Using spline.cal to create camera paths?

Hi List!

I am currently trying to do some first steps in animation with radiance. I
did some search in the list's archives, as well as older radiance digests and
some basic descriptions about camera path interpolation in general. What I
found to be useable (I am not a maths expert;-) was the method described by
Greg Ward using rcalc, spline.cal and cnt to generate a view file from key
point.

So I defined a list containing 15 key points and directions, processed this
with rcalc and spline.cal, made 150 views out of this and was quite happy to
see all this was working (after having some trouble with the format-files
used by rcalc first). Than, I used rpict to render my 150 frames, and I was
quite happy with the speed on my old machine (the use of an ambient file shows
it's benefits here).

However, I didn't really get the camera path I expected to result from the
key points. I know that there are lots of parameters that define a path, not
only the key points. The problem here is that the interpolation let me to a
path that doesn't even come close to the points. So I need a method to bind the
spline generated by spline.cal closer to the defined points, making the
spline less smooth. How can I do this? Should I define more or less points? And
is there a parameter in spline.cal that can be used here?

Thank you, CU, Lars.

···

--
+++ GMX - Mail, Messaging & more http://www.gmx.net +++
Bitte l�cheln! Fotogalerie online mit GMX ohne eigene Homepage!

Lars writes:

However, I didn't really get the camera path I expected to result from the
key points. I know that there are lots of parameters that define a path, not
only the key points. The problem here is that the interpolation let me to a
path that doesn't even come close to the points.

This is puzzling, as the Catmull-Rolm spline used in spline.cal is designed to pass exactly through the key frame positions. Are you sure everything is working properly? Did you check your generated .cal file to make sure it has the right values in the right places? You should see something like this:

Px(x):select(x,79,72.3046,71.1568,71.1568,71.1568);
Py(x):select(x,24,23.2029,23.1347,23.1347,23.1347);
Pz(x):select(x,5.5,5.5,5.5,5.5,5.5);
Dx(x):select(x,-0.970143,-0.970142,-0.990371,-0.977998,-0.999862);
Dy(x):select(x,-0.194029,-0.194029,-0.0484078,0.0306398,-0.0091765,);
Dz(x):select(x,-0.145521,-0.145521,-0.129696,-0.206351,0.0138471);
T(x):select(x,0,2.74756,1.09528,0.658805,0.765425);

Each select call has the same number of arguments, corresponding to the number of keyframes you entered, and the Nth argument of each should correspond to your Nth keyframe value for that variable.

-Greg

Hi Greg!

This is puzzling, as the Catmull-Rolm spline used in spline.cal is
designed to pass exactly through the key frame positions.

Hmm.. that makes it easy to check.

Are you sure everything is working properly?

Not any more :wink:

Each select call has the same number of arguments, corresponding to the
number of keyframes you entered, and the Nth argument of each should
correspond to your Nth keyframe value for that variable.

Yes, everything seams to be ok here. I still might have used the cal file in
a wrong way in the last step when I generated the views. I will repeat the
procedure with fewer key frames and frames to check this last step. Is it ok
to specify one spline with lots of control points in general, or should I use
multiple splines? I am not sure if my 15 points are already too many controls
to generate a path as I expected?

The really important information for me is that the spline is to go through
the points I define, I didn't know that before and will be able to try some
more paths now. Thank you!

CU Lars.

···

--
+++ GMX - Mail, Messaging & more http://www.gmx.net +++
Bitte l�cheln! Fotogalerie online mit GMX ohne eigene Homepage!

Hi!

This is puzzling, as the Catmull-Rolm spline used in spline.cal is
designed to pass exactly through the key frame positions.

That was the most important information to me, as I was able to search for
the error. The cal-file was ok, I had made an error in the last step,
(generating the frames' views with cnt and rclac) where I had a wrong time
sum. The path is nice and smooth now, and it all isn't that difficult ;-)))

Thank you, CU, Lars.