#flashing shadow
17 messages in this thread
Quick question….
We used to run all of our braodcast stuff field rendered, but ran into a
problem nearly everytime where we would get half a shadow somewhere in the
animation. So…we went back to frame rendering (longer animations now, of
course) but now we get these weird flashing shadows.
We are not ray-tracing and shadow maps are set at default values. Could the
problem be in the SHADOW-BIAS-ABSOLUTE setting in the 3DS.SET file? Any other
suggestions?
Thanks in advance.
Hi Chris:
Are you displaying via PAR? And the shadows flash?
-Kouros
Nope. This is just straight hard-drive to videotape. And the shadows flash
for the entire length of the animations.
Chris,
Very strange. Does this happen on more than one machine?
– G
Yes, both problems (the field rendering – shadow missing in one field
somewhere, and the flickering mapped shadows) occur on both machines. We tried
increasing maps size and tightening hotspot and falloff and still have the
mapped shadow flicker problem.
Bizarre… That's extremely unusual. Perhaps you could try to work with Jonas
or Brian on isolating a test case.
– G
It's DEMO tape time of year for us right now and we are quickly trying to come
up with quick fixes for these problems. After the tape goes to the post-pub
house next week we can take a closer look at it.
Thanks.
Gary
I just had the same shadow flash but haven't had the time to play with shadow
map parameters. I'm going ot be out your way this weekend so won't have time to
test it but will try next week.
TedB
!^NavFont01F0007MGHHGA0760D
ted,
i too have experienced a number of incidents of "flashing" shadows. i am at a
total loss to explain why this happens. i have been unable to isolate what in
particular (for any given animation) causes this rather bizarre looking
aberration. one time produces perfect shadows, the next generates throbbing
flashing shadows.
i realize this is in no way a solution, just affirmation that it does occur. i
have found it very difficult to explain to clients that this is a "special
effect!" hopefully, someone has a clue and will be willing to share it with
all.
michael
If someone could come up with a simple, reliable, reproducible case
for it, I'd be grateful.
gary,
sorry for the delay. i'd be happy to send a simple example. i've not
uploaded anything previously (i know, for shame…) and i'm not sure
quite where or how to upload a zipped file for you. if you could
enlighten, i'll send you a flashing shadow sample post haste.
thanks,
michael
Just simplify it down to the two relevant objects and no texture maps. Save the
.prj file and write up a short .txt file explaining exactly what to look for.
Then zip them up and send them to me via binary email.
Thanks,
– G
>> Could the problem be in the SHADOW-BIAS-ABSOLUTE setting in the 3DS.SET
>> file?
I ran into appearing/disappearing shadows that were corrected by that setting.
I believe the problem occurs in scenes that have a fairly large amount of
changing geometry. In my case, I had a robot and conveyors in a large room. As
the arm swung from position A to position B, and objects were hid/unhid with
hide keys, shadows would appear/disappear in other parts of the scene.
SHADOW-BIAS-ABSOLUTE=ON helped in this case, but required me to adjust my
shadow casting parameters.
– Dave
I messed around with map size but no help. Now I will give your suggestion a
try.
Thanks.
>> Now I will give your suggestion a try.
You didn't really elaborate on the characteristics of your shadow flicker. If
you still can't get it, post a more detailed description here and I'm sure
someone will jump in with other ideas.
– Dave
Are you net rendering? If you are you may have some parameters set differently
on some of your machines. I ran into this problem with texture maps once. The
give away was the net render log where we found the offending frames repeatedly
came frome the same machine
ME
Yes I am net rendering, but only on my own machine. Our machines aren't hooked
together.
I keep forgetting to adjust the shadow-bias-absolute value so I can't respond
on that, but in several models we continually see the problem in all test runs.