Nyeng!

Daisy’s shoes and socks are finally skinned, which means that now the only things left are little odds and ends.  A lot of them.  One of those things is the buttons on her shirt, attached with rivets -

I also modeled a tongue, though I’m curious how far I might be able to get without having to skin it.

My biggest annoyance right now is shading.  Since she’s gone through so many revisions, some of her shaders had to be pulled from older files.  Everything seemed to be going well, until I did a test render with the output passes enabled.

The beauty pass turned looked fine, until I saw the alpha channel -

Even with custom alpha channels enabled in the render settings, Mental Ray refuses to render a transparent alpha for her lenses.  Searching for a solution online didn’t turn up anything, either.

Things got a little weirder and a lot more frustrating when I loaded the passes into Photoshop -

First of all, I have no idea why her hair is yellow in the ambient occlusion image.  Her glasses are a bigger problem, though.
The only layers I wanted her lenses to render out were specular, reflection, and refraction.  But I didn’t expect them to render black in all the other layers…  I was expecting everything behind them to be visible.  Since I prefer to adjust the layer blending in post, having her eyes only visible in the refractions kinda throws things off.  Even if I get rid of the lenses, this could still be an issue, because I use the same shader for the out layer of her eyes.

Once again, her pupils are causing me problems.  Because of the modeled pupils, I added extra, slightly larger spheres to handly the reflections.  But because of the issues I ran into with the glass shader, now I’m back to experimenting with projected pupils.

Problems like the one with Daisy’s eyes really annoy me…  with all the other things still left to do, problems like that suddenly become top priority, which for me usually means at least a day will be wasted trying to solve the puzzle.

After playing around with the shaders and projectors, I may have found a way to keep the old eyes.  The first step was to create a spherically-projected black & white ramp, with no color interpolation.  The color positions and the projector were then adjusted so the front of the eye would be white, changing to black just beyond the eyelids’ max. range.

While this map will be connected to the shader’s opacity, an extra step needs to be taken for the shadows.  On its own, Mental Ray will ignore the transparency, and render the shadows on the surrounding objects as if the eye were still visible.  Connecting the ramp to a shadow shader (p_shadow_transparency) fixes this nicely.

Now for the really stupid part…  after spending this much time on a problem, it seems there’s always something stupid I’ve overlooked.  Sure enough, once I finished assembling the shader, I noticed that the roof of her mouth is actually at least half an inch above the top of her gums.  After pulling the geometry down her eyes still poke through, but now it’s much easier to correct the problem with a lattice deformer.

…she has balls in her mouth -

Lame joke, and an unexpected side-effect of her design.  Like many cartoon characters, Daisy’s eyeballs needed to be big (like the size of oranges).  I only realized they were visible through her mouth because of two things - 1.) Until now, her eyes weren’t even attached to the rig, and 2.) before I added her teeth, I had no reason to look inside her mouth.

Right now my options seem to be limited.  Aside from the pupils, the eyes don’t have a lot of detail, so a simple lattice deformer doesn’t work that well, and the sculpt deformer pushes and stretches the geometry a little too far, even with the lowest settings.

Of course another alternative would be to swap out the spheres for non-rotating surfaces, that won’t poke through into the mouth -

The only downside to this is that the pupil will need to be a projected map, which could make expanding/contracting the iris difficult.

I’m a professional.  Now, there are some hobbyists out there that do absolutely stunning work, so stunning that it’s almost criminal that they aren’t being paid to do what they do.  In fact, there are plenty out there that could whoop my ass without trying too hard.

A key difference between hobbyists and professionals is experience - a hobbyist would love to make a decent wage off their work, but they most likely lack the experience needed to back up their demands.  A professional, on the other hand, has the experience, and so is in a much better position to negotiate.

I feel like pointing this out because of an odd misunderstanding, that’s been festering in my mind for the past day or so -
I’ve never been asked to do an art-trade before, but I assumed I knew how it worked, just from the wording - two artists exchanging pieces of art, which I also assume are created with roughly equal amounts of effort.  Maybe these artists normally do commissioned work, but agreed to a trade just to make things easier.  Maybe they simply admire each others work, and/or want to make connections.  Maybe they’re both just really nice people.  Whatever the case, I imagine it’s meant to be a reasonable exchange.

