Thursday, 17 February 2011

Creating a Walk Cycle for an Alien Creature

Moving away from the musha scene slightly I wanted to try and get a walk cycle from the creature to get an understanding for his base movement, I thought of how he would walk and how much effort he would need to put in, and how long it would take him. I saw him just plodding along slowly through the jungle, heaving his weight with some applied effort oblivious to everything around him. Now I needed to apply this to animation. I wanted to find some footage to help make my animation with as few flaws as possible so I looked at animals that I thought would have a similar anatomy to our creature, from a range of small and large monkeys (like Capuchins to Gorillas) and also a few clips of lions, to see how they all walk about. Our creature has a build similar to a Gorilla but is the size of a smaller monkey and our creatures head is in a similar place to the lions, so I needed to draw a balance between to three and blend together their walks.

Because the creature is a quadruped, and I haven't really spent any time at all looking into animating them, I found this test rather tricky at first because I was trying to do everything at once, and everything looked too choppy. However once I sorted out a way to do the animation in sections, I found it easier to apply movement to the character. First I started out by getting the timing right in the front legs, then went on to animate the movement of the body (using both torso controls, shoulders and the hips control.) After getting this movement sorted I went and added movement into the back legs to finish the main movement off. When I was satisfied with the motion in the body and legs I would start the animation in the tail. (This process can be seen at the end of the post, next I'll show the key bits.)

In this clip below I had put in all the main movements I needed in the legs and the body:

The body still looked a little stiff, with not a lot of movement to imply the weight he is trying to shift from one side to other other. So I went back and emphasized this movement a bit more and to get this little swaying movement from side to side, showing us it takes a lot of effort to move his weight. After this I tweaked the snap of the feet, to make them move more smoothly onto the floor. Next I moved onto adding in the tail movement.

In the next video, while the tail is moving it feels and looks like it is rather stiff because of the lack of movement at the base of the tail:

After thinking about this and why this movement of the tail doesn't look right it is because I was taking the reference straight from the smaller monkey tail and the lion tail and because their weight is not shifting as much as the creature is their tail stays relatively still because it has no impact on the lion's/monkey's balance. Usually tails go from thick to thin from base to tip, but for our creature, he has been designed with a backwards tail. And this is why it wouldn't look right animating straight from the reference footage of the monkeys, because their tails are very light and are thick-to-thin, therefore they don't have movement in a tail to keep balance.

Going back to this scene, I deleted all the tail movement key-frames and started off by getting the movement into the base of the creature's tail first before adding anything else in. By doing this I've now got some flowing movement into the tail now, as if the base is influencing the movement occurring lower down in the tail. Whereas in the first tail test, there seemed to be no influencing movement for the big sweep being created at the end of the tail, and I felt no implication of weight. In the new tail test I needed to get this sense of weight back into the tail, and animate as if the base influences everything else.

For now, the final animation:

I like the way this has turned out, there's some nice, smooth upper body movement and tail now has some good movement to it, and has now got a sense of weight. I still feel as though the hands need tweaking slightly to to smoothen out the snap a bit more and the fingers and rotations need to be changed a bit on his right hand. For now I like this, and for next time I will animate the hands differently and keep the nice body and leg movement. While animating the back legs I think it would be great to have a bit of shoulder-like movement for the creature's bottom, to give a little bit more motion into his hips because from other angles (mainly the back and sides) his hips feel a little stiff. With this rig, I also felt a small limitation with his front fingers because I wanted to move them like back feet (a small amount of toe roll.) Due to the way he was designed he has arms and legs, therefore rigged as arms and legs, however he is a quadruped creature the size of a small monkey so needs different arms (more like legs). What I have to work with is good and if the hand movement he makes is good enough for the on-screen image then that's alright. It would be nice, if there is time for the rigger to add this control in, is a toe roll for the hands, so that the creature can walk on all fours like this monkey that you can start to see walking at 8-9 seconds in:

I think that adding the toe roll in the hands would give more sense of our creature being a real animal, but for what I have, I really like the walk I have created. Below are videos of the initial stages of the animation I did, just to show how I approached this quadruped walk cycle:

Adding in the body movement:

Adding in the back legs:

Tuesday, 15 February 2011

Musha Tests Commence

Start of the week and I'm on more testing of the rigs. Got hold of the hunter rig and after some poses and a quick play with the joints and many, many controls that get bundled in with the auto-rigger, there are some issues. The weighting is the main issue, with parts of geometry following stuff it shouldn't and a moustache that floats about when rotating his head. The other main problem is with the feet, because the control to rotate and translate is at the end of the foot there is no way to make a proper walk/run cycle through not being able to rotate the foot upwards on contact with the floor. Other than that it seems okay to use.

