In "lock head" mode, hip position can drift insanely far from hip tracker
complete
knah
The way to reproduce this that we found was to recline on your back at about 45 degree angle (gaming chair style) and simply move the head a bit. What you'll observe is the hip swinging wildly forward and back, as if moving between the tracked position and the animated position of crouch animation.
While in head lock mode hip drift is inevitable in some poses, I'd expect the hip to at least stay constrained to the line going through head and hip trackers (i.e. it drifts down, away from head), instead of just going to completely random positions unrelated to hip tracker position.
Log In
Kung
marked this post as
complete
Kung
marked this post as
planned
Kung
marked this post as
interested
The original behavior during early development was as you described (the hip shifting only in the head-towards hip direction).
However this causes avatars to bend their knees as soon as they look down when in a standing pose. The hip moving back as well is an effort to prevent the erroneous knee bend. However I was on the fence about implementing the back move.
If there's support for simplifying lock head to only shift the hip in the direction of head-towards-hip that's something I'm interested in. It would certainly make it more stable. I'm thinking the simplified motion would also be best for very non-human type rigs like dragons with digitigrade legs or things like that.
Does this sound like what the community would prefer? (Feedback please)
bd_
Kung: I'm not quite sure I understand the proposed change. Do you mean the hip would move directly away from the head to accommodate drift?
Kung
bd_: The hip would move directly away from the head to compensate for a crunched spine. There would be no backwards movement associated with the crunched spine. The current "drift" (I assume you're referring to the backwards movement of the hip) would be removed, but as a consequence avatars will have more knee bending (even without bending your knees IRL) every time the spine is crunched (which usually happens when looking down).
I was unsure about implementing the backwards motion in the hip to compensate the knee bend, because the other two modes (lock hip, lock all) are already available to anyone who doesn't like knees bending at the wrong time. Because the backwards motion is receiving a negative reaction I'm considering removing it. The question to the community is would the knees bending every time your spine is compressed be better than the hip drifting to the back? (of course only referring to the "lock head" mode)
knah
Kung: To be fair, the "knees bend when looking down" is also a relatively common complaint from users of [redacted]. There it's caused by solver not bending the spine when it's mostly straight, so the exact behaviour you described occurs.
I suggested a mixed "lock both, but lock head more" mode in my other reply, and I feel like it could solve this issue too - at least if it slightly bends the spine when simply looking down too.
bd_
Kung: I feel like part of it is the difference between sitting/standing regimes here. I already feel that head lock is inappropriate for standing, at least for my rig, and would prefer it for sitting, if not for the very visible knee and hip motion that occurs with the smallest head movements. Ultimately my preference is to reduce the knee motion when sitting, but I'll note that (as I mentioned in my other canny) there are some directions of motion (looking down) where i suspect distorting the spine would work better and others (looking up) where I suspect moving the hip would work better.
Ronsterwars
You can also easily test the same problem by standing up and looking straight down to your feet. The hip will then drift backwards.