Mantis - Croquet
Viewing Issue Advanced Details
1203 Any minor always 05-16-05 21:01 05-16-05 23:14
elenzil  
 
normal  
new  
open  
none    
none  
0001203: portions of vrml models being dynamically culled out
This is really a problem w/ the vrml importer,
but i've had little suvvess tracking down anyone still attached to that code, so.

With the VRML models at:
http://predictor.scripps.edu/scientific_update_gallery.php [^]

As one gets closer to the models,
various portions of it seem to be culled out of the scene,
as if by a visibility test,
even tho those portions are well within view.

Also picking is very dodgy.

The models are composed of several sub-objects,
but there appears to be no tricky transforms in the file.

Have loaded & re-written them with 3rd party software,
obtaining a differently-structured file but with the same results.

See attached screenshots.

 vrml culling.png [^] (429,553 bytes) 05-16-05 21:01

Notes
(0001494)
elenzil   
05-16-05 21:38   
actually i think this may be my fault.
more to come...
(0001495)
elenzil   
05-16-05 23:14   
confirmed: this was introduced in my own version by me.

it's frustrating tho -

the cause was that i re-configured the behavior of #scale:
so that instead of this:

---
SomeClass scale: s
  multiply my own translation by s.
  actually scale myself.
  for all my children:
    child scale: s.
---

it was like this:

---
SomeClass scale: s
  actually scale myself.
  for all my children:
    multiply child's translation by s.
    child scale: s.
---

  
this makes a lot more sense to me
because with the previous way,
if you scale an object, it translates !

however,
this ran into complications with the boundSphere of TMeshes
which i haven't been able to sort out.