Walkcycle workflow question

I am rather new to the whole world of Actions and the NLA editor, but I have now reached the production phase in my short (see wip here) and I wanted to start in the correct direction using a proper workflow.

One question which I came accros is the following:
Lets assume I have a walkcycle for my character and I use the NLA editor with its cyclic option and the modifier system to layout the path the character is going in a shot.
Whats necessary in the shot is of course to refine the animation with all sorts of variation, responding to the environment etc.

Doing this with multiple blended actions will work partially (I think) but is not a final solution.
What I would find very cool, if there was a way to ‘bake’ the NLA strip back to a new action and do the refinements on this newly created action.
The other way would be to do the whole walking by hand, which seems quite a tedious and repetitive task.

Is there any good workflow on how to tackle this task of doing a longer walk with refinements.

Thanks for your help.

Loramel,

to be honest, I have never been a devotee of the walk cycle, because unless you have a lot of very simple walking about to do, and are very short of time, so you have no option, you are better off to just animate the walk directly.

As you say, the problem is that you need to ‘tweak’ the walk to match the environment, and that takes almost as long as the original walk animation itself. People rarely walk more than a couple of steps with the same exact gait, and in most real circumstances, such as indoors or over rough ground, most of their steps are accelleration and decelleration, twisting and turning. Walk cycles assume that every step is the same length, just faster or slower, whereas in reality, people tend to take shorter and quicker steps when moving slowly, - particularly when positioning themselves to do something else like lifting or reaching. It is that change that makes great animation look so much better than quick and easy stuff. Compare Shrek with ScoobyDoo, as an example.

Looking at an average shot length of say 10 seconds, a person if walking continuously, take 10 steps roughly. This is 30 keyframes if the walk is simple. That really isn’t going to take that long to animate directly. Remember you can cut and paste keyframes directly in the action editor, so you can select a group of events and just slide them along the time line then tweak them to the new locations. In the time it takes to perfect your walk cycle so that you don’t get feet sliding all over the place when you start to move the armature, you could have posed those frames.

I personally don’t even move the armature, - only the bones. The armature stays where it started and only the bones controlling the character are manipulated. This totally eliminates any chance of foot slippage. I just animate the feet first, then make the hips follow, and then the hands, and then tweak the head and shoulders to get the feel of movement. OK, I have very limited experience, but it works and it’s fast enough. The downside to that is that you cannot copy a pose or invert it, - the pose is stored relative to the armature location. The upside is that you get more realistic movements, because it eliminates the ‘character sliding smoothly through space’ feel that many animations have. this is because the armature is going from a to b, and the limbs are waving around en-route, rather than each limb going from a to b, and taking the body with it.

As I say, I have very little experience, other than a lot of testing, but the results are at:
http://blenderartists.org/forum/showthread.php?t=176528

However, if you must do a walk using the ‘move the armature and superimpose a walk cycle’ thing, then just cut and paste the keyframes in the action editor, and then you are free to tweak them as you like. You can also scale them to speed things up etc.
Matt

Travellingmatt, thanks a lot for your detailed reply.

You made quite some interesting points.

Using a walkcycle directly is out of the question as it is too repetitive. My thought was to have a solid base to start with, but your point about different stride width and such makes me think that you are right about the amount of time necessary to adjust the base action.

Is set up my current walkcycle using the same technique you mentioned - keeping the armature in place and just move the bones. I found out just as you said, that this gives you much more control over the positioning of the feet.
I used then the offset bone feature i the NLA editor to create a walk over several cycles. See here for the result.

Before this walkcycle I did one using no cycling at all just animating all by hand. This seemed quite a lot of work, but then I was just starting animating.

I’ll give your method a try and see how I can copy with it. Thanks again.

btw: I found some hints and discussions concerning the current 2.5 development about animation baking, but even with a current 2.5 snapshot I was not able to try it out an see if its really convenient.

Travellingmatt,

Checked out your animation. Pretty cool. Your points are spot on. Also I don’t think armature is really designed to be moved. I mean, none of the great animators I have known of do that. And I learned from one of the best, Timothy Albee, to just keep it in place.

So you are on the right track there.

I have gone over the walk cycle subject a little in this thread here:

http://blenderartists.org/forum/showthread.php?t=175517

I would be happy to illuminate any points if you like. But in general, you guys are on the right track. Walk cycles should be flexible and allow for realistic movement.

Thanks Richard.

The link you provided has some great information in it and just confirms what I was researching myself.

As I have no formal education in any CG area my approach is somewhat unbiased ( this being good or bad is another thing :slight_smile: ).

My background is more technical with a deep founding in physics, so what I recognized from intensive studies of videos from my walking daughter, that walking is a highly physical process and that its an extremely energy efficient one too if you are having a normal walk.

