Building a Better Knee

Knee_Detail_FinalThis week I’ve been working on the design for the knee, and I’ve finally got a working design.  As I said in the last post, most figures I’ve seen have knee joints that could be improved upon, at that’s what I intended to with this figure.  Often they will have an oversized mid-joint area that looks awkward when fully bent.  Double joints also has have a tendency to bend fully at one joint before bending at the other, which is a relatively minor issue, but it’s something that I feel could be improved.  And improve it I shall.

I started with this basic concept of putting geared-teeth inside the mid-joint, and in theory it should make both joints bend consistently and at the same angles every time.  Before going forward, I thought it might be more effective to change the angle of the individual joints.  Looking at the way an actual knee bends, it seems like the kneecap stays with the bottom half of the leg, bending a lot more at the top of the knee.  To simulate this, I made the bottom joint larger than the top, making the lower joint turn more slowly than the top.

It took a few tries to get the gears to work right.  My first attempt used relatively small triangular teeth, and while they looked like they should work on the computer, I once again learned that doesn’t always translate to a physical product.  These teeth didn’t mesh at all and would grind a little bit, but that was it.  They felt more like ridges used for grip than gears.


For a second attempt, I tried to adjust the size of the teeth and make them big enough to work in the physical world.  Once again, Looked great digitally.  Physically…


Still just as ineffective.  They occasionally would mesh a bit, but not consistently, and therefor were still useless to me.  I reevaluated the design some, and realized the main issue was I was designing these gears with pointed, triangular teeth, and the printer was not able to handle sharp angles like this at such a small scale.  They came out rounded and prone to slipping and grinding.  I needed a design that wouldn’t do that.  So instead of the sly and shifty triangle, I moved on to the noble, sturdy square.


And BAM!  Working small-scale gears.  Fewer teeth with larger gaps, and every tooth is securely fitted into its complementary gap with no room to slip.  As long as the housing holds everything in place, they should work consistently, every time.  I printed some quick leg parts to better illustrate the full working leg.  This design works just the way I planned.  These are the bones for the final product, now I just need to add the meat.  Extra Knee_Exploded_Finalfeatures will surely make their way in before then, most likely a ratcheting mechanism somewhere to help them hold any pose.  I would like a somewhat complex knee pad with some movement as well.  But all things in time.  That’s for another day.


Fancy Footwork

The main body has been blocked in, so I’m now working my way back up with some additional detail.  Starting at the feet, I made my way up the knees with some more concrete design work.  I’ve also been fleshing out a lot of the specifics of construction.


The feet have taken some work.  I had a general idea of what they were supposed to look like, but they were admittedly not my biggest focus when creating the concept.  I’m probably going to be doing a few follow-up concept pieces throughout the build to get a better sense of the details.  For right now, though, I like the idea of bringing in some elements of trail shoes to this robotic foot design.  It also made me start thinking about different materials I could use when casting the final pieces.  I want to look at what’s available, but doing the soles of the feet and other parts out of a flexible rubber would really give the final piece some extra flair.


I’m trying something new with the knee design to solve a problem that most likely no one else is really  bothered by.  I’ve always found it a bit awkward how so many figures on the market use a traditional double-knee joint.  Many Play Arts Kai figures, for example, use a ridiculously over-sized joint, which makes for weird poses when bending the knee.


I’ve also had the experience that when you bend the leg, usually one joint will bend all the way before the other starts bending.  So, in an attempt to fix that noise, I’m adding geared teeth onto the joints to ensure the leg bends evenly every time.  It’s most likely over-engineering for such a minor issue, but this is my project, dammit, and it will be meet my expectations.

Knee_Detail.jpgI had gone through a few different versions of implementing this idea.  The last had exposed teeth on the joint, and although it was simpler, I hated the idea of being able to see the inner workings of a detail that should be invisible.  Currently, it’s fairly similar to many of the double joints other figures use, just hollowed out to make room for the teeth on the inside.  I will most likely be prototyping this piece next.  I plan on trying a few different gear ratios.  Right now it’s set up for two gears of the same size, but I think one large and one small gear may actually give a more realistic result.  Looking at the way a knee bends, the kneecap appears to stay in line much more with the shin than the thigh.  To achieve this effect, I think I’ll need a small gear for the top joint and a larger one for the bottom.  More to come as this develops.


Form and Function


Legs blocked in

Pressing ahead with the modeling process.  I spent some time this week getting most of the body blocked in.  Instead of trying to focus on one piece at a time to completion, I’m finding it easier to block in everything first, then come back and polish everything up piece by piece.  The block in process has gone surprisingly fast, but I’m sure I’ll get caught up with adding in all the little details and spending more time there than anticipated.

This is also the point where I’m trying to tighten up a lot of the articulating parts.  I’ve modeled the pieces to look correct (roughly) at this point, but haven’t tested the angles of articulation yet.  Playing around and rotating things in Maya is the quickest and easiest way to test a lot of these things before going to print.  The hip joints, for example, had only been roughly placed in.  I placed a sphere where they would go to find the center point of rotation, and moved the joint there.  Then I was able to move the hip piece and reshape it so it would still look right no matter how it is posed.

