Would you like to make this site your homepage? It's fast and easy...
Yes, Please make this my home page!
Larry Schultz currently has enrolled more than 10 students in
his online tutorial program at www.splinegod.com
, and he commented that the recording method he uses (a special
verion of Media Player?) takes 1/20th the time of composing html
tutorial pages. In his presentation last meeting, he modelled a
head and created a slelegon skeleton for an animal torso in
record time, while sharing advice about workflow and avoiding the
"world of hurt." He also pointed out that much of his
facility with the interface was born of experience, working
independently and for many studios like Netter Digital and
Foundation Imaging, where he helped create the Foundation
Institute. For the March meeting, Mr. Schultz began with a
grid-like figure object, and used "spin quads"
primarily to make the contours agree with muscle position, and
tend to "loop". "Edge loops" produce better
shadows and more natural bones deformation. He did this for the
major muscle groups like the trapezius, glutes, ab's and
deltoids. This method would also save on dozens of additional
vertices produced by band-sawing or smooth shifting to add detail
where needed. Another useful technique is to "smooth
scale" to match a background copy of an object, while moving
vertex detail to where it is needed. Mr. Schultz is not fond of
point-by-point point-dragging to approximate these structures;
instead, he said, use the most advanced tools you have, and stay
away from much slower, older methods. On the subject of overall
workflow, he had many observations to share: for team working,
use default menu's, but use tabs to create special menu's of the
functions you use most, like skelegons; have an undo setting of
100 or more; minimize the use of constraints or expressions,
especially since many bones may be selected at once and grouped
for expressions-like bones deformation; apply weight maps to
groups of (smooth-shifted) points progressively in Modeller,
assigning different values (100, 50, 20%) to rows of points, and
assigning all the points to a group, in order to deform features
like mouths and eyelids predictably, with a minimum of selecting.
This last trick was based on extruding two lips from a grid shape
by smooth-shifting and rotating two rows multiple times. Although
he uses weight maps for modelling, he seldom uses weight maps for
animation. Larry Schultz also pointed out that much of the
deformation and modelling involved in character animation is not
driven by walk cycles or other actions that would lend themselves
to expressions; much is shrugging, looking, and moving over
non-flat terrain. To rig a character for this kind of action, he
recommended having a "rested" or "base" pose
at -1 frame, and placing extreme poses at about frame 10, to test
the bones for stray deformation. The top of the head touched by
the hand, the hand on its opposite shoulder. The past meeting's
rigging demo was basically the review of one of the operations to
be avoided if one didn't want to avoid "a world of
hurt" - shift-r versus r, resting bones on their respective
point clouds. When Larry does this, it all looks easy. The
stationary bones that have not been "rested" yet, he
adjusts the rest length of, and this cloud of deformation
protection expands to protect the area being pinched or
stretched. (I think the book teaches this, but in the confusion
of learning about ctrl-r being different from r/shift-r, one can
really get detoured. Ignore r/shift-r until all of the bones work
without pinching/stretching!!!) If this doesn't work, he will
either add a bone for a problem area, like a bone for a face, or
try increasing the strength. He often mixes skelegons and bones.
He often uses a falloff of 128^, for the sharpest falloff. He
uses weight-mapping as needed, but will use them for large areas,
in order to isolate the head deformation from shoulders, or
shoulders from the body. Once the major adjustment have been
made: "record pivot rotations" zeroes the rotation
settings -- and keyframe. Selecting several bones and renaming
this "selection set" with a group name has already been
mentioned. Another trick for faster refresh times was to make a
"stand-in" for a highly subdivided model using its zero
subdivision source, but cut up into respective bone areas like
arms, hands, etc. These would be saved in separate layers THEN
parented to their respective bones (after Object Replace) for
pinch-free rehearsals. This spares one from using the technique
used for 5.6, of using no object, just bones, for fastest
refreshed proxy animation. The trick of "Enable Deformations
On/Off" I frankly didn't gather, being an INSPIRE user. A
request for an IK rig in the last ten minutes of the session lead
to a left ankle goal, anchored at the pelvic mid bone. At first,
nothing happened, then Larry Schultz toggled off "unaffected
by descendants" for the affected bones. He assigned bank
only for the upper legs and pitch only for the knee area and
"pre-bended" the knee, though I missed the details. He
had earlier emphatically advised not to have knee-goals, and
applied "match goal orientation." A comment about IK
"jittering" lead to a remark about big objects having
small goal strengths and small objects having large strengths. He
made the point that FK is generally used for the upper body,
because the shoulder moves the arm, whereas the feet remain where
planted. IK excels in animation applications where a surface is
followed, like the weasel animation in "Ice Age." This
workflow streamliing makes all the difference in practice -- it's
easier to object-dissolve between a model with a cup of coffee
modelled in its hand and one without, while object dissolving a
cup to disappear from a table in one frame, than to rig and
animate this using expressions, if assigned the sequence during a
job.