E.g. the conservation law of angular momentum is what makes the counter rotation of the upper body to the rotation of the hip a must, otherwise you would invest too much energy.

Anyway thanks for your pointers and the encouragement.

No problem. It is just a matter now of applying that information to the mechanics of a CG character and translating it to key frames.

Let me know if you need any help with that part and I look forward to seeing more tests when you get them.

Thanks a lot for this generous offer. I will gladly come back to you as soon as I need advise.

( … I really have to start a list for all the entries in the credits of the short …)

I think that when doing a true walk cycle, the armature IS intended to move. You keep the armature still while creating the walk cycle, so that it walks on the spot, then move the whole thing through space at a speed that keeps the feet still during the backwards moving phase. That’s how the book I have shows it. Trouble is it really doesn’t work if the character turns or changes speed.

I found that the trick is first to place the keyframes for the feet hitting the floor, then add copies of these to hold the feet still. Then move the whole body so that it has a keyframe for each passing frame and each extreme frame, and finally add the passing keyframes for the feet and any intermediate ones if the path is not quite right. As soon as you have the foot controller keyframes mapped out, you can run the animation to see if the timings are right, then do the rest later.

Interestingly, I read an idea from a professional animator who suggested that you could hide the legs of the model and skip the entire walking part of the animation so as to get the body and the arms right first, then do the walk stuff afterwards. He did suggest it was an advanced technique, but it is interesting. It does place the emphasis correctly. After all, in reality you don’t think about your legs- you think about getting your head and/or you hands to the right place, and your legs waggle around without any conscious thought to achieve that. If your animation places the feet and legs first, the position of the head and hands will often be less than ideal. There are cases, such as a person walking along talking to another person, where the actual location of the head and hands is unimportant, but generally the character’s aim is to see or do something, and their legs get them into the right position. I may need to think this through, but it does make sense to try to design the rig and/or mesh so that you can work backwards this way. CGI does have the advantage of allowing you to go back and forth in time adding keys.

Sorry, thinking as I was writing there, bad habit!

Matt

Actually I think in the end its easier to keep the armature in place and just move the controls. This mimicks more closely the behaviour of a real walking. You move your foot and place it down on the ground, where it stays until you move it again. There is no counter action to null the backwards motion. I found out that as soon as you start to
combine movements to achieve one effect you have a hell of a lot of work and fiddling.

But then, I am just beginning with all this stuff so you may easily ignore me here :slight_smile:

I have come across the same suggestion. I think Roland Hess mentions it in his book.‘Animating with Blender’. Its interesting to mention that in a dance course I attented just a couple of months ago, the instructor told us that to initiate a movement or change in direction the body moves first and then the feet follow. This would correlate with that technique. So you make a controlled body unbalance and counter with the feet to avoid falling.

Sorry, thinking as I was writing there, bad habit!

Matt

LOl.

Not a problem.

Actually, lets clarify one thing. By move the armature I mean in Object mode.

You should have a rig that allows you to move the whole body from the root. What I am referring to is moving the Armature Object. You should also be able to move the root bone leaving the armature objcet in place.

But regardless of that. You are right. Many books and tutorials teach that walk cycle. But beyond games, it is useless for good animation.

Anyways, I think you guys are on the right track here.

Another thing to think about along the lines of what you were thinking about. How many times are you actually going to see the legs and feet?

You should be able to do a good convincing series of poses regardless of the camera angle. However many of the shots won’t even include the legs and feet.

Bingo.

Lets zero that in even more. It is the characters thinking that is key. They intend to move for some reason, then the body follows - sometimes head and hands first. How many people trip and fall when trying to get out of the way of something?

Or a baseball player backing away from a high wild pitch? First thing he does is get his boby out of the way, feet still planted. And many times they topple to the ground as a result of imbalance.

I think you just have to get walking out of your system as an animator. It is really a technical thing good to have under your belt. Then use it when needed. That is why it is best to learn the most flexible method.

As said in those links. Walking, standing and dancing are more or less the same mechanics and should be animated the same.

Hi all,
I’ve been reading this thread since I noticed the first post. It has enlightened me. I have a character with various actions like - rest, rest2walk, walkcycle, and walk2rest. I combine them in the nla editor and the character goes from standing still (rest) to walking and back to standing still. The animation all looks very mechanical and I was blaming that on my lack of polish for each action. Like loramel, I have no training in CG animation.

Now, I am re-thinking my approach. I understand how not creating a walkcycle and repeating it, but instead animating the whole walk the character should take would be better. Not every foot placement or arm swing would be identical and not mechanical looking, but more natural looking.