I’m finding hips are a little more difficult to get right compared to the rest of the body, because you’re trying to mimic a very fleshy part of the body (i.e., the butt) with hard plastic.  Most figures I’ve seen handle it in a few ways: either the classic G.I. Joe T-crotch  (terrible), a plate-like design for the butt to cover the joint from behind (awkward in most poses), or make the ball joint that combines most of the mass with the rest of the leg (best option).  I’ll be going with option number three.HipJointEx


Another piece that’s been taking a while to figure out is the back and shoulder area.  Many of the figures I like incorporate a rolling shoulder for better articulation, and when posing, makes a huge difference between more natural poses and that infamous stiff action figure pose.  However, I want to try something new (at least I’ve never seen it before, that is).  I want to add in a shoulder blade on the back, connecting the inner shoulder joint to the outer body.  If done right, I think that piece could add some cool detail, and have one more moving part to make poses even more realistic.


This is taking a little bit of planning to get right, because it’s essentially a floating piece with two connection points, and needs to be able to handle a weird range of motion.  The piece attached to the shoulder needs to be a ball joint, while the part attached to the back should be a ball joint that can slide, hinge and rotate.

Again, I’m still in the blocking phase, but it’s never too early to start thinking about this stuff.  The sooner I get it figured out, the easier everything else will be in the future.


Body Work

Progress_UpdateI just wanted to post a quick update for the work I’ve done on the model so far.  Originally, I had planned to complete this project piece by piece, finishing the head, then finishing the body, and so on.  Instead, I’ve gone a different route and started broadly modeling the whole thing before getting carried away with details.  There are a few reasons I find this a better way to work.  I want to make sure the model works as a whole and solve any problems as early as possible before going in with the finer details.  I also know my style and technique tends to evolve and improve over the course of a project, especially one as long-term as this, and I don’t want to finish the head first only to go back and improve it later.

I plan to work from head to toe getting everything rough in, then toe back up to head with the details.  I’m going to have to redo what was already a very tentative schedule, but I think I will actually finish the modeling a bit faster than anticipated this way.  Hopefully I can have all the 3D modeling done by summer.  More posts to come as this comes together.

I think I will be designing the thigh holster mechanics next.  The design I have planned is actually pretty close to the mechanics used in the helmet, so that process should be much quicker.


Designing the Helmet


The initial concept

One of the biggest features I wanted to include with this figure was a helmet that can open and close.  I want to show her face beneath the helmet without having to make a second, swappable head the way most manufacturers do it now.  Also, having complex moving parts may make the build more complicated, but I think the end product is far more impressive than simpler figures.

I started this process with a basic idea of how it would function.  Maybe a hinge at the top of the helmet.  Then I added in a parallel set of arms in the helmet to make the movement more interesting and mechanical.  I started playing with the length and angle of the arms to see what effect that had on the angle of the open helmet, and even mocked up some quick ideas with Lego pieces I had available.  Once I had something I thought could work, I quickly laid it out in Maya and went to the 3D printer for prototyping…IMG_3068

…and my first attempt was a mess.  Parts printed poorly, critical pieces were designed too small, some of them broke immediately on use.  I was a little disappointed, but I learned a lot.  Just because something looks like it will work on the computer doesn’t mean it will translate well to a physical model.  So I went back and adjusted.



Success from failure

My second one came out far better.  It actually worked!  Pieces fell off here and there, but it was successful as a proof of concept.  However, after that initial excitement wore off, I tried to look at it more critically to see if there were any more improvements necessary.  I realized that while it did work, it wasn’t actually opening as far as I needed it to.  This design wouldn’t even get the helmet above her eyeline.


Tweaking for better results

Since I knew all the basic components worked, I went back to Maya and made a few different designs there.  I tried changing up the arms to see what would give me the most clearance, and once I had something that looked better, I printed out the new pieces.


Working prototype

The resulting helmet is even better.  These pieces will make up the internals, and I can build the body of the helmet around it.  I’m hoping to have the head completely modeled by the end of April.


Begin Sculpting Phase


I spent a little time in ZBrush this weekend to get a simple base mesh fleshed out.  From here, I’ll be importing the base into Maya to work on bit by bit.  This base is really only to serve as a guide, since the final version will be made of several little pieces.  Not concerned about the details (like fingers or muscles), I just want to get the basic proportions down to have something to work with.

In the past, I would’ve probably done an isometric drawing to work from directly in Maya, but I’ve found that takes a lot more time and probably isn’t quite as accurate.  This way feels much quicker and intuitive.  While I have had some experience working in ZBrush, I’m far more familiar with Maya.  I haven’t had much opportunity to use the two together on a single project, but I’m definitely seeing the benefit in being fluent with both.  The next step will be to get the first pass at a skeleton set up with all the joints in place.  From there I can build out to the bigger pieces.