adonthell-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Adonthell-devel] stairs-back collision issue


From: Kai Sterker
Subject: Re: [Adonthell-devel] stairs-back collision issue
Date: Thu, 6 Sep 2012 20:41:28 +0200

On Tue, Sep 4, 2012 at 9:41 AM, Kai Sterker <address@hidden> wrote:

> No clue yet what might be causing it and how to actually debug it.
> Looked at the verbose collision system output but could not spot any
> obvious difference between that one and any of the others. Might need
> to dig even deeper, but might also be something entirely different.
>
> If it is indeed a problem with collision, one possible cause might be
> that the backward faces are wrongly orientated. Might need to write a
> unit test to ensure that this is not the case. OTOH, I would assume
> that the first collision occurs against the vertex along the back
> upper edge of each part of the stair, so face orientation should not
> even matter, Will give it another shot tonight ...

No success on that one yet, but here's a few things I've tried/done.

The new unit test clearly showed that the triangles all our models are
made of face in the proper direction, i.e. outwards. To verify that, I
changed the direction between front/back and ended up with the
character getting stuck to walls.

I then spend yesterday evening comparing the collision implementation
with other implementations of the same system and did not spot
anything out of the ordinary. Note that I only did that for the
collision step, not yet for the response.

What I did find out though is that the left to right stairs model
exhibits the same issue, while the right to left model works fine. So
it appears that there is a certain kind of symmetry to the issue.

Today I just tried with a differently sized player model, but neither
22x16 nor 26x20 made any notable difference. So what I am left with is
adding additional logging to the collision and deflection stages and
then run tests on the working stairs and its opposite and compare the
result, value for value to see where things differ.

However, as I'll be gone from tomorrow until Sunday evening, this will
have to happen next week.

Kai



reply via email to

[Prev in Thread] Current Thread [Next in Thread]