Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
Also, one thing I want to mention that people tend to miss when doing walk cycles. People tend to lean into the walk. The slower they're going, the less they lean. When they're running, they lean WAY into the run. And it's all to maintain balance. If you're legs go forward but your body doesn't also lean forward, you'll fall over backwards. You ALWAYS instinctively try to maintain balance.
That's why some advanced animation tools provide, as I mentioned before, what are called "expressions". That way you can automate your animation rig to, say, automatically position the hips at the center of the position of the two feet, or even slightly forward so that the body automatically leans into the motion. But without those tools you need to consider and apply manually those basic "naturalistic human motion" principles that animators study in depth.
Another reason why I suggest people ACTUALLY use videos of people doing stuff and study those details. Because those details are what makes a good animator out of a poor animator.
Realism is about detail.
And speaking of comedic effect with the cat tumbling down the stairs...
I kinda wish that Mr. BCRice would chime in here about how to animate for comedic effect. I'm very impressed at his comedic timing and animations. I'm sure he can give some good insights on stuff to consider.
I'm still laughing at his "Ex-Men" animation...
Hilarious.
And one more contribution to the topic...
Here's a video showing a reference walk cycle, in this case a female acting "angry". It's a good example to exaggerate the walk motion, showing how she starts slowly, leans into the walk to position her body so she can accelerate, and does the opposite to slow down at the end.
There are many such reference videos available freely that you can insert into your background as a guide when you're animating
EDIT: And here is the home page to the above source, with MANY nice reference videos for animators. It might be a nice challenge to try to replicate some of those motions manually and see how close you can come with your Carrara animations.
https://www.youtube.com/user/endlessreference/videos
Hahaha !! I just noticed that mmoir and I provided the same reference videos...
My apologies. Interesting coincidence.
Anyway, his first link to "referencereference" is much better IMO. Good stuff there if you want some nice animation guides.
Interesting discussion! Can I add a word about "naturalistic" vs "realistic". I think you want to aim for your movements to look natural, but maybe a little exagerated, even with realistic characters - and even more so with more cartoony characters. I have studied quite a lot about animation over the years and if you try to slavishly copy, frame by frame, the movement of a real person, as an animation in comes across rather lifeless and flat. Good animation involves at least some exageration of the movements (it is one of the 12 rules or basic principles of animation as set out in the classic The Illusion of Life book by Disney animators and widely quoted ever since, even for computer animation). You will also note that most animated features avoid overly realsitic human characters, in order to avoid the "uncanny valley" of something appearing not quite human, which people find unsettling. If you want ultimate realism - hire an actor! Use animation for its unique strengths, and aim for natural movement rather than overly realistic.
Very true about the ready availability of virtually any reference material you can think of on the internet!
Phil, while what you say is mostly true, and surprisingly exactly like something I would say (or have said many times...)...
I'd caution that for most here the goal should probably be to start by slavishly copying real world human motion and trying to understand what people actually do. And more importantly, WHY.
Only then can you do as you suggest, and what artists generally do, and that is IMPROVE on reality, thru exaggeration and comedic timing, etc
What I've found is that too often, animators try to go to the IMPROVE phase withouth studying or understanding the reality and reasons and physics phases. And as a result, too often you see, even in feature films, what I call "light" animation, where characters have no weight sense of presence, and as a result their motions have no "feeling". Hard to explain I suppose, but when you see it you understand.
By the way, I find it kind of funny how I always hear folks denigrate the idea of "slavishly following reality". I hear it all the time.
I realize it takes work and is difficult. Physics isn't fun. Studying isn't fun. Well, at least for some people. But UNDERSTANDING reality in human motion, lighting, textures, etc., is absolutely KEY to producing images that people like. And that's ALL people, including you and me.
Only once you understand it can you improve on it.
Fair points - except maybe about physics not being fun. And about studying not being fun - what is better than learning something new!
Yeah, but I'm still waiting for the day when someone here posts something like "hey, I was just studying the Disney animation principles, and here's my video showing my attempts to put squash and stretch, etc., into practice".
But in fairness, I've only been hanging around here for, what, 8 years or so. I need to give it a chance...
In any case, for anyone out there who really wants to study human motion, I strongly suggest you take some of those side/front view reference videos that were linked previously, and make a grid in your favorite editing program and overlay it on the video. Then watch the videos and take note of how each body part is moving, either leaning or rotating or moving up and down, etc. Especially the hips, the most forgotten and important part of human motion.
