leg skeleton start where it is , but have an extra joint before it to make the rotational axis in the place it is on a real skelly...
still , its a larf looking at my dodgy drawing in red of a pelvis ... hehe
Thermidor, yeah JK is right. Tweaking the mesh after smoothing is totally defeating the beauty of the workflow of Model>Rig>Animate>Smooth.
He's not using Catmull Clark JK. 'Linear' smoothing in Maya is using a totally different algorithm which sort of simply doubles the amount of geometry you have. Messy.
[/ QUOTE ]
Sorry misunderstanding, I ment Catmull Clark is what he *should* be using (I am not used to the Maya terminology of linear and exponential).
@Thermidor:
If you wants something to be sharp, and an extra edgeloop, or move the existing edgeloops closer together.
Oh and subD is not box-modeling, but simply the method of creating a lowpoly-cage and smoothing is later on.
The point of rotation of the upper leg is definitly to low, it should be somewhere a little bit right (horizontally) of the middel of the blue diagonal line, that goes from the vagina to the outer part of the hip bone.
But if all of the blue line represent bones, than you definitly got quite a few totally useless one, since the hip is not moveable in itself, nor is the upper legbone at the top (the stuff you drew on the left side).
i dont actually have the blue bone pointing at the vagina , or the first bone on the leg chain , but im thinking of putting that one in , so the rotational axis of the leg skelly is more accurate ... when we lift our legs the rotation isnt just att eh hip , so its hard to work it all out ...
today , i think i will go back to the modelig , try to get the low poly one back to a good lvl of detail , and approach it all hoe you and daz advise ... then i think ill try redoing my leg skeleton to fit the jiont placement abit more ...... i will post again later , so u can tell me where ive gone wrong ...
i have gone back to the original low poly mesh and remodeled some details in , its quite a different way of working for me , as ive always tweeked little details on the high poly model ... not adding polys , but moving verts a little ... still , its looking ok ... ill post a shot when im done on hands and feet.
I still see alot of tris in that foot, and not very even poly density. The nails compared to the toes for example. Youre trying too hard to define the curve where toe ends and nail starts, instead of letting the smooth do the hard work with the curve for you. The form of your modeling is really nice for the most part. I just think you need to work on your topology. One of the best ways to do that is to look at a ton of other wireframes perhaps.
i definately have to work on my topoligy , but thats always the case
some shots of the texture un-lit... still some seams that need fixing.
this is where i am now, it dosnt look like im having much progress , ive sorted out some areas that i didnt like , but things just dont quite look right ... im thinking of trying a different clavicle system ... does anyone know how to move the pivot point of bones that have ik attached, without drawing another bone ...
this shot is using the standard bind with a little smooth skin weights , and then the smooth poly turned up after ...
After having animated a skeleton, where the bones and pivots are "out in the open" I noticed the animation was a LOT more "realistic" and facile. The thing about the shoulders and arms is that they are not "solidly attatched" to the body, but float on a layer of muscle, over the ribcage. That coupled with the fact it's a ball and socket joint makes the shoulders especially difficult.
So what I would suggest is that you pivot the clavicle, where it actually pivots, which is at the top fromt of the ribcage. The model is skinny enough you have a clear indication of the clavicles. The diagram shows the positioning. To keep the mesh under control you can add a couple of bones to act as scapulae to move the mesh, or a reference object so slide under the skin. An arraingement like this will give you ver realistic shoulder shrugs, as well as giving a better indication of the slide over the ribcage.
Another suggestion is to add an extra joint between the shoulder and elbow, so as to disconect any "twist" rotation from the shoulder joint, which will reduce the incidents of "Beer canning". You can then blend the transition between the joints so that there is a gradual twist all along the arm, and preserving the shoulder from any twisting. If you delete all the channels except the proper rotation,, and then limit the rotation on that same axis in the shoulder youy can include it in the IK solution, from the shoulder to the wrist, as long as you are using an RPSolver. I hope this helps.
