No. Most of the advancements in rendering today made possible even for relatively normal users to achieve realism ( ie. http://www.maxwellrender.com/ ). Most realistic CGI is the one I haven't seen.
I haven't said it doesn't take talent though, but it's easier to do than ever. Realistic stuff is being made each day by thousands of professionals in various studios around the world, most of which you don't even see because it's 'real'. What used to take a talent pool of ILM or BUF can today be done by a single person (rendering-wise). His work is great though, I wouldn't contest that!
Also there are some things that you just cannot shoot live... In CGI you can change the laws of physics, for example, you can bring people to live that have died centuries ago, or create imaginary world that are limited only by the human imagination - something that has been shown to be immensely capable.
Kneale believes what makes Roman so good is "sheer, raw, talent". For instance, he uses tricks such as reducing the depth of focus to draw attention to certain details, which also reduces the need to carefully reproduce every last blemish on each piece of fruit in view.
I don't think you know what the word "talent" means.
It's difficult to formalize what "realistic CGI" or even "CGI" means. Most of us probably have a loose definition that's something like "the rendered output of 3D models that looks like something familiar in the real world," but that's insufficient.
One could take 100 bitmap frames of video from a camcorder, then use a computer to generate 100 identical or nearly identical bitmap frames. This would be realistic by definition, but not very impressive. Even more absurdly, one could use a computer to generate some abstract visualization, then construct a nearly identical scene in the real world and claim the CGI is realistic.
And yet, CGI that uses 3D scans of objects, motion capture, or scanned/photographed textures probably fits our loose internal definition.
CG has always been and will always be a hack I think - debating the best CG is kind of like debating the 'most realistic miniature'. The more real elements you include (ie. real sunlight, real objects, photographed background plates) the less there is to offend your eye.
To me, the tools are getting so good (GI, HDRI lighting, MoCap) that 'seamless/not seamless' is soon going to be the only judging criteria. Of course along with visual aesthetic which will always occur in film whether you're using a physical camera or virtual.
Somewhat disappointing. The subject matter was not a difficult one. Not to say it wasn't well done, but it was not as challenging of a subject as a person or animal would be.
I think Sofronis Efstathiou sums it up nicely in the article. Nice job, but nothing groundbreaking.
I looked at the girl spinning around and was trying to look for errors but then the sony logo appeared... and I was looking at an advert... well played sony ;)
The artist behind this, Alex Roman, has produced a 12-minute CG piece (the "Third and the Seventh", briefly mentioned in the article) that is absolutely amazing:
He apparently quit his job and worked fulltime for a year to create it. It is, I think, the most amazing work of art I've seen created by an artist in my lifetime. I've watched it perhaps 20 times. (Sorry for the hype, I find it impossible to talk about this any other way.)
17 comments
[ 3.0 ms ] story [ 63.0 ms ] threadIf you're still doubting, just checkout Third and Seventh, it's amazing.
I don't think you know what the word "talent" means.
One could take 100 bitmap frames of video from a camcorder, then use a computer to generate 100 identical or nearly identical bitmap frames. This would be realistic by definition, but not very impressive. Even more absurdly, one could use a computer to generate some abstract visualization, then construct a nearly identical scene in the real world and claim the CGI is realistic.
And yet, CGI that uses 3D scans of objects, motion capture, or scanned/photographed textures probably fits our loose internal definition.
To me, the tools are getting so good (GI, HDRI lighting, MoCap) that 'seamless/not seamless' is soon going to be the only judging criteria. Of course along with visual aesthetic which will always occur in film whether you're using a physical camera or virtual.
I think Sofronis Efstathiou sums it up nicely in the article. Nice job, but nothing groundbreaking.
Interesting how the most advanced shading algorithms are almost able to describe the neurological bases of what we PERCEIVE as real.
http://vimeo.com/7809605
He apparently quit his job and worked fulltime for a year to create it. It is, I think, the most amazing work of art I've seen created by an artist in my lifetime. I've watched it perhaps 20 times. (Sorry for the hype, I find it impossible to talk about this any other way.)