You'll start to see a few general tendencies that you can store away in your brain. The hip moves up and down vertically. The hip rotates around the vertical axis. The RT arm moves forward with the LT leg, and vice versa. The motion of the arms isn't uniform, The toes bend as the foot steps forward. The body leans into the motion. And so on.
Put together maybe a dozen or so observations in your brain, and try to reproduce them in your animation.
First off, they weren't a few side comments. They were pretty significant comments that basically said, why even bother. They weren't only from you.
Even if you don't feel the animations were "naturalsistic," there is no denying they were all fairly complex. Some more so than others obviously.
Secondly, it is odd that you focus on the cat. Clearly it is a cartoon style. A medium noted for its lack of adherence to physics. You focus on the cat falling down the stairs, yet selectively ignore the other areas, such as when the window seat springs open and the cat is flung upwards and falls down (and the saucer), and then when it is "flushed." Both effects worked very well I thought. Sure, there were other timing issues in the cat segment, which I am very aware of already, but as I said, I am not Disney, and I had literally two weeks to build the cat, rig it, add morphs, build the set and furnish it with some pre-made content, kit-bashed content, and some built-from-scratch content, texture everything, light it and toon it up. Not to mention hand animate A3, the sway of A3's dress, the camera, the seat, the coat rack and the cat.
Again, even though it doesn't meet your standards, it shows a complex project can be done in Carrara, and in a timely manner, by just one person.
Thirdly, I did also specifically say that some were better than others. You mentioned the Aiko animation. First off, thanks for the backhanded complement on the upper torso animation. I owe some of that to Sparrowhawkes Jiggle Deformer. You of course, assume I did nothing of what you had suggested previously, because you are so caught up with this notion, that has been proven wrong time and time again, that people here don't use reference images and movies without you having to tell them to. The walk was modeled after an exaggerated fashion model walk, and while I could find some side view video, real life cameras don't really produce true isometric or orthographic views, so a direct line up just wasn't possible. What this means, was that, like many animators, I had to take artistic license. As experience grows, so does the level of quality of the artist and their artistic license.
Some of the other videos are clearly not meant to be realistic either. Take the Planet Terror/Cherry video. I could balance on one leg and stick my leg out to see what movements I made, and watch the footage in the film that it is based upon, to see Rose McGowan's movements, both of which I did. The real weakness is that I couldn't simulate the kickback from the firing gun perfectly, because, and this is kinda' important, chicks with machine gun legs don't really exist.
I put the videos out there, warts and all, to rebut the "discussion" that Carrara can't do, or handle character animation. I am more than willing to show how I did all of the things I did if anybody is interested, as I said earlier. It is probably easier to focus on one thing at a time, such as how to set up targets, or how break down the walk, what steps to take to make V4 ride a bike, etc.
Evil said: "Clearly it is a cartoon style. A medium noted for its lack of adherence to physics."
And that's the basic point where I think we disagree. Completely.
If you study cartoon style, it is an EXAGGERATION of physical properties that is generally employed in cartoons. When a character gets hit in the face with a frying pan, the response is a super-exaggerated "KABLAMMM !!!", and the character shakes wildly, his face gets squished flat like a pancake, etc.
Honestly, and once again, I suggest you study the subject. That's not a bad thing, and not a criticism. Cartoon animation is not just a matter of replicating a a kinda, sorta character motion, that looks a little bit like how a character moves, it really is an art, a combination of physics and art and comedy, an attention to detail, a consideration of timing, and so on. To be good at it really does take some study and practice and talent. Again, not a criticism, but one of the most common misconceptions I see in hobbyist forums is the assumption that anyone can pick up a 3D package and become a good animator. It's just not that easy, at least if you want others to enjoy what you produce.
To illustrate my point about physics being important in how we perceive animations, here's a simple animation of a ball falling to the ground.
http://youtu.be/wAG5OFPh62Y
Now, how tall is the building it was dropped from? Anyone?
(BTW, this was done using the Carrara rigid body physics simulator, and this is the raw render)
Thanks for all the helpful links. Using reference materials and observing human movement is very useful but there is a gap between doing this and harnessing animation tools to recreate it convincingly. I could study many of these films and still not know how to transfer that knowledge to Carrara or DS. That gap in knowledge is what I'm seeking to bridge. Perhaps what I'm referring to is similar to taking life drawing lessons but not being aware of techniques like cross hatching or mediums like pastels. Without that technical knowledge observation is more difficult to translate.
