Re: Virtual ClipTexture questions...

New Message Reply Date view Thread view Subject view Author view

Don Hatch (hatch++at++hell.engr.sgi.com)
Wed, 9 Dec 1998 17:30:56 -0800


On Dec 10, 9:02am, Matthew Whitaker wrote:

> 1) I have read /usr/share/Performer/doc/clipmap/HowToDoVirtual.html
> and /usr/share/Performer/doc/clipmap/IRClipmapBugs.html.
>
> These documents are dated 06/10/97 & 11/12/97 respectively, and appear
> to be draft versions. Do more up to date versions exist?

Those are the most recent versions (Oct 6 and Dec 11, 1997).
HowToDoVirtual.html never got polished but the content should be correct.

>
> 2) My understanding is that virtual clipmaps of up to 24 levels (8M*8M)
> are directly supported by InfiniteReality/Performer, but this can be
> increased to 31 levels (1G*1G) by applying the workaround described in
> section 1.5 of the bugs document. Is this correct? If so, is it an
> effective approach, or is it a more high-risk strategy, of limited
> benefit? I am very interested to know of previous experiences in this
> area.

I don't think you can use clip textures of more than 24 levels
(maybe even 22 or 23) without addressing this precision problem.
The suggested solution works well; its main drawback is
that it can be expensive and/or awkward to translate all the texture
coordinates of all the vertices every frame.
You might be able to get around this expense by playing games with
a texture matrix to do the translation
(expressing the tex coords in some suitable fixed local
coordinate system) instead of translating all the tex coords individually,
but I've never actually tried that so I don't know what problems
you might run into.
Also if you're using a pfTexGen to generate the texture coords
(which might be feasible since you're only doing Australia and not the whole
globe) then you need only change the tex gen parameters, which would
be cheap (but you'd have to work out the details of doing this
in a frame-accurate manner; I've never tried this either).

Don

-- 
Don Hatch  hatch++at++sgi.com  (650) 933-5150  Silicon Graphics, Inc.

New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2.0b2 on Wed Dec 09 1998 - 17:31:01 PST

This message has been cleansed for anti-spam protection. Replace '++at++' in any mail addresses with the '@' symbol.