Christoph Hormann <chr### [at] gmxde> wrote:



> Geoff Wedig wrote:
>> 
>> Well, sections are unioned together, so that should bound them somewhat.
>> So, for example, the front tower is made up of the base unit, the upper
>> unit, the outward sloping part of the crennelations and the wall part of the
>> crennelations.  That what you're talking about, or are you thinking of
>> bounding it at smaller segments than that?
>> 

> That surely already helps, but the more 'bounding levels' you have the
> faster it gets.  I'm not sure how you build up the walls, but for example
> horizontal slicing could work.  Then bound together two neighbored slices
> and then put two of these blocks together and so on.  

Unfortunately, there are rarely horizontal lines that go completely around
(if ever.  The algorithm is such that it tries to break both horizontal and
vertical lines)  I can do *some* breaking up by 

A. Reducing the number of blocks generated that aren't actually visible. 
This has already been done somewhat.

B. Breaking things behind the scenes.  The two foremost towers on the left,
for example, could be broken behind the crennelations and no one could tell.

But honestly, the untextured isos are pretty quick.  It's only when you
combine my complex textures with radiosity and so on that things get bad.  I
need to do some optimizing.

Geoff

Post a reply to this message