The way the human body moves is also related to emotion. For example, a sad or concerned person will sit down in a different manner to someone who collapses in a chair with exhaustion or sits on the edge of their seat in anticipation or anxiety. I've also observed that when people walk it's not completely smooth or fluid but subtly jerky so maybe ultra smooth movement is actually unnaturalistic? Noting these differences is undoubtedly very important but how do you translate them to a project in Carrara? What tools create smooth transitions between keyframes, emulate physics or make it easier to bring weight and expression into a figure and how do I use them? That's what I want to learn.
Concerning naturalism - my aim is to replicate as close as possible human movement even if this does stray into uncanny valley territory. CGI is used in films more and more to replace actors in certain scenes such as dangerous or impossible stunts, crowd scenes or, sadly, to replace an actor who has died before production has finished. It's an area of animation that I want to become more skilled at. I think toon animation is great and has its own complexities but that's not what I'm trying to achieve.
One thing I thought of is whether there are any post-render smoothing tools to help an animation flow better? I played around with Twixter in Sony Vegas Pro which can shorten or stretch video. I thought that by maybe adding extra interpolated frames to a slowed down animation sequence it might have a tiny smoothing effect on the animation when it is speeded up to normal speed again but so far this hasn't helped. I also tried Boris FX Motion Blur plugin on "subtle" setting but this didn't help either. Are there any other tools out there that might help?
Superdog, perhaps if you can give an example. You're asking about 25 different questions, and honestly it's not fair to have people chasing a topic in one direction when in fact you're looking in the opposite direction. One minute you're talking about feet sliding, then the next moment you're talking about naturalistic/realistic human animation, then about smooth transitiions between keyframes, then NLA's, and so on.
Do you have one particular example of a problem you're having? Maybe an image or a video would help us figure out what you're getting at.
Hey,
I will start this learning exercise by posting a quick video on setting up IK Targets for the Genesis figure so everyone is using the same character set up so the NLA clips work correctly. I will post this video later tonight.
I'm experiencing all these problems and was encouraged to start this thread to continue a general discussion about animating figures realistically that was sending another thread OT. One of the points of this thread is for semi/experienced animaters to share how they have solved specific problems they've encountered when trying to animate realistic figure movement. I can't post the animation I'm working but will try to upload a clip for feedback in a few days. Any animation project I've worked on invariably contains multiple problems which is why I'd like to learn how to solve not just sliding feet but all of them. In the meantime perhaps we could try the sitting example?
I guess you use Animate 2 in DS? When you try to join 2 aniblocks, you have to make sure that the movements you want to join are pretty similar. You have to crop your aniblock until they are (feet in approximately the same position, etc....). The aniblock transition can only do so much on its own. Also, remember that you won't see the final result of the transition until you join the aniblocks (see pic) or save to the timeline. You have to select both aniblocks and press on the "join aniblocks" button. If the transition is still a bit rough, after importing the animation into Carrara as a DUF ( my opinion is the aniblock importer is not very precise), there is Carrara's graph editor with its awesome tools found nowhere in Daz Studio.
I import aniblocks without translation using the plugin to join them
but since I make sucky videos and break all the rules not the one to ask.
Hi Superdog,
If you use NLA clips, you can click/drag the clip in the timeline to either shorten or lengthen it. You can also use the Animation Menu to add multiple NLA tracks to single character so that they can be layered. So for example, you could use one track for a walk cycle, add another for an eyeblink as Phil suggested earlier, and perhaps another with a movement to correct for a sudden transition between the clips in the track used for a walk cycle or other animation.
To avoid foot slippage in a walk cycle and translate the figure continuously forward when converted to an NLA clip, I create a simple walk forward that contains one step for each foot, making sure that there is absolutely no foot slip. Use the hip bone for the forward translation and not the top of the figure hierarchy.
When I am satisfied with the walk, I then create an NLA Clip using the hip as the offset. This way, when using the clip's loop option, I enable the offset and the figure continues the forward movement from the loop point without jumping back to its original position.
When animating a walk for a DAZ figure, I primarily use the side and front views in a medium scaled scene. However, I go to the view menu and change the default grid size from the standard 1 foot grid (if using US measures) to a .5 foot grid. This gives me a better visual handle on foot position, both forward/back and side to side.
