Instead of my usual pitfall of over thinking my idea several times just for the purpose of over thinking it I want to explore this one and fix issues that might come up with it on the way, to save time and frustration.
The basic idea was to involve speech and creature animation into the clip, so the easy conclusion is a person talking to an animal. I liked the idea of animating a big predator animal in a realistic fashion, so I decided that zoo would be a good set up to not get too cartoony and do a big, dangerous animal.
My plan so far involves a woman looking and interacting excitedly with a tiger in zoo, and while I have the clip roughly lined out I’m still unsure about the visuals, as there are several rigs I could use. However I have to keep the voice acting in mind which so far is a headache.
After finalising the animation I moved on to post production.
I wanted to convey the rich feel of jungle through a proper render and sound scapes. I started off by setting the lights for the scene. I asked for advice from a professional lighting technician who explained the set-up he would use if this was a real life production, and took his suggestions as my starting point. I adapted his three point lighting draft as my starting point, then added more secondary lights to get better control, for example two rim lights at slightly different angles for more contrast. Since maya can light objects selectively I could prevent the scene from overlighting:
First I connected my shadows to the main light source, as is logical, but I found that what was a nice angle for the key light formed some unpleasant shadows on the ground, disrupting the scene. So instead I disconnected my shadows from the active lights and created a negative light very close but slightly angled to the Key Light and created the shadows for the main character this way. Unfortunately something is wrong with my shadow render now so I can only show my very first render which only shows unadjusted raw shadows.
Another challenge was my idea of having the shadows of the trees fall onto the scene. I learned about Gobos, and how maya can project a flat 2D texture as a light source. The problem was (as you can see in this first progress shot) that the gobo still actually counted as a LIGHT source, adding light to the picture and that way disrupting my original lighting set up. I tried to find a way to have the light only cast shadows, and no light, but it wasn’t as easy as I expected. I even tried putting a slightly coloured texture in, hoping I could filter out the scene foreign colour and use the mask to cast the shadow again in postproduction:
It didn’t work.
As it proved impossible to tell the gobo directly to cast only shadows I experimented with some other alternatives, like creating the leaves as mesh above the scene and casting actual shadows. However I was painfully aware of how work intense this would eb and how bad the result would be in comparison, regarding complexity and realism of the shadows. I would have much less control than I originally had. Another problem with this idea was the need for all my objects in the scene to be receiving the shadows of the light above the mesh, without casting any unwanted shadows of their own. This should be really simple, but for some reason maya didn’t want to cooperate, so it seemed as if I had to duplicate every object in my scene, set their copies to invisible but cqasting shadows, whereas the visible objects were receiving shadows without casting any. The need for light linking (to tell maya exactly what light was casting which shadow and which light on which object) made for an incredibly complicated scene set-up with no guaranteed success, so I dismissed the idea (sorry Tim!).
Instead I went back to my Gobo light and learned that I could render a lighting pass in black and white, which would allow me to set the layer to multiply in post production, eliminating the unwanted light source of the whites and giving me full control over shadow intensity:
I wanted the gobo light to get darker towards the edges so I could then put a 2D background into the shaded area. My first rough background test was this:
It turned out to be slightly difficult to adjust to the surrounding lighting, so at this point I considered jsut adding more props and working with extreme light and shadows to create a background. I left the final decision of that for later, and wanted to compose my scene to know what lights I’d have to consider for the background.
With the gobo light in place I was hopeful I had all my renders sorted out. Unfortunately this was the point my shadow passes stopped working, and slowly dragging every element of my scene down with them: While I could still render out single images it proved impossible to render out sequences of the passes. So I’m forced to work with playblasts until I find a solution, but to give an impression I rendered out a few still images and composed them to what the final could look like (without shadows):
Since my area of practise is character animation I knew I wanted to create a short scene demonstrating nothing but acting and body mechanics. It was quite challenging to think of an appropriate action when being so limited in backgrounds and props (not wanting to spend any time on modelling or rigging), but when I read through a list of possible character animation training tasks I settled fairly quickly on a character opening a box. I dismissed my previous idea of sneezing because of how little emotional involvement it shows: Without a conflict in the scene it is very unengaging and easy, but when adding a lot of context into it it defeats the purpose of just looking at the body mechanics of a sneeze.
