Animation

Animating:

Before starting with my task of animation, I had to create a gap in the timeline, with the starting point at 0 referring to the default 'A' pose, and frame 24 as the starting pose of the animation. The reason for this is that the nCloth used in my character needs to have a chance to form before the character begins to animate, allowing it to move with the character and bend to suit the starting pose. Otherwise, the cloth would not drag to its designated position and would cause issues. 

Here is the starting pose of my animation. Initially I tried to start with the character on the ground, as per my Animatic. Sadly, my rig did not account for such extreme movement of the joints, especially with my character boasting a rather complex rig setup. As a result, I have had to tweak the starting pose of my character to him already standing up. This does alter my initial story, but given that the displacement textures also did not work, this actually lines up better to a slightly altered story. 

Instead, the story could suggest that the battle was very recent, with my knight character emerging victorious. Fortunately, the story has only been slightly modified, and the overall message will be maintained. The starting pose reflects this altered start to my story, starting in a very neutral pose whilst holding his Greatsword, which is half-balanced upon the ground by the tip of its blade. This was easily achieved, simply moving the controllers to a pose I liked and pressing the S key to keyframe these positions from the default 'A' pose discussed earlier. 

My next task involved blocking in the key poses of the animation. With reference to my Animatic, I proceeded to block in the relevant key poses of my animation for the first main sequence, where the character gets up and looks around to the recently fought-upon battlefield. This is a technique I had learned during one of my other assignments, in which it was suggested that animating is made more effective and efficient by blocking out the key poses of animation, which can then be refined at a later stage. 

Since the character already is standing (and does not stand up), I have replaced this with the character grabbing the Greatsword, hoisting it over-shoulder. This helps to portray my methodology discussed in my dissertation, giving me an opportunity to demonstrate the realistic and heavy weight of the Greatsword, as well as portraying the accuracy which was the fact that Greatswords would typically be carried over-shoulder due to their large size, something I had come across during my research mentioned earlier in my documentation. 

Here is my progress after blocking in the key poses, working in pose-to-pose. The character lifts his sword, swinging it over-shoulder. He then takes a moment to look around at the battlefield, littered with fallen warriors (Note that these are currently hidden to preserve performance whilst animating). 

To create this video as per above, I used the Playblast feature within Maya, which allows the user to create a test render of the current viewport. This is why you can see all of the controllers and such. The feature is great for a quick preview of the animation, to get a feel for the motion and how something looks when played in the correct playback speed, as the playback Maya does in real-time is often not very representative of the finalised motion, playing back in a lower/higher framerate than that of the end result. As such, Playblast can be very useful, and its also really quick to produce, only a few moments. I decided this would be crucial in completing my animation task. 

So far, my character raises the Greatsword over-shoulder, and then looks around. The key poses are good but the automatic interpolation between keyframes is pretty poor and needs refinement. The most obvious is when the sword is raised, where it spirals in a totally unrealistic motion. I proceeded to refine the animation. 

In the above video, I have tried to fix the motion of the Greatsword by adding keyframes which fix the position, so that the interpolation can do a better job. Right now, the motion is really sharp and unealistic, but improved from before. However, the main thing I noticed is that the Greatsword moves far too quickly, and you would expect a heavy sword to be raised much slower. I decided to try and slow this movement down. 

In the above video, the motion is greatly slowed down. The current motion is a lot more appropriate and feels like a good speed. Now that I was happier with the Greatsword motion, I could match the motion of the hands to the sword. I decided to re-tweak the keyframed key poses of the hands. 

Here is a work in progress refinement of the hands. You can see they are following the sword much better now but they still don't fully seem to grab the sword, floating or clipping in some instances. Again, this is an interpolation isssue. I did try to adjust the bezier curves in the graph editor but this did not have the desired effect, causing the hands to move far away from the character rather than adjusting the speed. 

This next Playblast video demonstrates the hands refined further. They now seem to grip the Greatsword much better. 

This video demonstrates a slight refinement. After the knight hoists the Greatsword over-shoulder, the hand still holding the Greatsword slightly bobs down before resting, adding evidence of the Ease-Out animation principle, to make the motion seem more realistic, since your hand wouldn't just instantly become still. 

