On 25/01/2013 02:31 PM, scott wrote:
>>> Seriously though, if your goal is to create a bitmap representation of
>>> the math curve, what difference does it make if you check every pixel in
>>> the image to see if it lies on the curve, or you convert the curve into
>>> a list of pixel-sized straight lines and then colour those pixels?
>>> Different maths but same result.
>>
>> Sure. But nobody does that yet.
>
> It's pretty much there though, not like the 90's where you got triangles
> 100 pixels wide on supposedly curved shapes. Arguably you're never going
> to get to the exact situation where every triangle is 1 pixel wide
> because nobody will notice the improvement over a small amount of
> approximation, and it's thus better to spend the GPU time on other effects.

It depends. If you're doing something like hair simulation, hairs with 
right-angle kinks in them are very, very noticeable. If it's something 
like a distant rock face or something, you're probably not paying much 
attention to it, and the rock texturing will probably hide it, mostly...

Post a reply to this message