bicubic texture mode

New Message Reply Date view Thread view Subject view Author view

Mike Goslin (goslin++at++cs.unc.edu)
Thu, 29 Feb 1996 15:30:40 -0500 (EST)


We've looked into the PFTEX_BICUBIC problem I asked for advice on a week
or so ago. It appears that the default bicubic filter is not correct for
our texture maps. When we replace the default filter with a notch filter
using TX_BICUBIC_FILTER, 1.5, -0.25 as part of the props argument to
texdef2d in GL, everything looks great and the artifacts are smoothed
away. Unfortunately, there is no clear way to set this filter value in
Performer. I've tried making a GL texdef2 call from my Performer app to
try and initialize the filter, and this didn't seem to work. Is there a
better solution or will I need to define and apply all my textures using
GL callbacks in order to use this bicubic filter mode? I hope I've
overlooked some easy solution to this.

Thanks,

Mike
-----------------------------------------------------------------------------

    Mike Goslin (goslin++at++cs.unc.edu)
    The University of North Carolina at Chapel Hill
    Department of Computer Science office: (919)962-1719
    CB# 3175 Sitterson Hall fax: (919)962-1799
    Chapel Hill, North Carolina 27599 home: (919)942-0043

-----------------------------------------------------------------------------

---------- Forwarded message ----------
Date: Wed, 21 Feb 1996 12:14:45 -0500 (EST)
From: Mike Goslin <goslin++at++cs.unc.edu>
To: info-performer++at++sgi.sgi.com
Subject: bicubic texture mode

I'm getting some unusual results when using TX_BICUBIC mode for
texturing. This is really a GL problem, because I have the same problem
in GL as I get in Performer. Could someone refer me to the appropriate
person on this (or pass this on to him/her)?

Here's the problem: I'm using TX_BICUBIC mode to filter a texture map
and I'm seeing artifacts that are texel-related. The smoothness of the
image falls somewhere between TX_POINT and TX_BILINEAR in quality. The
image itself is a texture map generated by sampling a bicubic lighting
function for a particular surface, so it is really an intensity
gradient. The image is 32x32, and I'm using 4 components with 16 bits
per component (TX_INTERNAL_FORMAT is TX_RGBA_12).

TX_BICUBIC seems to work okay for smaller textures (4x4), but I figured
this might be related to the kernel size for the bicubic filter because
it doesn't need to be shifted to filter the entire image.

Thanks,

Mike
-----------------------------------------------------------------------------

    Mike Goslin (goslin++at++cs.unc.edu)
    The University of North Carolina at Chapel Hill
    Department of Computer Science office: (919)962-1719
    CB# 3175 Sitterson Hall fax: (919)962-1799
    Chapel Hill, North Carolina 27599 home: (919)942-0043

-----------------------------------------------------------------------------


New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:52:29 PDT

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