I’ve essentially been rewriting the script over the past few months, both to make it so most of the functionality can be re-used for legs, and also to add independent controls for each joint.  What’s left to do – support for double-hinged limbs, parent-space switching, and IK/FK matching.

Getting it to this point is a huge load off my mind – the biggest problem I have with mel scripting is that whenever I get back to it after too long a break, I can easily spend half a day just going over my notes and analyzing the code, to figure out where I had left off.

The left scan was made with David 3, using the structured light method.  The right scan was made with the (not-David) laser scanner.  Apart from the lines on the David scan*, and the edge noise on the other scan, the results are fairly similar.  More importantly, they’re too similar to justify paying so much more for the oher scanner.
* these lines can be reduced with proper filtering of the raw photos, plus I’ve found that if the camera were placed more off-center, the object would’ve also been scanned using vertical lines.

While it’s a nice first test, the dark colors and shiny surface of the original object aren’t ideal. I’ll have to find a decent white powder spray before attempting to scan the entire figure.

Another concern was how well David could align and merge multiple scans, two tasks the other scanner failed at. To test this, I tried a different model – part of an unpainted and still-unassembled resin kit of Mai Shiranui -

While I would have liked to have scanned something a little more appropriate, this piece’s complexity and lack of color made it an ideal candidate. Since it combined the scans without any problems, now I may have to buy a David license.

Another clear advantage to David is its flexibility, in terms of hardware – while some use webcams to capture their scan data, and others use more expensive ccd cameras, I was able to use my dslr. While this method is slower, it means I can easily rescan the objects without having to retake the photos, unless doing so means I can get a better scan.

I also recently bought (and quickly returned) a certain desktop 3D scanner, which I will not name. I had a full review typed up, but fornnow I’ll just sum it up by saying it was a complete waste of time and money, and a disappointment on almost every level. The only positive thing was that I was able to return it for a full refund, and didn’t even have to pay for the return shipping.

While that scanner was disappointing, it wasn’t enought to put me off of scanning completely. It just gave me more incentive to try the alternative – David. More specifically, I wanted to try scanning using structured light from a video projector, rather than a laser. The beautiful thing about this, is that I didn’t have to buy any hardware – I was able to simulate the whole process in Maya, and have the software scan the model from the renders!


Even better, this could also mean that the only piece of hardware I’d need to buy is the projector, since I already own a decent camera.

Basically I want to sell the thing to free up some space (the money won’t hurt, either), but at the same time I want to keep it, for sentimental reasons.  I got the original version of this for Christmas back in ’87, and it quickly became one of my favorites.  Time wasn’t too kind to it, or at least I wasn’t – when I eventually got bored with it, I decided to take it apart to see how everything worked.  Unfortunately, much like Humpty Dumpty, I couldn’t quite put it back together again.

Skip ahead 15 years or so, and Takara re-released the toy as part of their then-current Transformers series, just with a new color scheme.  My inner geek won out, and I wasted no time ordering the thing, thinking it would be a nice addition to my desk at work.  What I forgot was just how big it is – the thing stands roughly 2″ tall!  It never quite fit anywhere, so it’s been sitting in its box on a shelf ever since.  Now that I’m attempting to model it, this is probably the most attention it’s ever gotten.

For the new year, one of my resolutions is to get through as much of my to-do list as possible, starting with various drawings and mel scripts. I’ve got a good head start on some of the new images over on Deviant Art, and over the past couple of days I’ve been working on some small mel utilities.  Once I’m out of those, I’ll have to get back to the big ones – my character-rigging tools…

With Daisy, the bulk of the changes are technical. Her rig needs to be cleaned up or replaced, clothes and hair should be set up to use dynamics, etc. There were some changes to her mesh a while back, but those are relatively minor.

Mary, on the other hand, will be a bit more extensive. The first thing I’ve addressed so far is an old one – the length of her legs. They were too long compared to the rest of her body, but using Daisy as a measurement made it clearer – if Daisy is 3’6″, then the old Mary is roughly 6’3″. The height of this new model is closer to 5’8″, and could probably go an inch or 3 shorter.

Her fingers were one area where attempting to match my drawing style was a bad idea. They ended up being too sausage-like. Her new fingers are more naturally shaped, with better (and clearer) joint proportions.

Her spine was another major issue. The curvature of her default pose was too strong, which was fine if she ever needed to arc her back even more, but if she bent forward, getting decent deformation was impossible.  Hopefully having her stand straight by default will fix these problems.

Her breasts will also need to be toned down. I don’t know when exactly they got as big as they are now, but I’d like to take them back down to their original size. One busty character (Daisy) is enough.  Whether I do this through modeling, or through rigging, I haven’t decided yet.  Either way, I’ll also have to do something about the “anti-gravity” pose, if only so I won’t have to explain it any more…

When I’ll be done with her is another question that I don’t have an answer for.  Right now, a higher priority will be finishing my rigging tool, and maybe implementing some of the more recent tricks I’ve picked up.  Once that’s out of the way, I think it’d be much better to try it out on a new character.

And here’s a couple of more recent pics by Dutch –

The other day I saw that Disney Animation is hiring pretty much every position, and got thinking that I should update my reel and send it out… but update it with what? The game I’ve been working on isn’t finished, so I can’t show anything from that, which basically means I have noting new to add.

Since I can’t show any of the models from work, a modeling section on my reel would be limited to Mary, Daisy, Bill, and Mercy. Bill and Mercy were never finished, but may at least be presentable. I’m not so sure about including Daisy, though, without having more to balance her out.

A rigging section would be fairly limited, since my most complete rigs simply aren’t very good for animation. A mel scripting section would also be tricky, since most of the utilities I’ve written might not make much sense in a video. Showing my arm rigger could help, but it’d be even better if it could also do legs.

So unless I start burning the candle at both ends again, it looks like for now I’m stuck. I can do this stuff, I just can’t show it at the moment…

The experiment, or at least the goal wouldn’t have been completely wasteful.  The idea was to take my multiShape sctipt, which is pretty slow, and turn it into potentially fast custom node.  Using one of the example scripts as a starting point, I was at least able to generate the node and its attributes without causing Maya to crash.  What I haven’t figured out yet, is how to get the node to actually generate the data that will be fed into a shape node.  With how little I know of the language and commands, or even object-oriented programming, I didn’t expect to get the node working, but I had hoped.

Having wasted enough time on Python,  I finally decided to use what I do know to write a tool for something my rig needs – stretchable limbs.  It’s not built into the limb procedure yet, but at least it works.

This idea hit me when mel script finally started to make sense. Even if I didn’t know the commands, suddenly I knew how a script should flow. Sometimes I would go to sleep frustrated that I couldn’t get a tool to work, and by morning I’d have the problem solved. More recently, I had to pull my car over on the Florida Turnpike, just so I could scribble some notes. It’s typically a 4-hour drive from Miami to Orlando, and the script I’d come up with was something I didn’t want to keep in my head the whole time.

Now that I’m making progress in Mel, I’ve been giving Python a more serious look. I was surprised to find that it’s not that different. The difference in syntax is minor, but the way the code flows is more or less the same. Python may look better on a resume, but for now I don’t see much advantage to scripting in one versus the other. If I decide to delve into Maya’s api, then that would be a different story…