From: Christian Skluzacek (c.skluzacek++at++fokkerspace.nl)
Date: 01/12/2000 08:41:56
Thanks Allen and Marcin. Upgrading to 6.5.6m seems to have solved my problems
and now I can happily go on using shared libs.
However, I thought I saw something in the release notes saying that textures
with an alpha component do not cast shadows correctly (which I have
verified). Will this be fixed in the (near) future?
Chris
Allan Schaffer wrote:
> [What OS version are you running? 'uname -R']
Was 6.5.5m, have now upgraded to 6.5.6m.
> Multi-part handwaving answer:
>
> 1. The static Performer libraries are version 2.2; (2.2.0)
>
> 2. I believe that the multipass-lighting features (such as shadows)
> had bugs introduced around Performer 2.2.2 and have been steadily
> rearchitected since then, currently version 2.2.6.
>
> 2a. More things are fixed for the version which will ship with
> IRIX 6.5.7.
>
> So if you're running 6.5.1 to 6.5.4 I bet the reason why the static
> libraries work but DSOs don't is because the DSO versions you have
> contain regressions in the shadow code. As much of that has been
> redone you might try upgrading to 6.5.6 which contains the most
> up-to-date DSOs -- 6.5.6 is available for web download from
> http://support.sgi.com/ or on CD.
>
> If shadows in 6.5.6 still have problem please let us know more..
>
> Allan
Marcin Romaszewicz wrote:
> Hi Chris,
>
> I've rewritten the performer multipass to take care of several problems
> and introduced a change in behavior to pfLightSources (they behave
> correctly now). Before the rewrite, the multipass code was not
> transforming the light source direction vector correctly into the
> performer coordinate space. This problem has been fixed, but the result is
> that the default direction of the light sources is now different. The
> default has always been the vector (0, 0, -1), but due to the bug, it
> behaved as though it was (0, 1, 0). In your code, explicitly specify the
> light source direction as (0, 1, 0) to get the old behavior with the
> dso's. The changes to the multipass implementation are detailed in the
> performer release notes (I think it was version 2.2.3 or 2.2.4). In our
> quarterly patches, we only ship the dso's, so your dynamic and static
> libraries are most likely using different versions of the algorithm.
>
> -- Marcin
-- ------------------------------------------------------------------------ Christian Skluzacek | FokkerSpace B.V. Software Engineer | Newtonweg 1 Program Operations - Simulation | 2333 CP Leiden +31 71 524 5718 | The Netherlands mailto:c.skluzacek++at++fokkerspace.nl | http://www.fokkerspace.nl ------------------------------------------------------------------------
This archive was generated by hypermail 2b29 : Wed Jan 12 2000 - 08:42:43 PST