After getting some feedback, I have made some big refinement to the animation. Firstly, at the start it was suggested that the character looked like they were sighing before picking up the sword. This wasn't the case, in fact the knight was breathing and picked up the sword after the first breath. The purpose of this was to give a bit of small motion to allow the nCloth to form better before moving the character by a lot, but this was not giving the right indication. Instead I swapped these key poses so it looks like the character gradually goes to pick up his sword, and no longer looks like an exhale. 

A big change to the above video was the implementation of Exaggeration, namely when the knight picks up the Greatsword. Now the knight bends much more significantly before picking up the sword, before raising it in an exaggerated motion, bending back a bit to give the indication of the weight and significant effort the knight has to put in in order to lift it. The implementation of this principle massively improves the animation in my opinion, it makes an otherwise mundane task look more interesting and believable. This does still need a bit of refinement though, especially in the case of the arm which bends weirdly as the knight lifts the sword, which I aim to fix. 

This test video demonstrates the fixed arm movement, where it no longer bends weirdly as per before. Although improved, the motion still looks a bit unusual and I will need to investigate further to refine this and stop this seeming the case. 

Above shows more progress in the animating of my character. I have tried to implement more Ease-Out in my animation, specifically when the Greatsword hits the shoulder where it slightly bounces off. However, this doesn't look good at all and is not very realistic. Originally the animation felt off because the Greatsword would hit the shoulder and become stationary, but the Greatsword wouldn't bounce either, certainly not to the extent of the video above. This was scrapped. 

I had scrapped the change from the prior entry, and the Greatsword no longer 'bounes' like before. Instead it very slightly jitters, but this may be subject to change. 

One other little detail I added was the helmet Visor slightly opening as the knight bends down to lift the Greatsword. This corresponds to the animation principle of Secondary Animation, and adds a little extra depth to my animation. I kept this quite subtle as the helmet does not have a face inside, but the motion feels in-line with the rest of the motion so I decided to keep this in. 

At this point, I wanted to look at refining the part of the animation where my character turns to both the left and right, once the Greatsword has been lifted. I started by smoothing out the turn the character does to the left. I wanted to have the helmet move left first before the whole body in a bid to incorporate the Anticipation principle, to anticipate the turn of the character. Though, this needs more refinement. 

I tried slowing the motion down, giving the knight a chance to aborb the surroundings. I did this by simply extending the keyframes slightly, which was done by holding SHIFT and using the Left mouse button to drag horizontally, dragging it to the right of the timeline to later keyframes. 

I started refining the second turn, which can be seen in the above video. I liked this a lot, since the head turn was quite quick and realistic-looking, anticipating the body turn which followed shortly after. I continued to refine the animation beyond the timeframe shown above. 

Here I have allowed the turn to last a bit longer and ease out, as before the video cut at a sudden stop. From here, I want to refine the animation by easing it out, as this will cut to the next shot of a close up of one of the fallen warrior characters. I don't want the scene to drag on too long but I also don't want it to cut immediatelly after the knight looks to the right, I want the viewer to think of the knight catching something to the right, creating a little bit of anticipation in the viewer before showing it. I would want the motion to ease out during this time until the shot is eventually cut. 

The above video refines the keyframes of the last one, and the motion feels like it continues from each key pose quite smoothly. The way I achieved this was by easing the keyframes between one-another. Within Maya you have access to different types of interpolation of the Bezier curves, being flat, linear or smooth. By selecting the keyframes and selecting the smoothed option, the motion moves between the key poses much smoother. From the above video, the main thing to refine is the left arm which doesn't really fit the rest of the motion. 

Here is a current preview of the whole timeline of the animation I am currently working on. Since my last update, I have slightly smoothed out the final part where the knight looks right, the arms move a bit more smoothly. 

I decided I wanted to smooth out the rest of the keyframes, since this had worked well previously. In the above video is the current progress of my animation, with no smoothing applied to the keyframe graph. If you look closely you can spot when the animation meets a key pose before moving to the next one, the animation ever so slightly stops before reaching the keyframe. Essentially instead of a smooth transition between multiple keyframes, the keyframe is eased out individually between each set of keyframes. 

