Friday, November 04, 2005

The Biff Diaries


Let’s start shall we?

Firstly the name of the blog? I am currently, at the time of writing, building a 3D model of a Bristol F2B aircraft. The Bristol F2B was a WW1 and post WW1, 2 seat fighter aircraft. It was known as The Bristol Fighter or ‘Biff’, by its crews in WW1 and post WW1 it acquired the name ‘Brisfit’. I will provide more information and explain why I chose to model this aircraft later in a future post.

Anyway Biff it is then.

What’s this about? Well firstly the term ‘3D vet’ is possibly a bit of a misnomer. I certainly have been dabbling with 3D graphic modeling and rendering applications since 1988 and did have a few renderings and digital paintings published in Amiga magazines in the early 1990’s. However I am hardly a pro and do not represent myself as such. I possibly have a bit more of a background in CG and 3D modeling than some of the guys currently modeling aircraft, who have come into this field through their enthusiasm for flight simulation and who have progressed through a variety of applications, from software dedicated to constructing aircraft only, to the likes of gmax and even 3DS Max. I am in no way belittling these modellers. Indeed there are many fine modellers out there whom I greatly admire and who can and do teach me a great deal. I guess what I am trying to say is that coming from a different background I have an approach, methods and philosophies that differs from those who have a more conventional, traditional approach and who have been modeling low-poly aircraft for a number of years. I would possibly have a different approach from those who have studied low-poly modeling in art schools in recent years. I am definitely not bound by tradition. Vive La Difference!

What’s in a poly then?

One question that people invariably ask modellers is ‘how many polys’ does the aircraft have? Usually they mean triangles – there is a difference – more on that in a future episode.

In modeling for the rendering applications of the late 1980’s and early 1990’s one had to be cognisant of polycounts back then too. The reason being, the more vertices and triangles, the longer a scene took to render and some could take days. Computers also did not have much memory in these days either – 512k was big RAM indeed. Modellers trimmed the polycounts by only modeling what was necessary and even deleting any polys that could not be seen. Any models that were likely to be close to the camera in the scene were usually modelled in great detail. Fast forward a few years to the present day, not only do we have high speed CPU’s we now also have GPU’s [graphics processing units]. I’m not technically qualified and am not going to embarrass myself by trying to explain how a GPU works. For an overview check out Wikipedia. Just know that GPU’s process a lot of graphical data, including 3D data, very quickly. Thus today we can do in real time what would have taken several minutes or longer to render in the early 1990’s. I can recall creating animated fly throughs of the Grand Canyon, back in 1991, using VistaPro on one of my Amiga computers, that would have taken probably half an hour to render. These days I can fly through a similar, if not higher resolution landscape in real-time in Flight Simulator 2004 [also FS2002]. Computer power has come a long way in a comparatively short period of time.

However we cannot just create any old model and expect it to be moved around in our game, in real time, by the GPU. [It’s fun experimenting though ;) ] Not only do we have to be cognisant of polygon or vertex counts we have to design our models efficiently to be effectively processed by the GPU. [I have only just learned this myself in the last 10 months myself and am still learning.] In any game there is much happening and many 3D objects moving around in what could be considered to be quite rich 3D environments. Thus 3d objects require polygon budgets. In a professional environment such as a game in development, the polygon budgets for each object would be determined by the development team, I guess most likely an art director working in conjunction with software engineers. In a professional setting then, especially in game development, 3D modelers have to be very creative to make the most of the polygon and texture budgets they are allowed. These modelers have to be very creative indeed. The term ‘creative modeling’ is starting to replace the term ‘low polygon’ modeling. Low polygon modeling is a relative definition anyway. As technology advances the high polygon resolution polygon models of today, become the low resolution models of the future.

As stated in the intro I do not work professionally in this field at this time. Whether I will work in the commercial [payware] side of flight simulation add-ons remains to be seen.

The freeware side of things is quite different. In one sense we have more freedom than our professional counterparts. Indeed sometimes I think freeware='free for all' when it comes to design considerations. Many payware modelers out there also appear to adopt a liberal definition of low poly modeling.

At the end of the day it is a question of balance. There is no right or wrong way. There are a few ‘do’s and don’ts’, but there is no absolute approach to modeling a 3D aircraft for a real time rendering system.

We’ll leave it there for the time being…. To be continued.


[Note: Edited for grammer corrections 10 November 2005]