But suppose one of them were asked to produce a 3D render, with custom-built characters, full texturing and lighting, etc., while the other does a traditional, pencil & paper sketch…   call me crazy, but that isn’t anywhere near reasonable, it’s insulting.  I don’t care how talented the guy doing the sketch may be, or what their usual rates are.

A sketch can take anywhere from minutes to hours to draw, depending on the complexity.  Maybe a day or more if it’s super-detailed, or requires a lot of thought.
In 3D, posing a character can also take minutes or hours.  Setting up the lights can take hours more, and the actual rendering can take several hours or even days.  And that’s on top of the MONTHS that it can take to build, rig, and texture each character in the scene.

Even if I just happened to already have all the characters and objects built and ready to be posed, the time it would take to produce a decent render would still be more than it takes to produce a sketch.  And using those characters to fill anyone’s request would also depend on the subject matter - if I think the subject is inappropriate, insulting, or just flat-out wrong for my tastes, then I’m not going to do it.

That’s not to say I’m not open to requests, just as long as I don’t have to build anything that I normally wouldn’t.

And commissions…  just be warned that I may charge an industry-standard hourly rate.  And for an image of two custom-built characters interacting in any sort of environment, the final cost could easily be that of a new motorcycle, or even a small car.

I cannot wait until I can afford to upgrade to Maya 2010 (or 11).  The whole time I was weighting the dress, all I could think about was how much better it might be if I had access to nCloth and Kickstand’s StretchMesh.  In fact, once I’m done with Daisy I just might download the 2010 and Stretchmesh demos, just to see what I’m missing…

This is probably as close as I’ll get to doing a nude pinup of Daisy, at least with this version -

Still a ways to go before she’s finished, but at least now she can be posed.

If I can offer one bit of advice to anyone wanting to model a female character -  keep the breasts small, or at least within reason.  The technical problems you’ll run into increase almost exponentially with the breasts’ size, so just save yourself the trouble.

Something I still haven’t quite figured out how best to rig, are toes.  I completely ignored the toes on the first version of Mary, and simply skinned them to the single toe joint of a reverse-foot rig.  I’m not even sure I got around to the toes in any of her revisions, either.
With Daisy, I tried applying IK to each to, then parenting the handles to the reverse rig.  This didn’t quite work as well as I’d hoped, so my next thought was to try spline-IK…  the only reason why I haven’t done that yet is because spline-IK takes longer to set up, and having to do it for ten toes isn’t too appealing.

Then it finally dawned on me - if her shirt is never going to come off, then why should her shoes?   Since I’ve already weighted them, I could revisit the toes later, like if I decided to do a swimsuit version.  For now, though,  they’re a waste of time.
Rather than trying to re-weight the foot to the reverse rig’s toe joint, I’ve opted instead to just leave feet out completely…  though this meant I had to model a pair of socks, to hide the fact that her legs end at the ankles.

I think the major problems with the shirt just might be solved - since getting rid of the holes made the shirt less interesting, I decided to cut them back in.  But since the mesh is otherwise seamless, there’s no worry of it being pulled apart in the front.

I also finally got around to reshaping Daisy’s breasts to fit inside the shirt.  Skinning everything so that the meshes don’t intersect is going to be a pain, though.

I’d only just started skinning Daisy’s uniform, when the issue of her breasts came around again.  Specifically, how best to skin the seam, to keep it from pulling apart.  The solution?  Remove it.

No more holes also means she no longer needs an extra mesh underneath, unless I end up using the new mesh as a wrap deformer…

The holidays are over, and I’m slowly getting back to it -

Her hair patches have been converted to polygons and merged together, so the next task is to fine-tune the shader.  Things will still be a bit slow for a couple more weeks though, since I’m staying in Michigan until the end of the month.

Something that didn’t take too long, was my dropping c++.  After only a few days of going through tutorials, my idea of making a platform game seemed quite a bit more unrealistic that I originally thought.

Thanks to Mel scripting, C++ doesn’t actually seem all that difficult.  But assuming I did manage to learn the language quickly, I’d still have to learn object-oriented programming…  bascially re-learning everything, in a different way.  Then I’d have to learn an API or two, such as SDL, Allegro, or OpenGL.  While experimenting with the platformer functionality and attempting to write an engine, at some point I’d also have to try to write editors for levels and sprite animations.  THEN there would still be the trial and error of making an actual game.

An alternative might be to use something like Torque, or even Game Maker, just to figure out for sure if I want to make a full game, or if I just want to dick around and show off some simple animations.