Re: New Enterprise: Model or CG?
It's the "simulated chaos" of CGI smoke, for instance, that makes it less convincing than the REAL chaos of real smoke. You can show a single snapshot and it'll look just fine, but the swirling and billowing effects, the diffusion and dispersion, never QUITE get there. At least not yet. Eventually, more complicated algorithms will be derived to make it MORE believable, but something run in a controlled, repeatable numerical simulation can never QUITE match the behavior of something in an uncontrolled, unrepeatable, random and chaotic situation.
To compensate, you have an artist who knows the subject matter and who manually tweaks each image, or the whole animation, in an attempt to reduce the "obviousness" of the CGI version of these details. Good CGI artists can do that effectively, but it's NOT easy and it's not something that can be reduced to an algorithm (well, not yet).
To be fair, Vektor, still images are much easier to make believable than animated stuff, for the same reason I mentioned above... chaos.Vektor said:
For those of you who think you can identify the CG image vs. the real image every time, I suggest you give this a try. I am a 3D artist and I consider myself to have a pretty discerning eye and I only scored 6 out of 10.
What's ironic is that some of the realism people associate with physical scale models has to do with visual characteristics that are not necessarily valid for real, full-size objects. In other words, much of that so-called "realism" is nothing more than what we are used to seeing after decades of physical model work. In various technical ways, a properly rendered CG image might actually be more realistic, but only if reality rather than a miniature--superbly crafted as it may be--is your benchmark.
It's the "simulated chaos" of CGI smoke, for instance, that makes it less convincing than the REAL chaos of real smoke. You can show a single snapshot and it'll look just fine, but the swirling and billowing effects, the diffusion and dispersion, never QUITE get there. At least not yet. Eventually, more complicated algorithms will be derived to make it MORE believable, but something run in a controlled, repeatable numerical simulation can never QUITE match the behavior of something in an uncontrolled, unrepeatable, random and chaotic situation.
To compensate, you have an artist who knows the subject matter and who manually tweaks each image, or the whole animation, in an attempt to reduce the "obviousness" of the CGI version of these details. Good CGI artists can do that effectively, but it's NOT easy and it's not something that can be reduced to an algorithm (well, not yet).