Opening a box offers a lot of room for different changing emotions, since the character has to constantly react and adapt to his or her environment: Seeing the box, deciding to open it, finding a way to open it, seeing the content, reacting to what he or she sees.
I struggled a bit coming up with a scene and some form of punchline or pay-off. I looked into script writing and found this great advice about scenes (for book writing): “A scene consists of a goal, a conflict, and disaster.” Winning is a very boring endeavour to watch and keeps an audience from relating to a character and getting emotionally invested into a scene or story. However, I find it extremely difficult to set up the disaster and let my characters fail in their actions. My severe dislike for slapstick doesn’t help much either.
I still didn’t manage to set up a truly witty and memorable scene, but I didn’t want to repeat the mistake made in our Digbeth short film and spend more time than necessary on developing a story, especially since it would barely matter for this exercise.Instead I decided to run with the first idea coming to my mind and chose the setting suggested by the rig:
Animation Mentor released one of their new tribal rigs for free, which promised to be a very high budget rig and with the lack of facial expression ideal for an action shot like mine. I kept to the tribal theme suggested by AM and made use of a Treasure Chest my flatmate had modeled earlier this month.
PLANNING AND REFERENCE
With the character and set in place it was easy to settle on the first idea and to draw out an animatic for the basic overall action:
My main acting beats came mostly from experimentation with drawing thumbnails. I have tried several times in previous projects to create the acting beats through acting reference, but I never found myself getting truly comfortable with the camera and therefore never seemed to deliver a good performance in terms of creativity and wit. Instead I draw out my first idea and then force myself to draw at least ten different versions for each pose. This way I have often found improved and more engaged poses that I then took on to my animatic and used as basis for video referencing.
The video reference for this project was relatively difficult to shoot due to a lack of props and the movements being very quick and the character more lightweight than I am. However I found it surprisingly helpful to figure out his centre of gravity:
When I first posed the character out I got a lot of interference in the legs. The knees would constantly be too high up and no matter how much I altered the knee controls or changed the footroll the knees would stay all over the place. Through video reference and studying the movements on myself I realised that my knees weren’t actually affected by the way I held my foot, but by my hips and where the centre of gravity was. From then on in it was easy sailing and the rig felt incredibly natural and responded great to subtle changes of the hips (hovering over the left foot first, then changing over to the right, leaving the left leg free to be moved around.)
I also took a lot of reference from different animals: The paw held up in excitement when he first sees the chest is the same movement that my friend’s dog does whenever something strange catches her attention. (though it changed slightly in the course of animation to turn into a more relaxed hold, which felt more natural)
When he jumps down to make use of the stick, it seemed very fast in the original first draft. His quick jump down, picking up the stick in the same movement and using it straight away seemed very choreographed and suspiciously convenient. However I didn’t want to put a whole different series of movement in, explaining how he finds the stick and tries out what kind of tool it could be, out of worry about finishing on time and having enough buffer time to really polish the shot. To avoid that unnatural moment to distract from the action of the shot I had to find a different solution: I found that by placing a simple break in the movement and giving him another 2 seconds to contemplate his next move, I could already break the staged feeling. I was approaching the shot in a mixture of pose to pose and straight ahead at this point, and when filling in the poses of his moment of contemplation it felt fitting to have some tension rise before he decides to jump down and in one quick maneuver makes use of the stick. I drew my inspiration of that from wildcats, ducking down low in anticipation of their attack when hunting:
At first it seemed quite difficult to get the feeling of concentration right due to his lack of facial expression or even the possibility to really manipulate the eyes. However, the more I analyzed the footage the more I realised that cats don’t actually change their facial expression at all during that movement. The unchanged stare expresses the concentration, and the contrast comes from the suddenly lowered head and the tense shoulders. Both could be transferred to my rig easily.
His walk was mostly inspired by Disney’s Tarzan and real life Chimpanzees & Gorillas. It was actually a bit of a challenge for my own shot video reference because it’s harder than I thought to keep the hips at a half-height like that.
