[Bf-committers] Blender success story, a couple of potential bugs and a feature request.
Jonathan Merritt
bf-committers@blender.org
Mon, 24 Nov 2003 14:55:23 +1100
Hi Everyone,
I recently put Blender to use on an animation that was shown at
Equitana, which is a big commercial horse show that was held here in
Melbourne over the last weekend. I gave Blender a big plug, pointing
out that the animation would not have been possible without the efforts
of Free Software programmers all over the world. Here's a single frame:
http://www.warpax.com/temp/forelimb-frame.png
The animation showed a moving horse limb (animated with real kinematics
data that I pulled out of a refereed publication), and illustrated the
function of one of the ligaments of the limb in storing elastic energy.
I'm reluctant to post the full animation at this stage, since there are
a few things that need to be fixed, and among them are the bugs listed
below.
The limb is an animated armature (animated by rotation and location
keyframes, not by an IK solver). It has a basic animation that is set
to repeat by having a cyclic extend mode. In order to control the speed
of the animation, the armature object has a Time IPO curve, whose slope
varies through the animation to slow the limb down for a while before
moving it back up to full speed.
I discovered that the rendering of objects parented by the "Use
Armature" method seems to have a bug after the first frame, when motion
blur is used. The following is an example:
http://www.warpax.com/temp/forelimb-ghost.png
Here, the ligament at the back of the bones is rendered in entirely the
wrong place for just *one* of the motion blur frames (the others may be
out of place as well, but I can't tell). Note that this bug *does not*
show up in the *first* rendered frame, only in subsequent frames, and
does not appear in Blender itself when an animation is played using
Alt-AKEY. Hence, you can't observe the bug by rendering a single frame
- it will only appear in the second frame of a rendered animation.
Also, the amount of blur applied to the animation does not seem to
respect the relative progression of time indicated by the Time IPO
curve. Hence, even when I slowed the Time IPO curve to a complete halt
(ie: horizontal line in the IPO window), the rendered objects were still
being motion-blurred. Is this the desired behaviour?
Finally, an option to render a flat mirror reflection map (as opposed to
a cube-face environment map) would be really wonderful. In this
instance, with a heavily set-up scene, I was able to reflect things
manually only by painstakingly selecting everything except the bone mesh
objects and reflecting them in the ground plane. This was quite a
hassle, considering the simplicity of the effect from a technical point
of view.
Jonathan Merritt.
PhD Student in Equine Biomechanics,
Faculty of Veterinary Science,
The University of Melbourne.