[Uni-verse] what is needed for the thursday testing?

[Schreck Akos]

Hello
Please precise if you could what is expected from me for the testing on
Thursday.
Should we modify the models, or such?
I got the info that i.ll need a linux machine, does it mean 2 laptops?
win + lin? 
thanks, bye
akos

-----Original Message-----
From: uni-verse-bounces at projects.blender.org
[mailto:uni-verse-bounces at projects.blender.org] On Behalf Of Peter Lundén
Sent: Friday, April 20, 2007 10:57 AM
To: consortium mailing list
Subject: [Uni-verse] Report from test session

Dear partners,

Yesterday Mattias and I was performing a test session. Here is the
protocol from that meeting.

Report from the first test session in the preparation for the Review.

Present: Peter Lundén and Mattias Claesson

The purpose of this session was to verify the applications on the front
side of UVAS and their communication. Back side is I guess the most
tested part of the Uni-verse software so this was left to next test
session when Raine is with us. Testing of the reduction was not
performed because we did not have a resent binary so it has to be tested
next week.

Two cases where tested, a simple shoebox room similar (but not the same)
to the case from the IST event and a slightly simplified version of the
testscene (7 rooms and a corridor). The models are available at:
http://www.pdc.kth.se/~slas/universe/models/

I have looked at the last model from -/+. It is a very nice model but it
will not be possible to run directly in UVAS. It it the worse case you
can think of for a BSP based ray tracer. one room with hundreds of small
cavities around the walls this will heavily fragment the space into
thousand of small subspaces.


Case 1 Shoebox)
The model was uploaded to the verse server using loader. Connecting Max,
connector(both Linux and Win) and UVAS to the server. Changing material
(with acoustics) in connector. Its very cumbersome to scroll in
connector, very many mouse gestures are needed to get to the bottom of
the object list.

Changing materials from Max introduces problems. The material change was
reflected in UVAS as it should but Max was destroying the acoustic
properties of the materials, the removal of the acoustic properties was
not correctly reflected in UVAS (maybe a good feature for the review
occasion but still a bug).

A wall and a sound source was repositioned in Max and it was correctly
reflected in UVAS. Further testing is needed for sound source handling,
creating a source from Max was not tested.

Materials where changed from Max which was correctly reflected in UVAS.
We had to remake the model to get this to work. The reason for this is
that the model was using quads and that it uses multi materials but can
only handles triangles so Max will triangulate the model and then the
materials will get lost as Max can not fully handle multi materials. The
solution is to have a already triangulated model then Max is not
creating any new geometry and the material associations are not broken.

Quelsolaar was connected, it was verified that the listener in UVAS was
reflecting the camera moments in the renderer.

SGRenderer was connected, it was verified that the listener in UVAS was
reflecting the camera moments in the renderer. SGRenderer is stealing
the object with id 0 object in the server and adding the listener tag
group to this object, SGRenderer must use the right object which is the
object returned when connection to the server. It is still quite
cumbersome to navigate in SGRenderer (its not very easy in Quelsolaar
either).


Case 2 TestScene_simplified)

The Testscene model was slightly simplified removing doors and doorposts
which generates a lot of fragmentation in UVAS.

The model was uploaded to a verse server. Connecting Max, Connector,
UVAS, Quelsolaar and SGRenderer.

Windows version of Connector was crashing and was not able to handle the
model same goes for Quelsolaar. The Linux version of Quelsoolar did not
crash but din not show the model, only one lamp was visible. We where
trying to increase the light without any success.

Some reflections where missing in UVAS.

Connecting Max. Ceiling was moved in big room, it was correctly
reflected in UVAS. Material was changed on the walls and ceiling (its
one single object) of the big room, reaction in UVAS was OK.

Conclusions:
Its problems with most software using a not so complex model as the
Testscene model. At this stage I think its ridiculous to try to aim at
more complex models. First step MUST be that ALL tools works with one
level of complexity before moving on to the next level.

The only renderer we got to work properly (almost) was the SGRenderer.
Debugging of connector and Quelsolaar is urgent.

The Max plug-in was working except for the material bug.

UVAS was working but some reflections where missing.

We propose that we fix the TestScene model to look nice and prepare it
so we can demonstrate the change of material and geometry. The wall
between the to rooms at the north-east corner (+Z is north) is made as a
separate object and is then easy to move. The walls, the ceiling and the
floor  of the other rooms are also separate object so the materials
could easel y been changed and so each room can have its own acoustics.
We can also demonstrate that we can move the ceiling in the big room
(nice acoustic effects). Furnitures and stuff without acoustic
materials(ignored by UVAS) are added to the model so it look nice.

--PLu
_______________________________________________
Uni-verse mailing list
Uni-verse at projects.blender.org
http://projects.blender.org/mailman/listinfo/uni-verse