zlacker

> a friend of mine was on the design team and he assured me that NURBs were going to be the dominant rendering model since you could do a sphere with just 6 of them, whereas triangles needed like 50 and it still looked like crap. But Nvidia was so tight fisted with development details and all their "secret sauce" none of my programs ever worked on it.

(1) Someone designs something clearly superior to other technology on the market.

(2) They reason that they have a market advantage because it's superior and they're worried that people will just copy it, so they hold it close to the chest.

(3) Inferior technologies are free or cheap and easy to copy so they win out.

(4) We get crap.

... or the alternate scenario:

(1) Someone designs something clearly superior to other technology on the market.

(2) They understand that only things that are more open and unencumbered win out, so they release it liberally.

(3) Large corporations take their work and outcompete them with superior marketing.

(4) We get superior technology, the original inventors get screwed.

So either free wins and we lose or free wins and the developers lose.

Is there a scenario where the original inventors get a good deal and we get good technology in the end?

>>api+(OP)
This is what patents are for, but in the real world that gets complicated too.

>>api+(OP)
Good point. In this context, it wasn't even superior though, at least not in the long run. Memory got bigger so that storing more triangles wasn't a problem anymore, it's more about computational resources. There, NURBS are only clearly better for very smooth surfaces (like the mentioned sphere), which are rare in natural shapes. For everything else, you get more details per FLOP by just using more triangles, which is where the industry went.