On Dec 7, 2007 2:28 AM, Kevin Gillette <[EMAIL PROTECTED]> wrote:
> i think that's a bit overkill -- if we're doing splatting, the point is that
> no one texel is easily distinguishable where the splatting occurs, thus if
> we implement splatting, we *should* use simple POT dimensions for our
> textures.

We will still need border to deal with linear and anisotropic filtering.

>  i agree that removing the rotational/mirroable seamless
> requirement would be a major benefit to artists, which could focus on higher
> quality and better looking textures that need only be tileably seamless.

Making it easier for people who want to work on the game should be
very high priority.

> if there is interest in the splatting idea, i think we should plan a
> deprecation schedule

I and Dennis were thinking of automatically updating old maps to the
new scheme. The current way maps are drawn can be expressed as a
subset of the features available with splatted textures.

> furthermore, i believe it would be prudent to consider a redinition of
> mip-map levels by "pixel resolution" rather than "pixel dimension", so that
> i could, for example, design a 2048x2048 "tile" that has a 128x128 "pixel
> resolution" -- that way, a single terrain type would just have a large
> texture repeat interval.  this would conceptually be somewhere between
> warzone's current discrete tiling system and megatexture <
> http://en.wikipedia.org/wiki/Megatexture>. of course, for such a system to
> work well, a "smart" method of texture packing and binding (a 2048x2048
> texture would get its own bind, probably) would need to be proposed, as
> would a method for detecting graphics processing capabilities and scaling
> the amount of maximum detail to be appropriate to the machine warzone is
> running on.

I am not sure what exactly you want here and why.

  - Per

Warzone-dev mailing list

Reply via email to