To set up the figure for a walk, I sometimes just use the rotation tools to rotate the limbs, and sometimes I use target helpers. There is no wrong way. The one thing I do in all cases is to add cylinders parented to the hip and to the shoulders. For the hip, I place a simple cylinder primitive stretched and thinned to look like a long rod. I place it at the junction of the thigh joint and hip, going across the body through the figure's body. I then duplicate it twice and turn the duplicates so that they point forwards and I position one at the right thigh/hip joint and one at the left thigh/hip joint. This gives me a better view of how I am rolling and tilting the hip as they are parented to the hip. I do the same thing for the shoulders. I know it sounds complicated, but it actually only takes a couple minutes. If I am having a hard time differentiating between the rods, I just create a couple simple master shader with different colors. When I am done with the walk creation, I delete them before creating the NLA clip.
If I am making an NLA clip, and because I have an older computer and an old graphics card, to help performance when I animate a figure, I do not dress it, and I change the image map based textures to the default texture by dragging the default shader, shader ball from the shader palette onto the model's multi-colored ball at the top of model's shading domain list. This replaces all the shaders with the same simple shader. Since my intention is to create an NLA clip that will be saved to the browser, the actual figure's shading is irrelevant.
Thanks so much for the technical information. I find it really helpful to learn how others tackle these problems. There are some techniques mentioned that I've never really explored such as bridging blocks. I'm wondering if turning poses into clips might work as bridging blocks? I might even try turning a keyframe at the beginning of a block into a pose block that I can use to connect two blocks in cases where there is a gap in the transition. I'll also try importing into Carrara as a DUF and edit using the graph editor. I'm applying some of what I've learnt so far and already there are visible improvements in the animation I'm working on. There are parts of the animation that look very naturalistic. Probably because I've worked very hard on setting up the lighting in Octane Render to create lots of light and shade to really enhance the contours of the figure which is a G2M morph of my own design using HD details. I'm also using the RedSpec OcDS shaders for G2M. It's finally entering uncanny valley at certain points but still a long way to go.
One of the techniques I've used to enhance realism is to research the way Rembrant and other old masters used light and shade in their paintings. A combination of HDR and emissive lighting can create some amazingly realistic scenes that can actually enhance a slightly dodgy animation. If there's a bit that isn't flowing right instead of panning the camera away design the light so the figure is obscured partially by shade at that point. This can work for internal and external scenes at any time of day.
Okay, well, now you're talking about "uncanny valley" and emissive lighting, so I give up trying to understand what issues you're asking about.
So I'll just proceed with my previous train of thought in case someone here is interested in the title of the thread "Figure Animation Tips and Tricks".
The simple video I posted was an attempt to get people to think about physics, and how bodies and characters and objects respond in the real world. And how we as viewers PERCEIVE things.
The video shows a simple ball falling under gravity. And if you look at the video, you might notice the speed the ball is falling. Either consciously or subconsciously, we ALL sense or feel something important about the scene based on that speed. Anyone know what that is?
Well, since I doubt I'll get a response, I'll answer it.
The ball fell from a relatively high place. In this simulation it was the top of an 8 story building (approx. 80 feet). And we can tell that by the way it appears the ball is falling relatively slowly. And that's very important. We INFER a lot by what we observe. If the ball fell faster, we'd infer it fell from a lower height.
Also notice that it starts out falling slowly, then speeds up. Again, from school physics, we learned that all objects, no matter their size or weight, fall with the exact same acceleration. And the graph of the object's Z position over time is shown below
My point is that if we want "naturalistic" movement of any figure or object, we need to be aware of how viewers perceive things, or else we risk losing them.
So if our character is jumping of an 8 story building, we expect that it will take a little over 2 seconds for him to hit the ground. If it takes only 1/2 a second, we don't buy the idea he jumped from 8 stories high.
And this goes with just about any animation you're talking about.
When he's jumping off the building, we expect his leg to push off the building ledge, and then we expect his body to accelerate forward from the push. And when his body accelerates from that push, it starts slowly as the leg muscle accelerates the mass of the body, which takes time.
So in almost all real character motion you see the graph I just posted in effect. Things start off slowly, and then accelerate. When you're walking up the stairs, your legs PUSH your body up each level. When you fall under gravity, you start off slowly and then accelerate.
It's that basic core motion that needs to be considered in just about anything you animate. It's what makes viewers "feel" the motion. And unfortunately it is one of the least-considered aspects of much of the animation you see.
And once you have that core motion, you can then improve upon or exaggerate that motion depending on your goals.
Nice work Evil! Not bad for a farmer ;)
And the good thing about most animation software is that it gives you a default mode where it interpolates (aka, guesses) motion between keyframes using a curve very similar to the graph I posted. Often it's referred to as "Bezier". When a keyframed object or bone transitions from one keyframe to the next, the software assumes that it will start off slowly, like real life stuff does, and then accelerate. Cool.
