View Full Version : CA: How do the studios do it?

02 February 2007, 06:56 AM
I was wondering about how the people in the studios do walks. It seems that EVERY tutorial I see involves a static walk cycle that is then moved and repeated afterwards, but I heard a long time ago that this is the wrong way to do it. I've since been actually animating the complete walk cycle and while it involves more work it seems to look more natural to me.

Anyway, after getting two more animation tutorial videos that show the static "walking in place" method, I'm curious how the "pros" at the big studios do it. I kow there are a lot of real professionals who use XSI for movies and TV on a daily basis around here, so I figured that this is the best place to ask. Do you guys find that there are times where it's best to do the walk cycle, or do you always animate the complete walk?

Where I work, I do a bit of everything, so I don't get to do character animation every day, and when I do I always seem to have to relearn a lot of what I've forgotten since last time. I have the basics down, but there is remarkably little information available on the intermediate to advanced 3d charcter animation, which is where I always find myself having problems.

I guess this whole question comes from me always asking myself if this is the best way to do things and wondering how the pros do it.

02 February 2007, 01:40 PM
My guess is that it depends of the animator, and also of the animation pipeline.

Some animators may use the Mixer heavily, so they animate one walk cycle, and then move the character around. They can also put some animation on top of that for the acting part.

However most of the animators that I work(ed) with never use the mixer for such things, instead they fully animate the whole thing. They probably copy-paste keyframes all over the place and modify them so at least they have a good timing, but it's still hard labor. The animators that work(ed) with, just like everyone else, like to have low-level control on their work, and keyframing the old way gives them that control.

I've never worked with animator that animates purely pose-to-pose.

Now the animation pipeline is important: animators will do only what they are allowed to. If a pipeline is based around the Mixer, then animators who don't use it will have to be trained. On the other hand if the animation pipeline is based on good old keyframing, then chances are animators will have to do keyframe animation, etc. Some more efficient pipelines will allow the animators whatever approach they like, but this is a lot of work for the TD.

Now this comes from someone who's never done animation full-time and never done rigging full-time as well, just worked with people who do that. Take it for what it's worth.


02 February 2007, 04:08 PM
Personally, I never animate in place.
First I block general movement in space, then hips, legs and the rest.
Some place may have a library of cycles to speed up work and stay in character, but those are usally just a starting point.
What Bernard said is true.

02 February 2007, 06:17 PM
Walk cycles are rarely used in feature animation - unless you need to save time and have a limited budget.

Walk cycles are used and useful in game animation which has different needs and requirements.

02 February 2007, 10:24 PM
Anyway, after getting two more animation tutorial videos that show the static "walking in place" method, .
This is a common way for exporting to games, because the game engine taking care of global moving.
However, animating a 'true' walk cycle is much better for visualizing (it's a simply natural), and I think it's easier: foots are static or they moving, there is no sliding. And there is a proper info about global translation, so programmer don't need to move character 'by hand' :).

You can always convert a true walk cycle into the static one, just by taking upper body z translation and applying inverse animation to COG. Reverse isn't that easy as it looking at first.

How to preview a true walk cycle as static: create a null, link Z translation to upperbody. Select in order, this null, then anything you like, open object view. XSI using the first selected object for camera compensation.

CGTalk Moderation
02 February 2007, 10:24 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.