he IWH Inquirer goes in-depth with editorial articles on topics, issues, news and reviews with a focus on health, history, science, literature, technology, philosophy and organic gardening.

Primary Format :


Also Listed as:

User Tags:

User Votes:

RSS Feed
Website

Visit Methings.com for the most recent listings of:

The great divide in 3D CGI between real-time rendering and pre-rendering

The great divide in 3D CGI between real-time rendering and pre-rendering

Play Now -->

DATE : Wed, 29 Oct 2008 07:44:58 PDT
Entered in Database : 2008-10-29 14:44:58
length :
Link to the Show / Show Notes

Listen Letter to the Editor Financial Times ref: Alan Cane’s article “Will we be able to tell reality from artificial imagery?” FT Europe 30th Jan 2008. Sir, Whilst Alan Cane poses some interesting questions about the narrowing divide between real and artificial imagery and how this might affect our future society, in his article "Will we be able to tell reality from artificial imagery". I think he underestimates the fundamental divide between the real-time 3d rendering processes e.g. computer games and the pre-rendered 3d movie or clip making process e.g. Bee Movie. With the former the GPU is indeed of prime significance it being required to render a 3d scene in real-time, and its rapid development by Nvidia and ATI has indeed spurred the game industry to create ever-increasing real-time realism in their games. However in pre-rendered 3d movie making or special effects the prime job of the GPU is only to aid the 3d operator within a 3d application to see what they are doing at a very basic visual level (wire frame or simple shaded views) not for the main rendering process. It is possible to tie your desktop GPU into the render process with some 3d creation software packages but generally main renders are sent to render farms and not rendered with the 3d operator’s computer. (Typically each scene is first broken down into elements and divided amongst the operators each focused on a specific element e.g. lighting, modelling/sculpting, hair and fur, etc. These elements are then combined in a render pipeline and served to a render farm for CPU processing. Render farms generally are stacked server systems designed for one thing, combined CPU processing power, the ability to crunch numbers quickly. Render farms then render each element and send it onward for compositing or putting it all back together for a final render that combines all the elements into a single two dimensional frame sequence. The movie is then colour graded etc. just like any movie and then it's ready for conversion and distribution through the various mediums like film for theatres, DVD's etc.) My point being that what you see on the big or small screen and its increasing quality and realism is pre-rendered and far cry from a real-time rendering computer game. It's a divide that is the difference between having a render farm of 10,000 CPU's for a single high quality multi element frame or a single GPU. As 3d creation software increases the quality of the realism that can be achieved in pre-rendered films, it comes at an ever increasing price of the processing power required to feasibly render an entire movie in a given production time frame. A simple way to take stock of the divide is to play the latest game on the newest PC and GPU you can possibly find, the difference between the game opening pre-rendered movie clip and the actual game is that exact divide. This is of course not say the divide will never be bridged that would seem inevitable, (and as a 3d Director that day we can render movie quality scenes in real-time will be a most amazing day indeed) with the incredible advances in computer technology, but I suspect it is further beyond the horizon than Alan Cane or I might hope for. Best regards Ivor W. Hartmann

Play in your Iphone