Scott
but the model is beautiful, i didn't mean to dishearten you, keep up the good work!
Thanks for the comments scott, im gona put a rotation in the arm like you have sugested. and im gona keep the clavicle and scapula seperate, maybe pivot the arm seperate as well , and use SDK to make realistic floating.
as for the 13 - 14 year old stuff
If you want some actual *constructive* advice on what would make her look older ( but retain the 'skinny' feel ) , here are some thoughts:
Head in relation to body size
Ive yet to see a full figure front ortho, but it feels like her head could be big. 7 and a half heads tall might be an average woman, so look into that.
Face:
Although its very cute, theres a certain 'chubbiness' to the face that doesnt fit with the very bony nature to the rest of the model so it might come across as teen puppy fat. Particularly around the mouth. And overall the 'convex' nature of her facial bonestructure makes her seem younger than she might be. Also check eyesize. If theyre big then thats a sure fire way of making her look young.
Hips:
Without ruining the skinny feel, you could definitely go wider on the hips. Really skinny flat chested women still have hips wide enough to give birth.
These might help ( Not Safe For Work images )
http://www.fineart.sk/show.php?w=710
http://www.fineart.sk/show.php?w=713
http://www.fineart.sk/show.php?w=731
http://www.fineart.sk/show.php?w=674
Keep at it man, youve the makings of a fantastic piece of work here.
the face is sorta based on somone i know , but yeah , the area where the cheeks meet the mouth is probly a problem area...
thanks for taking the time to help me out ... illl keep at it , as it is , im quite pleased with the face , and the body .. which is i supose why it hurts when non constructive comments are made. ill post when ive finnished remaking the rig
last time i did some solid life drawing my lecturer told me 7 heads is the norm, allthough my anatomy book says clearly 8 heads is the norm ... very good book other than that.
8 heads tall is the artistic ideal and its been for a long time and when humans get smarter and smarter their heads get bigger.
Don't get me wrong, I think your anatomy is really well executed on this. Sorry if I made offense though in the earlier post.
The other thing that Iam noticing about the rig you have posted to the top, isthe more complex the rig, the more it resemples the articulation on those high end 12inch poseable action figures. So there's another reason to buy that 18 inch tall poseable Spiderman, to get rigging ideas, and be ahead of the curve. :-)
Scott
so i did somthing like this ... the CVs of curves are attached to the bones around by parenting transform constraint clusters. heres a really good drawing.
and it seemed to work ok , apart from i couldnt attach the curves to the skin cluster. this is probly to do with no knowing too much about how this stuff works , and i thought that maybe it wasnt worth it anyway , as the binding method i was using was cuasing all sorts of havok. anyway , i saw a thing about somthing called muscle TK (i think) whick uses some fancy all singing system , but when i watched a video of a guy demoing there muscle tech , it was virtually the same idea as what im using.
wot i want to know now is , can this be done ? and is it all worth it ? im really sorry for my lack of communication skills here. (and spelling)
How are you doing Facial animation? the same way?
Scott
the face stuff is done with bones too ... mostly just useing stretching of bones to simulate muscle .... since i started binding to closest distance rather than closest join its fixed all the hirachal skinning problems i have ... it seems the closest joint works on the hirachy , so it gets it wrong alot ... hmm , maybe somone could talk to me about this on msn or somthing , coz i dont know anyone at all who is into or knowlegable about 3d , so this forum is my life line , although having a conversation is slow :S
as for the model .. ive come upon a revelation as to why my shoulders were being so hard to deform , i think they were rotated forward too much in a un natural pose on the model, so im looking atr fixing this , i havnt tested it yet tho ...
is there a reason why we all keep the T pose? i can tthink of several areas where a slight bend on joints would really help deformation later ... like on the hip joint.
im also redoing the face to a major degree, some big changes , i just hope i can get it looking nice ....
and im generally tidying up edges and stuff , all though im still not convinced that a tri here and there is a bad thing ...