Back then modeling with NURBS was the norm, not the exception. SI3D didn't have any texture unfolding projection methods (only planar, cylindrical or spherical). Creating custom UV layouts was possible, but a real pain as the UV layout editor was designed for low resolution polygon meshes used in early 3D games, not movie quality high resolution geometry, and you could only manipulate one UV at a time. Graphics hardware capable of displaying textures on geometry still cost a premium, so workflow often included jumping out to the renderer to check in on your progress from time to time. To create custom UV layouts required a huge amount of manual labor moving points individually, or you'd resort to clever repurposing of other tools such as RenderMap. But any respectable studio with a decent budget would've exported the geometry to a 3D paint program like Amazon Paint or DNA's Flesh for that work. Working with polygons pre-2000(ish) was a real chore due to lack of decent hardware acceleration and tools. Finally, don’t forget SI3D was limited to 60,000 triangles per scene. You could increase that limit by modifying an environment variable, but doing so ran the risk of corrupting memory and other issues with the graphics hardware. When you have to animate scenes containing many bugs like in Starship troopers, working with NURBS allowed a much higher number of bugs to appear onscreen before you hit those limits as you could reduce the bugs' geometry to 1x1 interpolation in U and V. That's another prime reason why NURBS were used so much back in the day.
NURBS had significantly better system performance during animation playback compared to polygons. Still does today. A lot of it has to do with the geometry description as NURBS requires only a handful of control point positions as input even for large and complex shapes. The rest is derived from interpolation which can be performed in hardware and highly optimized without much fuss as the geometry has a very well defined organization which is highly scalable and predictable. Graphics libraries, like OpenGL, can use triangle strips and other optimization methods to draw large amounts of geometry quickly with minimal overhead. Polygons (and subDs) are arbitrary and often cannot take advantage of those optimizations. XSI has many core features optimized for working with NURBS such as auto-LOD control when manipulating the camera, manipulating the geometry, or performing animation playback. While many people, especially today's generation of artists, are deeply against using NURBS at all, I think that mentality is a big mistake. A lot of that thinking has to do with not properly learning what NURBS are or having a decent environment to work with them. NURBS aren't meant for every modeling or animation task, but for some they provide elegant solutions which do not exist (or do not compute very well) for other types of geometry. Matt Date: Sat, 10 Feb 2018 20:05:38 +0100 From: "Sven Constable" <sixsi_l...@imagefront.de> Subject: RE: Friday Flashback #330 To: "'Official Softimage Users Mailing List. I wonder if they designed the bugs *a bit* with NURBS modeling in mind. I once modelled and rigged that bug in XSI as kind of a training session when I switched from SI3D to XSI. Except for the lower part of the torso maybe, there are no parts with singularities or other patch modeling difficulties. Pretty much all parts are rigid with easy UV topology. Ball joints and no enveloping for the most parts if not all. Maybe it were just simple parent child hirarchies when they rigged it. I don't know it of course, but I would guess it was relatively fast to animate in terms of performance. Sven ------ Softimage Mailing List. To unsubscribe, send a mail to softimage-requ...@listproc.autodesk.com with "unsubscribe" in the subject, and reply to confirm.