Like mentioned previously, I found the rig very useful, and many times when I got frustrated about it not behaving the way I wanted, it was actually my unawareness of certain bodyparts and how they should be placed. I found this problem is easiest resolved through acting the shot out myself.
I transferred my poses from the animatic into the first blocking stage, incorporating general rules of posing for every keypose. I tried to pay as much attention as possible to a clear silhouette (for this reason I placed the camera very early on instead of finding the best camera angle after blocking)…
…, line of action and great contrast. For example I made sure to implement a lot of weight changes and moving his hip up and down to avoid stiffness. I remember my previous animation often being far too subtle and lifeless, so I went with the general advice of pushing everything 3 times more than what feels right and then only turning it down if necessary. It made the blocking process far more fun while more focused at the same time and saved me a lot of improving later on.
The fixed camera angle enabled me to treat the whole animation more in a 2D manner (a workflow which comes much more natural to me) and ensure that timing and spacing are correct and everything is following appealing archs. There were many stages of little hiccups caused by counter animating (since his spine is almost always in an extreme C curve and his hips are very flexible there was a constnt high risk of the hip and the spine controllers conflicting each other) which I found easier to spot and eliminate than in my previous work, because I knew which camera angle mattered and where exactly the arch had to be.
BREAKDOWNS AND SPLINE
After the first blocking I found it incredibly difficult to put down my breakdown poses. I had put a lot of planning and effort into my blocking and was quite pleased with it at this stage, but there was a lot of empty time to be filled with in-betweens and I wasn’t sure how to generate fitting and appealing breakdown poses. The solution was a little plug in I found, written by Brian Horgan. It is an improvement of the ghosting tool in maya and lets you create a solid outline for your character, serving as a reference point. This was another step towards a 2D animation workflow which really helped me to focus purely on animation principles and appeal.
I planned my breakdowns quite meticulously and was happy with the look and feel, but I made the same mistake I made before and didn’t stick to the rule of keyframing every 4 frames. Especially the moving holds I kept as simply stills, which works fine for stepped mode but turns everything very mushy when turning into spline.
Even though I had put down a lot of breakdowns and paid attention to snappy timing, it still felt incredibly floating when I started to work in spline. I realised fairly quickly that I hadn’t spend enough attention on the elbow controllers, only keying them every other pose since the IK wrists pose themselves out nicely. This only works in stepped though, and once in spline unattended elbow controllers make the entire movement feel floating.
I had never animated interaction with a prop before and it turned out to be more challenging than I expected. From asking several animators I heard the preference to work with FK arms, but I didn’t ask or consider that that was assuming there where no props involved. I thought the difference between the two was simply of preference, and for some reason it had just never occurred to me that there is a simple and strong reason, even necessity, to use spline: It looks hands into certain positions.
Some rigs I used were incredibly helpful and were able to switch between IK and FK, keeping the approximate pose of the arm. This rig however isn’t, so when I tried to introduce the stick or chest into the action I realised I had to remodel the now IK arms for the entire shot. The ghosting tool saved my life in the process, since I could ghost the FK position, switch to IK and bring the arms back to where they belonged.
But for the stick IK definitely wasn’t enough and it took me several tries and retries to figure out the complicated world of controllers and parenting constraints. To add to the confusion the rig actually has to regrasp the stick several times due to the wide range of movement performed. A controller only lets you lock one certain position though, so for every time the hand lets go and repositions slightly a new controller is needed – even if it’s just a slight shuffle of the wrist to make the pose more natural!
My former tutor Tim suggested duplicating the stick, even though I found it easier and more logical to just have a new controller added onto the same prop. I got away with three controllers on each side and a few frame by frame hand movements to bridge the difference.
What was much harder to decide was the order of constraint. It first seemed logical to parent the stick onto the hands, but it already caused some trouble since the stick is held with both hands, using two locators, so the average had to be calculated, which left me no possibility of efficiently manipulating the stick. Instead I decided to have one hand lead the stick and the second hand follow the stick. That way I could control the stick with one controller and be sure the other hand followed just fine.
