Conclusion and Reflection

My animation in my opinion wasn't too bad, but there are a few things wrong with it. The speed of my character, like I've said previously, was a big issue for me. I just couldn't get him to run at a realistic speed because of my error in over populating the dopesheet.

I had to constantly fight errors in the sense of those annoying rotation errors, where I had to use euler filtering on each rotation of the arms. It produced ok results but the arms were still not fully under my control and therefore I feel I could not produce the exact movement I was trying to animate.

I think the rig I used was possibly too basic; it didn't have many capabilities in terms of actual movement which I think was producing those rotation errors. What I mean by that is this rig as far as I've researched can only be used for forward kinematic animation, which is fine, but doesn't possess the smooth and less error prone movement of inverse kinematics. I've used inverse kinematic rigs before and they have no issues with limb rotations as it doesn't really matter what pose the rig was set in, it always moves from its last rotation without its  first transforms interfering with its new transforms. A rig that can blend between the two would have been excellent, but this was the rig that our class was advised to use so it was best to stick to that advice really. Its good for very basic animation, but when you apply movement and rotation it has a tendency to produce unintended results.

Throughout my animation process I have tried to keep a close watch on important animation factors. For example silhouettes; I've made sure that my animations have as strong silhouettes as I could make them have.

The only time I thought I could really show anticipation was in the start where the character is getting ready to jolt up and run. You can see the momentum in this area too when he thrusts up, his momentum starts at the hips and travels up the spine to the shoulders where you can see his body squash a little.

My choice in animation doesn't really have the need for rules of squash and stretch in it, but I have made sure to use those rules when I could, like when the character is tucked in the start position then jolts upright.
I've tried to keep my character looking balanced at all times, making sure his weight is correctly placed across the limbs and his centre of gravity is kept at a point where he doesn't look as if he is going to fall. It was hard though as the videos I've seen of Olympic sprinters shows them lifting their feet off the around at the same time in some parts.

All that being said, I feel I've learnt a lot from this assignment and particular animation assignment. I will be more aware of my dopesheet layout, as well as going about movement in a mathematical way of thinking. For example, next time I do a run or walk cycle, I think I'll research or record myself running or walking and note down how long it takes to move so many feet, then set up those values in Maya's grid units so I can more accurately animate.

I am most interested in animation within this course, and even though my results on this assignment were not great, It has only strengthened my interest and enthusiasm in it and I will be doing many self appointed animations to bring the quality of work up in the future.

Final outcome

Here is the video of my animation as its final cut. I will post about my reflections and conclusion on it soon.

Front view render:

Side view render:

Perspective view render:

Real time speeds

This module requires that we animate in 24 frames per second (real time), and therefore I set it to that playback speed in Maya. Maya already recognises that real time is 24 frames per second and so there was not many settings the required changing.

After animating and putting the final tweaks into my animation I realised that my animation after the starting blocks was too slow in some parts and too fast in other to look realistic. To try and correct this I used the 'Dopesheet'. I spent a fair few hours trying to select and move the keyframes left and right to speed up and slow down the animation. The result was better than what I had to begin with, but it was still looking quite jerky and I couldn't figure out how to resolve the issue. The only way to do it that I knew of was to use the dopesheet.

I had been taught to make sure when I move part of the rig to keyframe all of it so that on the dopesheet you could see thick black bars across all of the different parts of the rig. It was great advice as it makes it easier to see what is happening, and makes sure that no one part moves randomly by accident due to movement further down the line in the time slider.

I think though, what with trying to correct random arm rotations and such, that I over complicated the dopesheet needlessly. By this time though It was far too confusing to sort out at the end and clean up / simplify.

Something that I'm definitely going to keep watch on next time I animate.

The below image illustrates what I mean by over complicating the dopesheet. When I look at it I feel I've made it too intimidating to try and clean up without destroying my animation keyframes.

Error Prone Rotation

Once I had got the legs animating as they should, I began working on the swing of the arms. After I has set the keyframes for the arm rotation I noticed they were not doing what I wanted and expected, instead of rotating in a manor you can expect from when you run - in a semi-circular motion - they rotated out and back in along the way. You can see what I mean from the image below:

In this image the arm is in mid rotation as you can tell from the graph editor, yet it is not by the side of the hips like I expected to happen, it has rotated outwards out instead. I believe this is because the 'Max' model I am using was rigged in a T-Pose, and because of this it for some reason interferes with its rotation as it is not rotating from its initial rigging pose.

No matter how much I tweaked the tangents I couldn't correct the outward rotation. I finally remembered a technique shown in class about trying to correct such errors. We were told to use something called 'Euler Filtering'. I'm not quite sure what happens when you use this filter in terms of its technical details, but it does a fairly good job of correcting it. I would imagine Maya interprets which axis you are rotating and tries to cancel out the rest of the rotation, or averages them so that they help the rotation instead of messing with it.