For example, you have key poses A, B and C. The ideal outcome is for A to transition to B to C. However, what currently occurs is that A starts slowly, accelerates then decelerates when reaching B. This then repeats when moving on from B to C. 

For comparative purposes, here is my current timeline in the Graph Editor, before any easing of the curves has begun. 

In order to ease these curves, you need to select the keyframes and then select 'Spline Tangents' which will ease all of the Bezier curves selected.

Here is the resulting timeline after the keyframes have been eased. Now the keyframes will transition smoothly between the preceeding and following keyframe. You can see a clear example of this taking effect near the start of the timeline, looking at the blue curve towards the bottom. The curve has now been smoothed, whereas before it would ease to a straight line and then back to an eased curve. 

In the above video this has been done, and this completely solves the issue I was having, since the animation now looks very smooth by comparison and flows nicely. A very easy and simple tweak which had a big impact. 

In the following video, I have worked on refining the motion of the Greatsword, this time focusing on fixing it to the ground. The difficulty is that the controller of the Greatsword is at the Pommel, the base of the sword, so animating it to appear fixed to the floor (whilst it is leant on the ground at the start) can be difficult as the 'pivot' for moving it is on the wrong side essentially. During the rigging stage I did not know of a way to implement multiple controls to control the same object, as otherwise I would have implemented this then to prevent such a thing from ocurring. Though, I am sure there would be a way to achieve this, and this demonstrates the lack of my knowledge in this area. 

As a result, I have had to manually tweak the position and rotation of the sword. This took some time to do, having to use Orthographic views to achieve the result as I needed to keep moving from different views and match the position across the times the sword gets moved and do so in each view, as I would rotate it to fit one view and it would be dis-aligned from another. Though after trial and error as well as repitition, I was able to achieve a pretty good result where the Greatsword appears to stick to the floor pretty well. There are a couple discrepencies but these are so minor you wouldn't see them without knowing about them beforehand and looking for them. The next step was to re-adjust the keyframes of the hands to match the movement of the Greatsword, since moving this meant the hands were no longer in the correct position as these need to match the hands. 

Here I have been able to adjust the hands so they match the movement of the sword pretty well. 

At this point I felt pretty happy about this scene, and moved on to the next one which shows a close up of the hand of a fallen warrior. 

Here is my current progress. I have enabled visibility of the scenery elements now as I need them to position the shot of the close up of the warrior's hand. Instead of creating a new project file for the next shot, I decided to move the camera the keyframe directly after the last keyframe of the first shot, so the camera would 'teleport' to the next location in a sense. My reasoning for this was that I was scared the nCloth would not match up between shots if this were the case, since it is simulated. Plus, I would most likely need 24 frames at the start of each new shot to allow the cloth to take shape, which would add to rendering times. Although this way of working with project files in animating is flawed (as if something goes wrong, you have lost all your progress), I felt this was reasonable justification to proceed with this workflow method. 

I have implemented the following shot of the warrior's hand, which is close in view with my knight character in the distance. The idea would be that a depth of field effect would be applied, transitioning between focus on the hand to the focus being on my knight character. To make sure the knight is not static in the background, I created a short idle cycle of the character subtly breathing in/out to prevent this being an issue, since the movement didn't need to be anything major. 

Following this shot is the final shot, the reaction of the knight. Initially my Animatic had proposed this would cut to a long shot across the blade of the Greatsword, where blood could be dripping from the sword to indicate he is the cause, and suggest to the viewer how he was feeling about that fact. However, I felt the inclusion of blood like this was a bit out of place, since real Mediaval fights may not have seen as much bloodshed as is typically depicted, and may not be enough to make a shot of this worthwhile. I was also a bit skeptical to add particles to the scene, in the midst of rendering times as well as the fact that particles had not worked before when I had tried adding in rain for the environment. 

Therefore, I modified this scene to a close shot of the knight, where the knight character lets out a deep sigh of possible regret. I felt that it was pretty obvious that the knight was the cause of all the death of the warriors in the scene as the knight is the last person standing. Hopefully this decision was the right one, but I felt there was a good sense of story and characterisation from this small segment. This would then fade to black. 