So, here’s my question. Let’s say I have a scene where a character walks up to a table and grabs an apple, takes a bite of the apple, sets the apple down, and walks away. Should this be animated in just one action, or broken down into separate actions so maybe they could be used over again at some point later??? Like, say 3 scenes later, the character walks up to a table, grabs a baseball, and throws it. So walking up to a table and grabbing an apple or a baseball would pretty much be the same actions and motions. Would it be better to animate each action, walking to the table, grabbing the object, and repeating them 3 scenes later, or just animate each action as one big individual action? Walk to the table and grab the apple, then 3 scenes later, make a totally new action of walk to the table and grab the baseball…

Thanks,
Randy

Personally I would just treat the whole scene as a single action. If you are not animating using walk cycles et.al., then the whole action/NLA system is overkill.- Just start at the beginning and keyframe you main poses, then fill in the gaps till it’s done. If you need to re-use part of an action it’s easy enough to select, say the body and leg controllers and cut and paste the keyframes to your new file, then add new arm movements. It sort of defeats the whole idea of making every movement meaningful though.

What I mean is this: In my opinion, the idea of the animation is to convey not just action, but emotion, thought, life. If this is the case, it is going to be very unlikely that the character would approach an apple with the same attitude to approaching a baseball. There is an anticipation of what is about to happen in each case, that permeates the preceding action. Someone reaching for an apple to eat would position themselves differently to someone going to throw something, and that doesn’t just happen as the person picks up the object. It is the action of throwing or eating that is in the person’s mind before they ever walk to the table, not the action of picking up. Picking the object up is, like the walking, only a means to an end. A similar issue occurs with both expressions and speech. An expression, either facial or in body language, always precedes the speech or gesture that relates to it. You will ‘act’ mad, then get a mad expression, then shout at someone. It never happens in the reverse order. Equally, it never occurs in the same keyframe. There has to be thought behind the action, and the thought shows before any overt action on the part of the character.

I’m kind of off topic a bit, but you can see how this whole process gets defeated by the idea of cut and paste of movements.

Matt

Cool. Actually OK, A bit off topic strictly from a walk cycle. But not so when you consider that these same things need to be taken into account.

I actually agree that I would treat such an animation as one action. I am not much into automation. For the reasons stated above.

But also I always try to drive this following point home to animators.

Think in terms of shots.

If you are putting together a reel to show that you can move a whole character that is one thing. In this case it is still a good idea to show a variety of shots. And I understand from an academic perspective, one is working toward mastery of the full character. An in these animation contests, you often see here and on other boards, the animators usually show the whole character the whole time through every action.

But when you get to the job - be that a position as an animator in a group project or one of your own - you will be handed a shot from the director. Even if you are the director. Story and direction come first. Storyboard and animation come second. In that order: Story, Direction, Storyboard, Animation.

The reason is because film has a language. Shots are like words and strung together make sentences that can convey entirely new meanings than the words themselves.

If you only show wide angle shots, you are not using the entire vocabulary of cinema.

So although academically speaking one should be able to walk over and pick up an apple in one continuous action, this is only one kind of shot.

So animation should be thought of just like film - a series of shots to convey a meaning.

One might have Character A walking for 2 or 3 seconds.

Cut to a reaction of another - Character B.

Cut to a hand reaching for the apple.

Cut to Character A anticipating eating the apple. (apple and hand out of shot)

Cut to Character B wincing as you hear the apple being bitten into and chewed.

Cut to the Character A spitting out the apple chunks.

Now you put this sequence alone. It can mean anything.

Start the sequence out by first showing Character B who was reacting having just seen a worm crawl into the apple then put it down and walk away. Then he hears someone enter (Character A) , he turns and looks. Then the above sequence has a new meaning.

A bit off topic but the point is, to convey meaning in animation just as in other forms of cinema you do so by shot selection and what action is shown in each shot.

More meaning can be conveyed in some cases by what is not shown. You can cut out parts of action to move things along. Does the audience really need to see the character walk all the way across the room and pick up the apple?

No.

You could even have a character come in and look. Apple in foreground on table.

Shot on apple.

Hand enters frame grabs apple.

Shot on character putting apple to mouth.

In this sequence the walking and other action is completely removed. The audience can accept that the character walked over to get the apple. They don’t need to even see this. They don’t even need to see a wide shot of the table and the character with the apple. This can all be left out to save time and to convey specific meaning.

This is a very common technique in filmmaking and animators can use it to their advantage.

If something is very hard to do all in one action. Figure out ways to break it up. Don’t be a purist. You may find there are more clever ways to show something and convey more meaning.

In short, film rarely happens in real time. That is boring to an audience. They don’t need to see everything all the time.

As an animator/storyteller use that to your advantage.

It may be a little bit off topic but I for my part do not mind, as these are some very interesting aspects being discussed here and in my case they are spot on my current project.

I just finished the book ‘Directing the Story’ by Francis Glebas and it opened up my eyes about the usage of visual language and the workings of filmmaking. All the points made by Richard are portrayed and emphasized there too. The effect was that I reworked my storyboard massively and it now lays a solid foundation for the actual shot planing and thus for the animation part.
There will be a lot of shots where you don’t see the whole body, or have such a wide shot that details will not show at all.

