Volca on 18/6/2009 at 11:59
Fine! The plan for 0.3 is roughly:
* simulation
- Room service
- Physics
- A/R system (maybe, this one is quite huge task and may span to 0.4 or 0.5)
* scripting - design of the message system, script messages for services that are implemented
* GUI basics - Mouse cursor support, Thief/SS2 menu system, later also SS2 window system
* save/load game support. Code exists, needs to be fixed and improved to the state of usability.
* others - some ideas:
- container service
- sound system
- particle systems would be nice addition
- writing a new rendering pipeline in parallel to the old fixed-function one, that would support modern effects (dynamic lighting/shadows, normal maps, parallax/cone step mapping, maybe a deffered shader or light pre-pass, you name it :) )
- other ideas that are not out of band (we have to follow the structure of the original and can't just come up with different solutions)
- toolkit for dromed users - there is some code present to export GAM to text file, we could also provide diff and patch tools, for example.
If you'd be interested, let me know, I'll be happy to discuss the possibilities :)
*Zaccheus* on 18/6/2009 at 13:05
I'll see if I can get at the source code and have a look around.
I don't know how far you got internally, but I think being able to bounce off the walls without going through them would be a huge improvement. :D
Volca on 18/6/2009 at 13:22
Let me know if you'll need assistance - the DEVELOPERS file in the SVN gives many clues but probably is slightly outdated. I'll give you my hand whenever possible to get you going.
You'll see that collisions can be a complex thing to solve - BSP/Portal+Cell structure collision detection system has to be written(or a fitting one found, which is unlikely) first in order to do that. The idea I have is to find the cells object is present in (via BSP traversal), then do a collision detection with the cell's polygons. We need cell+box, cell+sphere and ideally also cell+mesh collision routines.
SiO2 wanted to work on the physics but I didn't hear from him the last month or two (Hello? :) ).
*Zaccheus* on 18/6/2009 at 20:09
Have the source code now, will take a while to read through everything. :cool:
bikerdude on 18/6/2009 at 20:56
sweet, nice to see this is still alive..
*Zaccheus* on 19/6/2009 at 12:37
Quote Posted by Volca
The idea I have is to find the cells object [Garrett] is present in (via BSP traversal), then do a collision detection with the cell's polygons.
I was hoping we could keep track of which cell Garrett is in, such that we do not need to re-find the cell each time he moves. I'm assuming that there will be (or is) a simple way to determine which cells each cell is connected to.
Is there a separate forum where this stuff is discussed or are you happy to discuss it here?
Volca on 22/6/2009 at 06:14
Yeah - the BSP traversal is done at least once every frame (for camera) anyway and it does not introduce any performance problems. Dark separates the sleeping and awake objects for physics to lessen the load on CPU, that seems to be enough. What will be needed is material aware collision resolution for static geometry v.s. movable objects (and sure it has to handle all the other specifics of dark engine physics as well - friction, object density, water interaction in general).
We have a project members only phpbb forum on sf.net - we didn't want a guest/public area as it would mean moderation had to be done. I'm not sure I want to change that :erg: So for now we can keep the discussion here (or move to PM), no problem. We can also discuss using opde-devel mailing list if you'd prefer...
Albert on 22/7/2009 at 23:54
If it doesn't bother you to give me a small package containing a compiled barebones copy of the current OPDE source, I would be happy to test it...
At least, peruse it. :rolleyes:
MoroseTroll on 23/7/2009 at 07:21
Why don't you download it by yourself? The package is small enough (4.7 MB), and I've downloaded it a week or two ago.
