PublicVR Forum

Online discussions of everything having to do with PublicVR, especially current projects, tech support, educational practice and theory. See http://publicvr.org for basic information and send questions to jeff@publicvr.org
It is currently Sat Feb 24, 2018 7:04 am

All times are UTC - 5 hours




Post new topic Reply to topic  [ 2 posts ] 
Author Message
Unread postPosted: Mon Aug 10, 2009 10:10 am 
Offline

Joined: Tue May 27, 2008 11:29 pm
Posts: 558
This thread is for discussion and results of all issues related to testing and packaging CaveUT for final distribution. The reader should also be aware of a similar thread for VrpnUT.


Top
 Profile  
 
Unread postPosted: Mon Aug 10, 2009 10:13 am 
Offline

Joined: Tue May 27, 2008 11:29 pm
Posts: 558
Testing and Packaging For Distribution

This thread is for discussion and results of all issues related to testing and packaging CaveUT for final distribution. The reader should also be aware of a similar thread for VrpnUT At <give link>

*

The latest version of CaveUT is 2.5.8 nicknamed CaveUT RC-8 on this forum <give link>, while the Earth Theater at the CMNH is still running RC-3 with good results. So I tested all versions, RC-3 through RC-8 for performance. These are the most important results.

Apparent translation and rotation in the spectator views is nice and smooth in RC-6 using the standard keyboard/mouse controls, under the following conditions using:

    The base CaveUT package plus CaveUTCursorExample.ini

    Listen server or dedicated server (both worked)

    EmptyLevel, a textured box which imposes no performance issues of its own.

    SmoothTime=0.4 in CaveUTCursorExample.ini

    Nearly identical Server and Client (MAC Pros)

    OpenGL

SmoothTime is important. The default value 0.02 in RC-6 zip package does not work well in my test setup, while 0.4 does. The higher the value, the more CaveUT will make for spectators speed (at a given moment) be the average of the speed specified by the server (for that moment) in the speed for the most recent moments. As you increase the value, it makes your speed increase/decreace more and more gradually. At 0.4, I could not "stop on a dime", but it did smooth out all of the apparent choppiness created by the net traffic. It looks great!

EmptyLevel will be part of the standard CaveUT distribution from now on. It's just a big box with a few lights that lizard skin default texture, perfect for testing.

Movement with RC-8 (2.5.8) was choppy, Regardless of how I set SmoothTime. However, it works best with VrpnUT 2.5.4 (VrpnUT RC-4). Obviously, we prefer to have a single version of CaveUT that works well in both situations, but we can live with what we have now.

GroundSpeed is much slower with CaveUT than where it is set for standard UT2004 movment. We did this deliberately, because CaveUT is intended for cultural and media projects which require more delivered movement and navigation by non-game players. CaveUT RC-4 delivers excellent performance at the high (default) speed, so we recommend that for gamers, for now.

We would like to be able to adjust the ground speed as a parameter, but that would not be so simple. CaveUT has to maintain a delicate balance between ground speed and acceleration in order to defeat aspects of the built-in movement algorithms in UT2004. We could shut those off completely, but then we would have to reinvent a large and complex chunk of engine code with no guarantee of success. In theory, we could find the proper ratio curve of the ground speed to rate limit, and use that to always smooth motion. An experiment of that type is on wish list, but again the current situation is satisfactory.

A HEADDESK MOMENT: Testing with the Virtual Egyptian Temple, the spectator view updates faster, providing smoother apparent motion, When I use OpenGL, rather than D3D!!! IMHO, this is probably because I was running both spectators on "MAC Pro" computers running WindowsXP using Apples BootCamp drivers. It is not hard for me to imagine that they might do a relatively wimpy job of writing the D3D graphics card drivers. Or maybe it's because CaveUT's OpenGL driver is much more recently written than UT2004's ancient D3D drivers. Bob Kuehne's VRGL includes shader code, which did not exist before the final version of UT2004.

Whatever the exact cause, this is likely the source of many WTFs in my performance testing for CaveUT. Before this week, I had been using our Shuttle PC for one of the Spectators and a MAC Pro for the other. The Shuttle is dead at the moment. When I get it fixed, I'll do some comparison testing.


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 2 posts ] 

All times are UTC - 5 hours


Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
cron
Powered by phpBB® Forum Software © phpBB Group