However, you need to be careful. Sometimes you don't want stuff to move like that between keyframes. Sometimes you want it to stay fixed. Or move in a "linear" fashion.
You just need to be aware of what parts of your character move in what manner. Hands and feet tend to move very quickly, with almost instant acceleration. And feet tend to stay in one spot on occasion.
Typically, the real difficulty with doing naturalistic character animation is that it requires thought and planning. Lots of it. And thinking about movement and timing and physics. Dry stuff that many people consider work, and not fun. But necessary nonetheless if you want to do a good job.
So you take your animation in small steps, and plan what your character will do, how he feels, what the physics of his situation are, etc. That's why it's good to see videos of actual movement. It helps you visualize all of that stuff. Because typically, none of us is really good at bringing up and recalling and understanding all of that tedious detail.
And one of the biggest immediate challenges is to figure out where the "key" points in each small segment of motion will be. Those are where you place what are called "keyframes".
But where do you place them?
Well, it depends. Generally, you break down each small section of motion into beginning and ending of each section of motion. For example, when you're walking, one foot gets planted on the ground while the other leg "passes". And then it gets picked up while the opposite foot gets planted. So each "planting" and "lifting" point are good places to create keyframes.
So one of the most critical procedures of any character animation is to figure out in your head where you will place keyframes, and allow the software to interpolate between those keyframes for the motion.
And that happens after you've stepped thru, in detail, exactly what motions you want your character to perform.
Lots of dry, painstaking detail for some people. It's certainly do-able in any app, including Carrara. The question is, am I willing to put in the undertanding and effort, or is it best to drag-n-drop?
So for example, let's take a walk cycle.
You KNOW the easy parts, that your left foot will be planted on the ground for a period of time. So you know you'll need keyframes at the start and end points in the "plant". And between those points you don't want any fancy Bezier or nonlinear motion, just a flat interpolation where the software assumes it stays put.
And you know that the opposite foot will do the same thing, but at a different time.
And the best way to do that? Well, you use an IK target. What is IK? Well, that's a whole different subject, but suffice it to say that when you grab something with your hand, your hand instinctively goes to the object and your arm bones and muscles follow. That's IK. Bone chain follows a target point. So you add an IK target for each foot, and the leg IK chain will always point at that target and not go moving around between keyframes. And when you raise your hips, say from getting up from sitting in a chair, your feet stay planted as they track the foot goals, but your leg bones react naturally as the hip is raised.
So next, if we focus on the legs, you need to study what's called "locomotion".
Here's a good example. Stand up from your seat right now, stand naturally with your feet planted on the ground. Then rotate your right thigh bone out maybe 20 or 30 degrees, so that you're standing on your left leg. That's the first move you'd make in a walk cycle, correct?
Well, no.
You'll quickly notice that you're just standing there on one foot with your opposite leg extended. You're not going anywhere. You're not walking.
Why? Because walking is NOT about leg movement. It's about BODY movement, and the legs only keep you from falling to the ground. You first need to lean into the walk.
Everything in character animation happens for a reason. It's not just about duplicating some leg rotations, it's understanding WHY that stuff is happening. You first need to animate falling into the walk, which means your hips move forward. And since the leg is also moving forward, you are no longer standing upright. You're getting shorter as your legs SPAN your walk. Which means your hip not only moves forward, it drops on the vertical axis a bit.
So the hip bounces up and down. And it also twists a bit around the vertical axis.
And when the legs are spanned the furthest apart they need to exert effort to get your body moving forward. And so on...
And also keep in mind BALANCE. In fact, walking is an incredibly difficult balancing act. Seems easy and simple cuz we're doing it all of our lives, but if you ask any of the guys who program (or try to program) robots to do bi-pedal walking (two legs), it is incredibly difficult. And the biggest challengs is BALANCE. You have a 180lb person standing on two sticks, and those two sticks are all that keeps him from falling over.
Understanding BALANCE is absolutely critical to understanding character animation. Your body and your brain are always maintaining balance so we don't fall over. Our body is always centered over the span of our legs so we maintain balance. Or, depending on our intertia, we may be leaning PAST the point of balance, and leaning INTO the motion.
But those tiny intricate details are, like so much of 3D, what sells the concept and makes people believe they're watching a guy running, or jumping off an 8 story building, or giving a roundhouse Ninja kick to knock some guys head into tomorrow.