I have another test I've been working on this late afternoon, after working out a small scale problem with the creature rig with Matt B, where his eyes would not scale with the rest of his body and rig. After this quick temp fix, I followed some of the animatic and looked at the poses the creature had been put into and posed the 3-D rig to match. There is a small issue with the way he is drawn and the way he can be put, mainly due to not being able to squish much and his huge front which implies a huge weight so can't do some of the drawn poses. But when we get to these scenes the poses and actions will be sorted out as we go if the animators encounter problems.

Below is a test for a part of scene mushroom where I used the bouncy mushrooms from the tests done last term and the creature rig:

I based the jump on from what I remembered doing this test last term with the squirrelly rig and from the jump test I made on Friday. I like the way this quick test has turned out and the movement is flowing nice, however, he flows too much and doesn't have much of a sense of weight, too much like a small light squirrel. So I went back and tried to add more emphasis on his land on the mushroom, therefore implying more weight by making the creature lean down more into the mushroom , holding the contact for longer also and more of a push off the mushroom. Here the movement seems too jolty and stiff for the take-off jump:

Below I have added in the bounce of the mushroom and a few slight tweaks to give some refined movement, as a could-be shot and side-on. This was a nice little test to do, and have more ideas on how to improve and make the creature move much better for next time. Because I animated the creature bouncing on a still mushroom, it made it far harder to add anything other than a little more hip movement, and I couldn't get him to sink into the mushroom emphasizing him further giving the illusion of more weight. For the next test I need to make the back legs come down a few frames later than the front, leaning down more into the mushroom, more anticipation for the jump off the mushroom and the timing of the mushroom bounce altered slightly. But so far so good, and I do like how this animation flows, but to get his character across I need to make the above changes/keep in mind for the next test.

Friday, 11 February 2011

Creature Tests

After a very interesting visit by Double Negative today and a spot of lunch it was time to get back into animation and rig workings day 2. After countless tests yesterday to see where the rig could possibly go wrong and break, and if keying for animation works properly, the only way to really know for sure if the rig will work is to put it to the test. So this afternoon I've been putting our little creature into many poses and just went from there.

Below is a short test of the creature jumping, made in about 2hours with tweaks, but from this I've learnt the possibilities and restrictions of the tail's movement. Currently there's some nifty controls for complete tail movement horizontally and vertically and individual controls. Animating with just Tail_Curl (vertical rotation) was great for the starting anticipation. However to gain a real sense of weight from his tail (because it does look rather heavy) I needed the extra controls to move tail parts individually. So when it came to the end of the jump, for his heavy land, the tail naturally needed to stay relatively low after the initial impact. When it came to use these individual controls, they wouldn't key and keep the animation in place. Because I used the Tail_Curl control, I thought the individual controls weren't set up to work in conjunction with it, so I opened up a new scene and tested the tail. For some reason, probably because certain rigging connections weren't linked properly with each other I needed rigging help from Matt B our teams modelling, rigging and now some dynamics extraordinaire, to overcome this problem.

With this new scene it seems, for now, the easy way to overcome this is through creating a bunch set driven keys. So here's the process:

First we select a tail control, then in the Channel Box Edit -> Add Attribute
Name it Tail_Curl and a Minimum and Maximum Value of -10 and 10
Then we add a Set Driven Key (Always Load Driver and Driven for each new control)
Set the Neutral Pose first so Rotate Z - 0 and Tail Curl - 0
Set the Minimum value in Tail Curl as -10 and Rotate Z as (in my case) -70 for the amount of rotation
Do the same for the Maximum value so Tail Curl as 10 and Rotate Z as 70
Done and Repeat for the others

Restrictions are also evident in the hands/fingers and movement in the toes but from an anatomical perspective movements I'm thinking of wouldn't be possible. However, for that something more in the animation it would be great to have individual finger rotations for use in certain poses and for more freedom in movement/walk cycles/hand gestures just to give something a little more into an animation.

Anyway here's the first playblast I got, the first rough test:

Second with tweaks:

I also wanted the arms to land at different times and places to get the desired result, and for a character as described as a bit 'dopey' you want a bit of this to show a bit of character, so below is a video for both arms landing just a frame apart, which looks too quick and too thought out and too 'practised' a jump. Whereas putting the arms out by a few frames and exaggerating movement you can get that bit of dopiness across, like he is still learning to jump. And in the final tweak above I got what I was looking for. As noted above though, through the limitations of the rig I couldn't get the desired movement at the end for the tail, so at the minute it feels a bit weightless.