So far the animation is good, and clearly conveys the overall motion/action. The animation is exaggerated to a suitable extend and there is clear evidence of anticipation when the knight inhales before exhaling. However, I felt the motion was a bit fast after using the Playblast to test it out, and felt it could be eased out for longer. I decided to refine this with these points in mind. 

Above is the refined motion of the final shot where the knight lets out a deep sigh. I have increased the duration of both inhaling and exhaling since I felt that a deep sigh might take a bit longer, and felt this way upon experimenting with the motion myself to give myself a bit of a better idea. In addition, I allowed the motion to ease out for a bit longer to seem more smooth and give the animation a chance to fade to black. One other thing I tweaked was the motion of the arms, specifically the left arm of the knight. Originally this bent significantly, too much in my opinion, so I reduced this so something more appropriate. From here the animation looked pretty good for this final shot. 

Something I had not yet considered was some form of establishing shot to showcase the scene and overall environment better. I felt this was something pretty important to include especially considering the nature of my project with such a heavy focus on the visual element. To do this, I included a simple camera pan in the distance from right to left, giving the viewer a chance to absorb the environment. It also clearly depicts the whole environment and all the characters deceased scattered around with only the knight still standing. This establishing shot is very simple but really helps to set up the narrative of my animation, especially considering the fact that a typical viewer will have no context to the animation. This is something I find myself forgetting, and find it easy to forget that something like this is really important since the viewer has no idea what my project will show or be about. I am glad that I had finally included this in my animation. 

Based on some peer feedback, the shot which showcases a close up of one of the fallen warriors was a notable point of concern. They had suggested that the close up of just the hand felt out of place, and came off a bit confusing to them. The problem seemed to be that the fact only the hand was in shot felt like the shot was sending a different to message than that of what I was trying to convey. That shot might - to a typical viewer - imply that the character is important, gradually showing them in the shot only to cut away from it and confuse the viewer. Based on this feedback, I decided to modify the shot to something which shows the whole warrior in frame, clearly deceased. Hopefully this conveys the message of my story better than before. 

nCloth - Problems encountered:

Test #1: Chainmail (Original)

At this point I decided to test the nCloth of my character to see how it might look in the finished animation once rendered. This was using the Chainmail preset mentioned before in the nCloth section of my blog. 

After viewing this preview of the nCloth, I came away disappointed - specifically at the nCloth covering the tassets and pelvis section of the character. The cloth simulation felt very flexible and moved a lot more than I had expected it to. I had imagined it to stay a lot more formed and solid. Though, the simulation was correct, and was moving as it should. It was my flawed model which was the issue. 

I had created the plane for the cloth based on the reference images I had been looking at, and making educated guesses as to how the chainmail was actually held and pieced together in conjunction witth the rest of the armour. My educated guess was that the chainmail connected to the bottom of the chestplate, and draped down. However, in order for the chainmail to protect, it would need to maintain a fairly stretched pose to cover the front of the legs. In the test video above, the chainmail just collects at the bottom in a very vulnerable state, not protecting the upper legs much, if not at all. 

Based on this, my assumption is that the chainmail of the armour references I had looked at were either attached to another part of the armour, or extended around the body a lot more to help it maintain its shape when the armour moves. Though, this is something I had not spotted before and my lack of research meant the nCloth deformed in a way I had not expected or wanted it to. Unfortunately it was too late to go back and fix this, as it could mean re-texturing and re-rigging if I was unlucky. I was left with two options, I could either remove the chainmail altogether and be left with a suit of armour which lacks the level of functionality I had originally aimed for, or I could use different cloth presets to get more of the look I wanted. I decided to try the latter option and see if I could get a result that better suited my needs. 

Test #2: Burlap

I started with the Burlap preset for the nCloth physics. The result was better than the chainmail for what I was looking for, but it creased a lot and still did not stay as rigid as I had hoped. I continued with different presets. 

Test #3: Chiffron

I tried the Chiffron preset, but the result was very glitchy and the cloth seemed to get stuck in itself, causing it to break apart. This was not suitable, and I continued testing. 

Test #4: Concrete

I was intrigued by the Concrete preset since I was curious how this would actually work with an nCloth object. Since I wanted something more solid, I decided to try it. It was good at maintaining its shape, but perhaps too good. The deformation in the cloth seemed to stay permanant, and was not suitable. 