The results can be seen below, they were not perfect by any means, but I honestly couldn't find a solution to this. I have a feeling its because of the rig as it is fairly basic, or maybe I did something at the start which caused problems later on, but I can't think of what it may be.

Eventually the arm rotated to a decent extent of what I wanted as it no longer rotated outward, however it wasn't as satisfactory as I had hoped for as it still came too close to the torso when rotating and at times part of the hand went through the body.

Lip Sync

This is just a very short animation to try out audio files in Maya and to apply synchronisation of lips to that audio.

Linear Tangents

Whilst animating I started to notice that when my character placed his foot down and the moved forward, his foot would slide back a little, or would rotate slightly the wrong way then carry on the way it was meant to go. I decided to look into the graph editor and sure enough i noticed a slight spike in the corresponding transfer or rotation. To remedy the problem I decided to change the error prone tangents into linear tangents. To do this I selected the problem causing points in the graph, clicked 'break tangents' (as they were currently unified) and then select one of the tangent's weights so that I could then click 'linear tangent' which changed the spikes in the graph to be flat.

It can be better explained by the images below:

After the tangent was changed to a linear one, the problem was solved and the foot no longer rotated abruptly and in the wrong direction.

Reference gathering

This post is just to show the references I gathered to work from on creating my sprint start animation.

I used this image just to see the type of angles a person would place their body in at the starting blocks of sprinting, before the actual gun goes off for them to being the race.

This video helped me a great deal in referencing how an athlete prepares they body on the starting blocks, and how they position their body in the crouch position and then transition to the upright position.

I also used this image as a sort of rough guide on how to position the body within the first half-cycle of the feet from their starting position.

Decision for final module animation

For the final module animation we were presented with the choice of covering one of a number of Olympic sports to animate. I chose to go with sprinting as I thought it would be good to get practise in run cycles.

I thought it was best to animate the start of the sprint rather than just a looping sprint animation, therefore I'm going to animate the start where athletes begin with their feet on the blocks, and after that only 2 or maybe 3 loop cycles of the sprint.

Jump Animation

Here's a video of a jump animation i started. My first attempt at this had the model jumping at unrealistic timing, where the anticipation movement of the jump was fairly short compared to that of the jump motion itself. After feedback i changed it so that the anticipation lasted a little longer, as it would in real life situations when a person is preparing them selves to make a jump. I also had to correct the jump movement itself too; the character jumped but was slouched over in mid air when doing so. It now forces its spine and head back as you would expect in realistic movement.

On the landing i have tried to show the follow through of the top of the body and head moving forward sharply after its feet contact the ground. The arms haven't had too much attention paid to them, and so are quite stiff and lifeless, however i have tried to make sure they do not mirror each other in their movement (twinning), as well as the movement of the legs and feet.

Walk Cycles

The video bellow shows my first attempt at a walk cycle. The animation doesn't show many characteristics in its movement, nor does it have proper arm animation, and it also starts from a stand still rather than being a proper walk cycle. Now i know what is really meant by walk cycle i'll be creating a new animations, as i thought what was meant was a walking animation starting from zero movement. Its a starting point though.

Poses update

After taking in some feedback, I've changed my poses slightly in order for their silhouettes to be more recognisable and strong.

Poses after feedback

 Silhouettes after feedback

The main differences:

• The first model to the left now has 'negative space' between his elbow and chest; though the photograph doesn't have this, it also doesn't allow for the silhouette to be strong enough to be recognised. Its right foot has also been rotated more to ensure twinning (or mirroring) does not occur in the feet.
• The second model now has his left foot further behind him in order to be properly balanced, however the front orthographic view of this doesn't really prove it too well. The model also has his left hand moved to the side more in order to see it clearly as a silhouette, as well as having his hips and collar bones rotated in such a way that complement each other on balance.
• The third model also has its left hand moved to the side slightly for the same reasons, and its shoulders/collar bones have been moved up into a form of shrugging pose that better communicates the whole pose to the audience.
• The fourth model has its right foot rotated more for the same reason as the first. It also has its left arm rotated as to create a gap, or negative space, therefore allowing better recognition.

Animation Editors

There are two editors that can be used for animation that can be very powerful tools. These editors are called the 'Graph Edtior' and the 'Dope Sheet'.

Dope Sheet

The dope sheet can be used in a number of ways, for something as simply as moving a set key forward or back in frames. However it becomes quite powerful when you use it for 'overlapping' and 'follow through' in animation.

Overlapping is when a model has something like long clothing or long hair, and when the model moves the clothing or hair follows its movements, but at only a frame or so behind. This is extremely useful for making animations realistic, as in real life objects such as clothing or hair are bound by the laws of physics as is everything, and so forces such as inertia have to be taken into account to produce something believable. The dope sheet allows us to do this because you can select the keys on the bone that controls the hair, and then move them back a frame or two. Therefore the hair (hair-bone in the rig) will move as it would do in real life.

