igl vs. ogl default PFTEX_INTERNAL_FORMAT

New Message Reply Date view Thread view Subject view Author view

David Florek (dflorek++at++nvl.army.mil)
Mon, 9 Sep 96 19:31:42 EDT


I was checking to see that 'pfconv' (beta code to convert among file for-
mats, available along with 'libpfpfb') would convert a '.obj' file to a
'.pfb' file the same way, whether compiled as an IGL or OGL Performer app.

It does not. To make my life a little easier, I converted the '.obj' to
'.pfa' format instead, and found that the difference was a single token
over a 32MB file, specifically the value for the texture's internal format:

        dflorek++at++toltec[135]: head -20 testground_igl.pfa | tail -6
        #---------------------- Texture 0
        "ground64_512.int"
--> 6 1 1 1 1
        0x8010 0x4000 0x4000 0x4000
        1 1 1
        0.000000 0.000000 0.000000 0.000000
        dflorek++at++toltec[136]: head -20 testground_ogl.pfa | tail -6
        #---------------------- Texture 0
        "ground64_512.int"
--> 4 1 1 1 1
        0x8010 0x4000 0x4000 0x4000
        1 1 1
        0.000000 0.000000 0.000000 0.000000
        dflorek++at++toltec[137]:

This corresponds to:

        IGL: pfTexFormat(tex, PFTEX_INTERNAL_FORMAT, PFTEX_I_12A_4);
        OGL: pfTexFormat(tex, PFTEX_INTERNAL_FORMAT, PFTEX_IA_8);

This appears to be tied to the use of the '.int' texture image format,
as the '.inta' format generates 4 (PFTEX_IA_8) under both IGL and OGL.

Is this in fact a difference in default behavior between IGL and OGL
Performer libraries, when using '.int' formats? Also, will this affect
the appearance of the texture, given that it has no explicit alpha
component? Om this case, the '.int' file in question is 8 bits per
pixel, which is converted to 16 bits of texture information; is
PFTEX_I_12A_4 interpreted as 4 bits of intensity information and 12
bits of alpha, where PFTEX_IA_8 is interpreted as 8 bits of each)?
Would it make more sense for '.int' files to be converted to textures
with PFTEX_I_16 (presumably 16 bits of intensity, no alpha) format?
This might aid in preserving quality during texture filtering....

Any thoughts?

Dave

-----------------------------------------------------------------------------
  David W. Florek EOSoft, Inc. Phone: 703 / 704 - 1767
  Senior Engineer NVESD Night Vision Lab Email: dflorek++at++nvl.army.mil
  Computer Graphics Ft. Belvoir, VA 22060
-----------------------------------------------------------------------------
=======================================================================
List Archives, FAQ, FTP: http://www.sgi.com/Technology/Performer/
            Submissions: info-performer++at++sgi.com
        Admin. requests: info-performer-request++at++sgi.com


New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:53:31 PDT

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