I agree fully about what Matt said about emotion, intention and life and thus repetitive animation would not be fitting for character animation. If one looks at the root of the word animation it come from the latin word anima which means soul and thus animating is putting soul into an inanimate object.

Though for all sort of other animation work (mechanical, physical simulation) the NLA/Action editor can come in quite handy.

Cool. Good to see you are thinking in a cinematic fashion. If you don’t already, consider doing a lot of shooting with a real camera of live action. Try getting some people together to act out some basic actions, then cover it from many angles and then play with it in editing. Forgive me if you are already doing or have done this. But thought I’d just toss it out there.

Of the many film making books I have read there is one that is very good to get specific on the editing process from a creative, emotional and story telling point of view.

Here is is on Amazon. (not spam just a link I dug up from google)

Looking forward to your progress and I hope you post things on the forum as you go.

Richard, thanks a lot for the link. From a very first glimpse at its content I think this book is what is still missing from my library and may turn out as enlighting as the other one.

About your tips on shooting with a live action camera I have planned this and already done a bit of it but more for a reference for animating. I think it is a good idea to widen its application to angle selection and editing.

I document my progress in my wip and on my blog for everyone interested to see and maybe even give feedback.

On the subject of how camera shots work, and cinematic conventions: I am part of the OWF movie team, and there is a discussion on that subject on their website at:

http://openworldfilm.com/Forums/viewtopic.php?f=54&t=702

Which covers a lot of this type of stuff.

Also take a look at a lot of the storyboards that are on that site, ( Many by myself I admit) and you can see how many problems of CGI movies have been addressed. Choosing the shot is not just a matter of directorial whim, but often of necessity, as there are things that CGI does not do well, or could or should be avoided. The entire script was written with the minimisation of dialogue, character count, sets, and organic modelling. The original script had 32 characters and I think 5 sets ( with trees, distance shots, passing vehicles etc) in the first 4 minutes. The later one had something like 1 interior set and a couple of space shots and three characters.

The script in production at the moment is just a segment of the whole movie, being a flashback, but is self-contained, so is being made as a test.

Matt

Cool. Yeah, I’ll take a look at the link. You are spot on about the necessity angle. Big part of it alright. Then the challenge is to tackle these things in a way that does not look like you had to. Part of that is in style decisions and so on.

Hey guys, I just wanted to thank you all for one of the most interesting threads I have read in awhile.

I am a 3D hobbyist, and I have a tendency to run from one tutorial to another and try to “learn” without ever asking: “Gee is this something I should be doing?”

The idea of a walk-cycle was always a gimme for me. It is mentioned in virtually every animation book I have, and in most animation DVDs and classes I have seen. Of course we should be doing walk-cycles… right?

Until I read this thread, I never even questioned it. Now, after thinking about it… it kind of makes me laugh. You all make really good points as to why they really are not necessary.

I have a theory, btw, as to why the walk-cycle remains so universal: Because once upon a time, they were essential.

Early Disney, Warner Brothers, etc, made use of Walk-Cycles to fill in time cheaply. If you look at the old cartoons, they do have lots of places where the character is just walking (or running) along and the background is rolling behind them. It was a cheap way of getting animation on film, in a day and age where everything was so man-power intensive. (Hell, the idea of tweening hadn’t even been invented yet!)

So, 2D animation invented this technique, and it has been passed down like an old pair of pants that are hopelessly out of style.

We do it because they did it. It is tradition. Even if it no longer makes sense.

I don’t know much about animation. But I do know about writing, and the funny thing is we still see Radio’s influence in today’s TV scripts.

Ever notice how often a character will say the name of the person he is speaking to? I was watching Burn Notice the other day, and it is really noticeable. If you had a drinking game, and took a shot every time Sam and Mike say each other’s names, you’d be drunk before the third commercial break.

“Sam, can you check…”
“Yeah, Mike, I did, and you know what…”
“I know Sam, but I need to…”
“Ok, Mike. I sure hope you know what you are doing.”

No one talks like that in real life. Think about a conversation with your friend. How often do you ever say their name? Especially if there are only two of you in the room! It’s not even close to “real dialog.” So, why do TV writers do it? Because in early days, when the best writers in the biz were writing for radio, they had to put those names in there so the radio audience could follow along.

When that group of writers started writing for TV, they just continued the habit… and now here we are 50 years later, and TV writers are still doing it. It can’t be habit any more… now it is a hand-me-down technique that no one really thinks about.

I think it is the same with walk-cycles.

And… when you think about how long that strange radio script thing has stayed with us, I imagine that animators will be working on walk-cycles for a long time to come!

Thanks for the thoughtful discussion!