Follow through is different to overlapping, and an example of this is the idea of a moving object/model suddenly stopping. What I mean by that is if you take, for example, a person running at their fastest and then tell them to suddenly stop as quick as possible, the first thing to stop will be their feet and then their legs and then their upper body will probably lunge forward a little, until eventually they are stood still. So when i say follow through, i mean the domino effect of physics on a body; what happens after a force is applied (be it a stopping force, etc.)

Graph Editor

The graph editor is good for tweaking things such as the path of a ball between set key frames. In the image you can see the line descending and then suddenly ascending after the second key frame (2nd black dot). This would show the ball in the 3D environment falling, hitting a floor, and then suddenly bouncing back up. This type of behaviour can be difficult to produce without using the graph editor because in the graph editor you can choose the key frame's tangents and rotate them into an acute angle which produce a sharp point for the path to follow. Whereas in key framing without using the editor, you would have to constantly tweak the speed of the ball on the impact and bounce in order for it to be accurate and life like. It can be done, but is far more time consuming and less practical than the graph editor.

Emotions and posing

I'm going to post eight emotional poses in total soon, but I only have four for now. Underneath you can see photographs that I have taken of myself expressing certain emotions. Under that there are 3D model versions of these emotions, and finally underneath those I have shown the silhouettes of the 3D models. Silhouetting is important as an audience sees a strong silhouette and can usually determine the action or emotion that is being played out or provoked.

From left to right: Success, frustration, confusion, and fearful

The silhouettes do not show facial expression, so they have to be strong in order to be recognised

Practise Makes Perfect

Here's a video of a very quick animation i made where i am trying to replicate the bounce of a bowling ball.
Underneath this video you can also watch a video to see my reference.

My animation.

Reference video.











One thing I've noticed that is slightly wrong in my animation is that on the first bounce, whilst in mid air, it moves too quickly to the next bounce. I think it is as simple as adding one more frame to the space between the first and second bounce as its path in the graph editor looks fine, but practise makes perfect.

Reference Creation for Ball Bounce

For our first small assignment in the Intro to Animation module, we were given the task of creating a bouncing ball animation. The best way to understand the real world physics of a bouncing ball, for example how its mass, velocity, material etc. affects its bounce, is to get personally involved with reference gathering. 

With that being said, myself and a few others got into a group and bounced two different types of balls, and did so with different variables. I recorded the video, Jack dropped the ball and timed how long it took to stop bouncing, Oliver counted the bounces and Josh wrote all those statistics down for later use.

We video recorded them bouncing, as well as took recordings of the number of times they bounced, the way in which their movement started (i.e. throwing soft, hard, or dropping.) and how long it took them to stop bouncing.

Here are the results and explanations:

Tennis Ball Drop:                                             Statistics:

So here is the tennis ball drop. For the drop the ball was simply allowed to just roll off Jack's hand.

*Note that when I say tennis ball for this video and all other videos with the tennis ball, the ball is in fact just an imitation of a tennis ball; it is a bit more 'squishy' and foam like compared to a real tennis ball.

Bounces:  10

Time: 4 seconds

Tennis Ball Soft Throw:

Bounces: 11

Time: 5 seconds

Tennis Ball Medium Throw:

Bounces: 12

Time: 6 seconds

Tennis Ball Hard Throw:

*Note: Ball hits ceiling

Bounces: 15

Time: 8 seconds

Football Soft Throw:

*Note that the football is also just an imitation of an actual football

Bounces: 15

Time: 4.5 seconds

Football Medium Throw:

Bounces: 17

Time: 7 seconds

Football Hard Throw:

*Note: ball hits ceiling

Bounces: 17

Time: 10.2 Seconds

Self referencing conclusion:

Both types of balls bounce according to the initial downward force applied to them, in other words the greater the initial force the more they bounce, and vice versa. The only time this isn't shown to be true is when the ball collides with the ceiling, but of course in doing it loses some of its energy.

The football bounces a greater amount of times than the tennis ball because in ratio to its size it has less mass. So even though it's larger it is also not as heavy, this is because the material that it is made from, as well as the fact it has more air inside.

The material the football is made from allows it to be more flexible and so store more energy on collision with the floor, therefore allowing it to bounce back to a greater height and repeat the action. The tennis ball however is not as flexible, and compared to the football is a lot heavier in terms of mass to size ratio.

Its good practise to note the materials in which objects are made of and that of which they are colliding into. For example the football is smooth but the floor has carpet on it, so when it collides the floor absorbs some of the kinetic energy from the ball. Had the ball also have been 'furry' then it would lose even more energy on collision. 

The opposite would have happened if the floor was made from concrete; the ball would have bounced to an even greater high and more than likely would have increased the amount of bounces it made overall before stopping.