Test #5: Heavy Denim

Heavy Denim was the next preset I tried. Unfortunately the result was similar to that of Burlap, with lots of creasing in the cloth as well as not staying very rigid. I continued with different presets. 

Test #6: Soft Metal

My next test involved the Soft Metal preset. It was pretty good at maintaining its shape and keeping fairly rigid which is what I wanted. However, whilst soft metal looked good initially, I nocited that it started clipping throught the armour. I don't know why this happened but clearly this was not a suitable choice. 

Test #7: Thick Leather

Finally, I tried the Thick Leather preset. This option yielded similar results to the Burlap and Heavy Denim, but seemed to be more rigid, and of those three particular presets this one was the most suitable as it covered the area it was designed to protect the most. Since I was running out of time in my project, I decided that this would be suitable for my character based on this test. 

Animation - Final refinement:

At this stage, I wanted to start rendering. After doing a couple test renders, I noticed there were a lot of clipping issues going on in the model of my character. This can be seen in the test renders prior, but this was something I had originally missed since the Playblast gets done in a lower quality to get done quickly, and things like that are harder to spot. Though, this is something which was very obvious once I had done a test render. 

In order to fix this efficiently, I enabled the Shaded Wireframe view, which draws an outline around all the faces in the topology. This made it much easier to spot any clipping ocurring, such as in the large tasset plates which can be seen in the image above. This was easy to fix, simply moving the model via the controllers each time I spotted it ocurring, and it mainly ocurred during the time the character went to grab the Greatsword. There were a couple clipping issues I couldn't fix though, such as a couple times with the right shoulder which would slightly intersect at the helmet. This was something I was able to minimise as much as possible, but not fully prevent. Hopefully this does not cause too much of an impact in the final piece. 

Futhermore, I took a bit of time to refine some of the motion in my animation, since my test render helped me to identify some flaws. One very obvious flaw regarded the neck plates during the start of the animation, just before my character goes to pick up the Greatsword. It seems to oddly bend to the right before normalising. I was able to fix this easily, and I am glad this didn't make it past this final check and into the full render. 

In addition, I still felt that the motion when the character picks up the sword looked a bit off, specifically as the knight raises it over-shoulder. Specifically, the right arm seems to move and then suddenly jump into a different motion, and it doesn't feel very smoothed or refined. This was something I had picked up on prior, when the arm had been deforming oddly in the earlier stages of the animation testing. The reason for this as far as I can tell is that the IK of the arm would bend too much, causing the Pole Vector to bug out and move in an unusual manner. My solution was to add lots of keyframes here to try and fix this, but the motion as a result looks a bit jumpy and not very smooth. 

My solution to this problem was simple, I moved the position of the arm and sword during the time this occurs and moved them a bit further out. By doing this, the arm did not need to bend so sharply, and the problem did not happen. This did mean the Greatsword did not line up as well as it did before when it is leaning the tip of the blade on the ground, but I felt this trade off was worthwhile, as the new motion looks much improved as a result. 

Finally, just before creating a test render I went ahead and eased out all of the keyframes like I had done before. This is because creating new keyframes causes the Bezier curves to become unsmoothed, and so doing this a final time means that all of the animation should be nice and smoothed where it may have not been before. 

Here is the final resulting animation. I believe the motion is great, and I am unable to identify any major issues or flaws in it. I feel that I made made good use of the Animation Principles to help me in animating, using principles such as Anticipation and Exaggeration appropriately and effectively. Despite being altered from my initial pitch and animatic, I believe the animation still portrays the overall story effectively, and does not hinder the final result. My methodology discussed in my demonstration is evident in my animation, where the character displays a clear weight to the Greatsword. Visual storytellling has been utilised in the animation too, creating additional layers to the story where appropriate. For example, the deep sigh at the end of the animation suggests regret and guilt, giving other denotations about the story. Whilst the camera angles used do not necessarily help to showcase the character in full from all angles, this will be aimed to be achieved in my 360 video, allowing me to focus more speicifc and effective camera angles to instead convey the story instead. I believe this was an effective decision. 

Overall, I believe the animation is solid, and is ready to be rendered.