I should imagine so, yes...
Tbh, its got nothing to do with poser or daz rigs specifically... the same problem applies to any software.
Often (in fact, always) its impossible to get smooth, or rather CORRECT deformations at the shoulder through simple weighting (unless we're talking a VERY simple cartoon type character). One common way of fixing this is to make a series of morfs/blendshapes which "model out" corrections for certain poses of the shoulder. For ease, these can be tied to the rotation values at that joint to autocorrect (morf) the shape as you animate.
This by itself is fine. However, due to the range of motion in the shoulder, and the way that rotations are commonly interpreted (euler), the only way to transition between 2 poses in a certain direction involves flipping some of the channels at a single frame in time. When viewing just the bones, you cant see it... cos as one channel flips, an alternate channel also flips, compensating the overall pose. This ofc, gives you a hard value change in these channels between one frame and the next, and so if shapes have been tied to "follow" these values, they too will "snap" at the same frame.
Ofc, in other packages, where you have more thorough manual rigging tools, there are various ways to compensate for this... You can either euler (or gimbal) filter the curves to correct the flipped channels, or, you can resample the rotations at that point using quaternions, and use the results of the quat rotation data to drive the shapes instead (more or less whats referred to as a pose space deformer).
So yes, MB, maya, etc, all have channel filters that allow you to resample the curve data to try and avoid, or cancel out these flips as much as possible (in MB there's various types, gimbal killer, unroll, etc, each of which perform different reinterpretations/resampling of the curves).
Thing is... even with these filters... it isnt always worked out for you auto-magically... You will find that you may still need to perform manual changes here n there... or use different filters to resample different parts of the curve data.
Ofc, the simple fix, in poser, or wahtever else... is jsut to dump the morfs that are linked to the bone rotations. The flip of the bone will still happen, but you wont see it... cos the effect you are "seeing" isnt the bone flip, its the morf flip. Ofc, doing this will reduce you back to basic weighted shoulders, which will often lose volume, or look, pinchy.
Point is... you still have to make your own rigging decisions, and curve management in order to get things behaving exactly the way you want or need them to. Mocap is wonderful, but its by no means a panacea that can save you having to do other stuff manually... tahts just the way it is.
And tbh... as far as iPi goes, and wonderful as it is... I would consider it pretty useless by itself... too many inacuracies, too much foot skate, etc to be of any real value use withou having another package to edit and cleanup the resulting mocap data.
Not that thats a bad thing, ofc... Im not aware of a SINGLE mocap system out there that DOESNT require any manual cleanup.
As for what package to choose... MB is by far the leader, but if all you're looking for is some basic clean and fliter stuff, then you wont be using about 80% of the features in the package.
With that in mind, you might do well to look at Maya... it has mocap retarget and cleaning tools (brought in straight from MB, effectively giving you a "motionbuilder lite" right inside of maya). Add to this the fact that Maya is actually cheaper than MB, plus has a whole cavalcade of rigging, skinning, and animation tools that MB doesnt, and you may find it more suited to your needs.
|