402accessdenied

Ok, so it’s been over a month since my last post. Yeah, yeah, I know… it’s like my posting schedule when I was in college all over again.

Whatever.

Apparently, the quiz I posted in my last post about real vs. CG images wasn’t as hard as I thought it would be. I even tried to trick you. Images 1, 2 and 5 were CG, images 3 and 4 were photographs. The way I cheated was to use HDR images for the photos – images of the same scene that are taken at different exposure settings and then combined to form a composite image. I did this because traditional photography (especially with film) has difficulty with scenes that have a wide range of brightness – usually they are either skewed toward dark, light, or straddle the middle and lack contrast. CG doesn’t have this constraint, and this difference is one of the primary ways of easily differentiating between real and CG. Also (especially in the case of the river photo), HDR images often have interesting colorization, making them appear unnatural.

Purists might argue that calling the HDR images “photographs”  was misleading. It probably was. However, they were in fact images of the real world, and were not the result of computer rendering.

The impressive thing to me was the fact that the CG images I presented were generated in real time – not pre-rendered. Pre-rendering can achieve amazing quality, with results that can be deceptively life like. Up to this point, real time rendered images were significantly lacking in realism. The images of the girl were screen shots of a digital model of model Adrianne Curry created by NVidia to demonstrate the capabilities of their new graphics processor. I have to say, I’m impressed. While not perfect, the progress is impressive. At this rate, it might not be long before scenarios such as S1m0ne (imdb.com) become a reality.