View Single Post
Old August 18 2008, 02:17 PM   #618
Cary L. Brown
Rear Admiral
Location: Austin, Texas
Re: USS Grandeur - One... More... Time!

Vektor wrote: View Post
Well, I’ve hit a bit of a snag. I’ve been working on detailing the underside of the saucer and in the process I’ve discovered that I didn’t plan ahead very well in structural terms. Both the top and bottom sides have a number of features that are supposed to align with the implied radial and concentric structural members of the saucer, but I have never previously worked out exactly what the configuration of those structural members are. As a consequence, the details on the upper and lower saucer surfaces failed to magically line up as I had been hoping they would since I first realized it could be a problem.

Now, I could fudge things just enough to connect the dots, so to speak, and hope that nobody notices that certain details are a little off or slightly askew, but I have a really hard time cutting those kinds of corners after all the work I’ve put into this thing. The alternative is to basically erase most of the radial details on the saucer and rebuild them in a way that lines up with a unified internal structure. Some of the larger parts and pieces might have to be adjusted as well. Ultimately, the amount of work required is what scientists and engineers like to call “non-trivial.” It’s a significant step backward in order to move forward, which is the kind of thing I was really hoping to avoid at this late date. I swear, sometimes I wonder if I’m gonna have to leave this project to my descendants to finish.

Anyway, I have pretty much made up my mind that if it’s worth doing then it’s worth doing right, so I started the detail removal process last night. In most cases, I can go back to earlier iterations of the model and steal the pieces I need, but I can’t do it wholesale because there are a lot of other modifications of the same parts that I need to keep. It’s actually going more quickly than I expected, if I can just figure out a structural configuration that will work with all of the design requirements I’m trying to satisfy.

One bit of good fortune in all of this is that it gives me the opportunity to fix something else that has been bugging me. I’ve been thinking for a while that the gridlines were too wide and too deep for a ship this size, to the point where they actually detract from its sense of scale. Since I’m redoing most of the gridlines anyway, I figure I might as well scale them down a little bit. Of course, that means I will have to redo ALL of the gridlines, including the ones on the warp nacelles, but in for a penny, in for a pound.

At least you will all know what to expect in the next set of WIP images, but don’t count on seeing them before next weekend.
Sorry to hear that you've gotta take a step back... always sucks, huh?

You're a much better rendering-guy than I am (and probably than I'll ever be) but I'm a pretty damned good engineer. I've been through this sort of thing on many design programs. And as a result, I've developed a basic "standard practice" that I use on everything.

Before I make any physical elements of any kind, I create a very detailed "skeleton" made up of datum planes, datum axes, datum points, and sketches (never intended to be seen in the final model). I then build my geometry using those as references. That avoids the very problem you're encountering now.

If dealing with habitable spaces, I generally create datum planes to reflect the floor and ceiling of each individual deck. I create sketches (on individual datum planes) reflecting basic structural elements. It's a long and involved process but it's invaluable... because you already have geometry (which will never be visible in the final model) which you can reference as you create every downstream feature.

The problem with this approach is that you lose the ability to "massage" certain elements... you're pretty much tied into the general layout you create up-front. But if you do that layout in a parametric fashion, you can then "tweak" the layout in those datums and sketches and everything else downstream will update "automagically." (Or "most everything" anyway.)

Just tossing in my own 2cents... you obviously have a workflow that works pretty well for you, but what I'm suggesting might avoid the sort of problems you're encountering now. Take it or leave it.
Cary L. Brown is offline   Reply With Quote