disappearing faces! hel
14 messages in this thread
Hi to all!
I have the following problem:
I made a cube and copied a few hundred times with the create/object/array
function. I mapped the cubes with a bumpy box mapping. The cubes are standing
on a flat ground with a bump mapping too (I used a tiled planar mapping).
In my 2300 frames large animation I fly through the cubes with my camera
(linked on a CamDummy). Now the problem: sometimes the faces wich are close to
my camera (the cubeside to the left and to the right side of my camera, and the
ground) are dissapearing. I don`t hit the faces with my camera, my Z-clip
setting is the default value of 1.0 units. What can I do to solve the problem?
BTW: I use one spot light with raytraced shadows. On my Pentium/100 Mhz with
64 MB RAM a image in a resolution of 720×576 takes more than 8 hours to
render… (without swapping to disk). Well I use some animated opacity maps to
show some special effects, I use some LenzFX Aura`s and have some spaceships
flying in the space. But 8 hours seems very long. I think there are some ways
to optimize the raytraced shadows <g>.
Thank you very much in advance.
Michael
Michael:
Z-clipping is the first thing that comes to mind. I'd try a setting of .1 and
see if that works. I've used this setting and have had success, FWIW. If this
fails I would link everything together in the Keyframer and scale the whole
model up. To cut down rendering time have you deleted all of those cube
bottoms which are sitting on the ground? Can you delete any other faces which
never get pointed to the camera? Can you exclude some cubes or other objects
from your light source? Can you change the attributes on some objects to keep
them from casting shadows or receiving shadows? A couple ideas anyway.
Bob
Hi Robert!
Thank you very much. I`ll try a Z-clipping of .1. I already deleted all faces
that I don`t need. But still much rendering time. If I turn off the raytraced
shadows the Pentium/100 renderes an image in a few minutes. I tried to use a
shadow map but this needs a very big shadow map size (much bigger than my
memory). My solution is to make two weeks holiday. The best way I think <bg>.
Michael
Hi Robert!
To set Z-clipping to 0.1 doesn`t work to solve my problem. I tried to scale up
the whole scene but then I loose all my animation pathes and they are very very
complex and important. I can`t animate my scene again (very much sync.
points)…
Any other idea to keep my faces away from dissapearing ? Do you think it`s
because of the box mapping? But the ground doesn`t have a box mapping, only the
cubes. Hmmm, seems like a strange thing…
Michael
BTW: I use a 28 mm camera lens and I thought it could help if I use a 50mm
lens. It doesn`t work… 🙁
Michael:
You could try setting the z-clipping to a number even lower than .1 (.01 or
even
001). I don't know what the practical limit is but the explicit limit is 0.
A keyscript might be a way to scale up all of your camera keyframes. I'm not
even a novice at keyscripting yet, so I don't know if this idea will work.
Is there anyway to avoid using the low to the ground camera angles?
Bob
Bob:
Yeap, I`ll try a lower value. I also have no experiences with keyscripting.
Originally I was a programmer and when I heard that 3DS will have keyscripting
I was quit happy to make some nice scripts. Now I have no time to read the new
manuals… 🙁
Michael
Hi
I do have the same problem with z-clipping (see also 3DS forgets polys) and put
the z-clipping down as much as possible but this still didn't help.
Please any suggestionss are welcome
Tom
Hi Lars,
yeep, I saw your message. Putting Z-Clipping to 0.001 doesn`t work for me and
scaling up the scene to more than 1000% doesn`t work, too.
Then I spent some time under the shower (I have my best ideas there <g>) and
had the following idea: I use a basic box (with two faces) and put my whole
scene on it. Now my idea was: how decide the Z-Buffer when a face is visible or
not. For e.g. when I`m on the ground with my camera sometimes more than 90% of
the face isn`t visible (because it`s behind us). To make sure that my idea goes
in the right way I used a complex grid with a few 1000 faces as the ground.
After rendering a testimage I saw that the most faces are visible. Only some
holes appear close to the camera. The more faces I use for the ground the less
holes are rendered!!!
This technique works fine for my cubes but not for my very large ground plane.
I already use more than 20.000 faces but I still have some holes. Any ideas?
Michael
Michael, Got your ground problem solved? What do you do the ground object in?
If not 3DS then have you checked to make sure it's not flipped normals or
something instead of a clipping problem. If you scaled up your scene 1000% then
it's most likely not a clipping thing. Can you look at the object in optimize
or view or something and see if it's normals. Hope you get it solved. Let me
know.
DM
Hi Dave,
sorry for the delay but I was in HOLIDAY!!!!!! Yeap, now I`m ready for strange
new things…
Well, I made the ground in 3DS and it wasn`t a normal problem. The problem
only appeard when I used box-mapped cubes on the ground. Very strange…
Michael
>> A keyscript might be a way to scale up all of your camera keyframes.
I believe you can link everything to a single dummy object and then scale the
dummy as needed at frame 0. Everything including your keys should scale
accordingly. (I've heard it discussed, but haven't tried it yet.)
– Dave
Dave:
I suggested the scale idea but I don't recall the magic of doing the scaling on
Frame 0. I'll have to try that one on my own, thanks for the idea.
Bob
Scaling with a dummy object is the right idea, but if you want to preserve
animation paths, you should ONLY link the ROOT parents to it (root parents are
the objects that are on the far left in the show tree list), NOT their
children! It's probably also best to do the scale-up in frame one, not zero –
this leaves your editor alone for continued adjustments and new object creation
(when trying to compare them to objects already existing). Actually, other than
placing pivots and linking, I don't do anything else in frame zero – I copy all
track info to frame one and do all my keyframe work from there. My editor is
then always how I last left it!
Ernie.
Yeah, spot on Ernie, I do that too! It's a v sensible technique.