Re: [Adonthell-artwork] height levels

[Alexandre Courbot]

> > I have a few questions about the new height level thing on mapobjects:
> > 
> > Do you specify a height/ z-offset (or whatever it was called) for
> > whole mapobjects or is there something like a height-map file which
> > determines how high the individual pixels of a mapobject are?
> 
> It's for the whole mapobject. 

Don't listen to him James! (my turn, now ;p) Height is for the whole
mapobject FOR NOW. Just for my testing purpose. But I planned to give
the possibility to specify a height level per pixel.

> > Food for thought, I guess. For now I'll just improve the look of those
> > gfx...
> 
> Poor Alex :). One idea I could offer it to divide a mapobject into
> 'zones'. Say 5x5 or 10x10 sub-tiles that can have a different height
> each. That wouldn't be as accurate as a height map, but less CPU
> expensive and probably sufficient. But still a bitch to program, I'd
> say.

Here's the dillema. Something accurate, or something fast? As far as I
know, such sub-dividing wouldn't help much with CPU usage. But pixel
based height (and walkability) seems to be extremly CPU intensive.
Actually, as the movments are vector-based, it looks like I'll have to
borrow some 3D world techniques. Problem is, I'm not very good at this
(never did any 3D programming before), so I'm documenting myself for
now. The problem being, that you have (AFAIK) to check
walkability/height for ALL the pixels. That is, 1600 instead of 1 for
square-based walkability. So yet I'm searching fresh ideas, and
consulting some people familiar with 3D programming, hoping to sort that
out. If someone has interesting ideas regarding this, I'm ok to take
them! :)

But to answer James: this has been planned as you described it, no
worries!

> Of course, the easiest thing would be cutting the graphics into small
> pieces - at least for us programmers ;).

Yeah, was what we did for 0.2 and such. It was heaven time for us! ;)
But since the situation has inverted... :p

Alex.
-- 
http://www.gnurou.org