Richard Kain
|
|
« Reply #4 on: January 04, 2019, 10:52:45 AM » |
|
Well, in this particular case, I can see several areas where the models in question weren't actually optimized for use in-game. There are multiple details on each model that don't contribute to the object's silhouette, and thus would have been better implemented through a normal map. The excessive decoration on the front center of the chair/stool would be an example, as would the individual loops on the hangman's noose.
Back in the day, optimization of a model for in-game use was a craft unto itself. I've seen a lot of fantastic artists strip complex models down to their bare essentials. I've seen incredible works done with nothing but 800 polygons and a 256x256 texture without any shaders and an 8-bit palette. So when I see models like these and you tell me that they have been optimized for in-game use, I can't help but purse my lips a little and perhaps raise an eyebrow. Sure, by modern standards, they MIGHT pass muster. And the power of modern GPUs can eat up and spit out polygons as older systems could never hope to. (making polycounts far less of a concern in modern rendering) But old habits die hard, and someone trained to optimize models couldn't help but consider this level of optimization to be a little on the weak side.
Just so we're clear, I don't feel these models are bad. Quite the opposite, these strike me as pretty good works. Some solid, respectable prop and environment modeling, with some decently applied textures and material work. Good stuff, the artist has every right to be pleased with their work. If I saw any of these models in-game I would be just fine with their effect on the experience. Most of my above comments are just the product of a training that treats every triangle as precious. It's nit-picking, from someone with a pedantic approach to optimized modeling, pure and simple.
|