[Openal-devel] new distance model
dpeacock at creativelabs.com
dpeacock at creativelabs.com
Thu Apr 3 09:28:16 PST 2003
Hi,
This applies to the 'DirectSound3D' path only ...
If we want to support any non 'Inverse Distance Clamped' model in Direct
Sound 3D, the only way to do this is to set the DS3D Global Roll-off factor
to zero (so sources don't attenuate at all), and then manually control the
volume of each source as it moves around the 3D World. However, with a
scheme like this, it means potentially making lots and lots of SetVolume()
calls ... e.g if the Listener position changes, it will affect the volume
of *all* (playing) Sources (potentially up to 64 on Audigy 2) .. so in
response to one Listener Position call, OpenAL will need to make up to 64
SetVolume calls ! Every time a Source changes position, you will need to
make an extra SetVolume() call too. All these extra calls will slow
things down .... I know because I implemented the Inverse Distance model
and did some benchmarking, and it was so bad I decided not to release it
!!!!
Plus, in addition to the above, it is necessary to correct the reverb send
levels from each Source because as far as the EAX driver knows the sounds
are not attenuating with distance (rolloff factor == 0), which messes up
the wet-dry mix.
I can certainly look into this issue again, and do the best job I can to
minimize the hit when using 'DirectSound3D', but as Garin said, at the end
of the day, the 'native' openal32.dll's shouldn't have this problem, so it
will encourage everyone to create native openal32.dlls ! Definitely time
to fix our device enumeration stuff !!!!
Dan
Daniel Vogel
<vogel at epicgames.com> To: Garin Hiebert <garinh at turkeytrot.dhs.org>
Sent by: cc: <openal-devel at opensource.creative.com>
openal-devel-admin at opensource.cr Subject: RE: [Openal-devel] new distance model
eative.com
04/03/03 06:40 AM
On Thu, 3 Apr 2003, Garin Hiebert wrote:
>
> I've got to get ahold of Bernd and encourage him to join this list and
Just subscribe him manually ;-)
> performance penalties under the non-hardware-accelerated versions as it
> doesn't match up with existing DirectSound models. (Daniel Peacock was
> telling me this earlier tonight.) From the perspective of an audio
Could you or Daniel elaborate on this? The "DirectSound" codepath has to
do the attenuation in software anyways, right?
BTW, something I forgot to mention - per source rolloff doesn't make sense
with the new distance model so it shouldn't affect it.
-- Daniel, Epic Games Inc.
_______________________________________________
openal-devel mailing list
openal-devel at opensource.creative.com
http://opensource.creative.com/mailman/listinfo/openal-devel
More information about the Openal-devel
mailing list