> In some mid-1990's 3D hardware, texture coordinate is not represented by float-point,
> but by integer, like "pixel-index" in those 2D graphic hardware (namco s22 is a
> typical example).
>
> If we emulate them using current float-point poly.c, by the (coord +
> 0.5)/size_of_texture method to convert to float-point, some graphic glitch may happen
> at polygon borders, because the range of texture-coordinate is from [0, size-1], not
> [0, size].
Is that not similar to/exactly what M**G wrote once on his blog, regarding N64 textures?
Check 'p=39' & 'p=113' on his blog, because the Canadian Administration has even the links censored here.
|
Index 
Flat 
Re: About integer pixel-based texture coordinate(mathematic issue)
Reply
About integer pixel-based texture coordinate(mathematic issue)
