This picture was rendered with unimegapov v0.3. I'd like to
know where the white zone comes from. Here are the symptoms:

* it appeared when I put the stonework around the window
frames;
* it only occurs with radiosity, and it's there no matter
what recursion level I use;
* it's the same no matter what color I give to the stones;
* some tests showed that it appears as soon as I put 5
objects or more in one or the other of the windows, not
matter what objects or what color they are;
* it dwindles to nearly nothing when I increase the
neares_count and decrease the error_bound, but then of
course the render time increases drastically;
* the scene is lighted mostly whith radiosity (the ambient
is very low, there is just the sun outside, and a slight
shadowless fill-light at the camera position).

	Anybody knows what could cause that?

		Jerome

-- 

* Doctor Jekyll had something * mailto:ber### [at] inamecom
* to Hyde...                  * http://www.enst.fr/~jberger
*******************************

Post a reply to this message

Attachments:
Download 'radio.jpg' (7 KB)

Preview of image 'radio.jpg'

Jerome <ber### [at] inamecom> wrote ...
> This picture was rendered with unimegapov v0.3. I'd like to
> know where the white zone comes from. Here are the symptoms:
  [clip]
> Anybody knows what could cause that?

My guess would be that light is 'leaking' in from outside of the wall due to
too large of a value for error_bound.  Other than that, I'd only be
guessing, unless it is a bug.  I don't know why the number of objects in the
scene would have an affect on the radiosity (at least when the color of the
objects has no effect).

-Nathan

Post a reply to this message

On Thu, 17 Feb 2000 13:17:53 -0500, Nathan Kopp <Nat### [at] Koppcom> wrote:
>
>Jerome <ber### [at] inamecom> wrote ...
>> This picture was rendered with unimegapov v0.3. I'd like to
>> know where the white zone comes from. Here are the symptoms:
>  [clip]
>> Anybody knows what could cause that?
>
>My guess would be that light is 'leaking' in from outside of the wall due to
>too large of a value for error_bound.  Other than that, I'd only be
>guessing, unless it is a bug.  I don't know why the number of objects in the
>scene would have an affect on the radiosity (at least when the color of the
>objects has no effect).


Nathan is the expert on this sort of thing, but I seem to remember something
of a similar nature happening with an image that someone here was working
on over a year ago, something to do with autobounding kicking in.

Try with 5 100 per cent clear objects. Just a thought. 

-- 
Cheers
Steve              email mailto:sjl### [at] ndirectcouk

%HAV-A-NICEDAY Error not enough coffee  0 pps. 

web http://www.ndirect.co.uk/~sjlen/

or  http://start.at/zero-pps

 11:48pm  up 3 days, 10:51,  5 users,  load average: 1.04, 1.02, 1.00

Post a reply to this message

Steve wrote:
> 
> On Thu, 17 Feb 2000 13:17:53 -0500, Nathan Kopp <Nat### [at] Koppcom> wrote:
> >My guess would be that light is 'leaking' in from outside of the wall due to
> >too large of a value for error_bound.  Other than that, I'd only be
> >guessing, unless it is a bug.  I don't know why the number of objects in the
> >scene would have an affect on the radiosity (at least when the color of the
> >objects has no effect).
> 
> Nathan is the expert on this sort of thing, but I seem to remember something
> of a similar nature happening with an image that someone here was working
> on over a year ago, something to do with autobounding kicking in.
> 
> Try with 5 100 per cent clear objects. Just a thought.
> 
	Thank you both. Lowering error_bound does the trick,
eventually (1 is still too much), but the render time
skyrockets. Removing auto-bounding fixes it too and allows
for faster render (actually, by manually bounding my
objects, I'm able to get render times similar to those I get
with auto-bounding) so I guess there must be a bug somewhere
:(

	Anyway, thanks again to both,
		Jerome
-- 

* Doctor Jekyll had something * mailto:ber### [at] inamecom
* to Hyde...                  * http://www.enst.fr/~jberger
*******************************

Post a reply to this message