Casually Hardcore devlog

Lost in Translation

October 2nd, 2008

It’s been a tough month - I’ve been ill or away at weddings or random conferences. At least I’ve been awarded a Telefilm extension on the project!

Anyhow, I’ve had to let go some junior/intermediate artists unfortunately. There’s a well known axiom called “Cheap, Fast or Good - pick two” and I’ve encountered it.

I assumed most intermediate artists were pretty similar - given a detailed design (i.e concept art / model sheet) I could pick any 3D modeler and have a good final result. With programmers, I do not think this way. I prefer the most senior programmer over a dozen junior programmers, but with managing artists, I’m new to i. Well, junior/intermediate 3D modelers are typically cheap and try to be fast, but the end result is not very good.

The problem is of translation - between design and 3D model. I’ll provide an example from “The Art of Game Characters” book. It has pages upon pages of concept art (and sometimes their final result in game).

Original:

Final result:

This is a pretty extreme example as the woman looks nothing like the design and the general is not even fat! The original designs exudes personality and had a distinctive style, but the final models are dull and generic. The modeler (or modeling supervisor) decided to go in a different direction or did not have the skill to execute the original design (which is quite challenging). I suspect the concepts were deemed too difficult and beyond the skill of the modelers so it was dumb down to traditional human proportions.

For The Supers, I performed a test - I gave 5 modelers - from junior to senior - the same building design and told them to try to match it but they were free to add to it if they feel it makes the model better.

The result: 5 completely different models were returned - varying polygon counts, varying artistic consistency, closeness to design, etc. Being a non-artist, this really surprised me! The original (2D) design was pretty well-defined to me, but each modeler took a slightly different direction… and best models were by senior artists who have that “artistic sense” lacking in the the ones by juniors. How did I define best? Lowest poly count (technical), pleases everybody who sees it (artistic), cleanest geometry (technical), matches the original design (artistic). With the non-best models, I and many others would say “there is something wrong” but be unable to pinpoint it.

So I let go all junior/intermediate artists and kept only the senior ones on-board. The past few months, I had major problems with the junior/intermediate artists failing to match the designs which I originally attributed to poor communication between the art director and them, but my tests proved me wrong. The junior/intermediate artists also botched rigging and other 3D tasks rather badly so I began to clue in earlier, but blamed other factors (again, poor communication).

Seniors are more expensive, but at least their work doesn’t need to be re-done! So I’ve chosen Fast and Good over Cheap.

Entry Filed under: Game Development