The problem with this approach is that for a long object such as a stick it is nearly impossible to create appealing archs, since the tiniest shift of the wrist gets amplified in the stick tip, leaving again little control. Instead I changed the constraints once more and parented both hands onto the stick and animated that. Figuring out nice timing and spacing for the stick took me quite a while since I had to keep in mind the forces, the perspective and the hands holding the stick. Once I was happy with the stick there would always be some trouble with the hand’s archs, creating a little break in the animation and making every movement look awkward. In the end I solved this through a mixture of revising the stick and breaking the hand constraints for a few frames at a time, manually putting the hands where they should be to follow along a clear arch.
Unlike my previous animations I had no problems with knee popping. Probably because there wasn’t a normal walk involved at any point. The entire shot got me much more confident with the way hips move and how much movement is allowed or even necessary – and how much a good placed centre of gravity adds to the feel. I also understood for the first time how to look at only the torso, regardless of head, arms and legs, which I was never able to do before. This allows me to fix problems like knee popping or many other walk cycle issues early on and more confidently. Yay!
When I blocked in the last little crawl I did it without actual video reference, only performing it myself a few times to get a feeling for the movement and looking at this beautiful example in Pocahontas:
In my own version I was worried about the relatively stiff spine at first. He seemed to fall into the positions without moving his spine at all, but the timing and the little space available didn’t allow for any of the reverse techniques I had been using previously. After studying my own movements and the example it became clear though that there is actually very little movement in the spine during a crawl like that. Instead my problem could be solved by adjusting the timing – making him move less frantic – and smoothing the movement of his torso out using the hip controls.
LACK OF FACIAL ANIMATION
While I was quite happy to find a rig without facial animation (because it eliminated needless distraction and saved time) there were a few moments that made me wish I had eyebrows or something to convey a stronger emotional response. However, being forced to work around this has actually improved my understanding of body language, and as recent research shows the face only conveys emotion clearly in connection with body language. Without those cues the facial expression of pain and joy are extremely close to each other.
The moment before his jump off the chest I wanted to convey the concentrated tension of calculating his jump and preparing for the action he’s about to perform:
The timing had to be adjusted until the very last moment. I kept making the moves slower and adding more holds, which without any additional changes made everything feel more natural and settled. Especially the added anticipation before the jump onto the box conveys a lot of character, while the slowed down moves keep it from feeling too hectic. Here are some of the passes I did..
Next up is post production!
Over the past few days I have analysed several Animation Showreels to find out exactlz what distinguishes a good showreel from a bad one, and while solid animation is obviously a big part in it there is a lot of work around it: The appeal of characters and the quality of the render can go a long way. While a simple playblast is great for showing that your work is solid and doesn’t need any fancy effects to convince, it is just much more pleasing to watch a character in an equally simple setting, but nicely rendered and shadowed (for example in this showreel: http://www.youtube.com/watch?v=cwnafMmrdwo )
Staging and pleasing frame composition as well as chosen colours and timing of the chosen shots are essential for a fluent show reel. A lot of shots can go bad simply because of bad staging and composition. (compare for example this http://youtu.be/X-_ezYLCP3I to this http://www.youtube.com/watch?v=b8CxjArUKQ4. Both show basic scenes, often jsut involving one character and props, but the second example is much clearer in both content and visuals.)
All this is certainly something to wathc out for when planning your shots and putting together a showreel, but I realized that my previous idea of studying staging possibilities first to then transition into poses might not work. Instead I tried to look for a good example of Acting, and decided to study a long shot by the 2010 film “The Artist”, which imitates silent films of the 1920 era.
The scene focusses on two actors in Hollywood and their evolving careers and relationships. The extract I chose shows the female lead character secretly entering her idol’s dressing room and it’s an interesting example to study, since it shows her going through a whole range of emotion, delivered without word or sound, only accompanied by music:
There are a lot of interesting acting choices in this one, since the actress is required to portrait her entire character and emotional range in the seemingly subtle gestures. The film adapts the 1920 common acting style of expressionistic acting, which makes it an interesting piece for my question of how subtle acting can be in animation. Due to the medium of film being relatively new, a lot of stage actors found their way into the film business, bringing their techniques of stage acting with them. Unfamiliar with the intensity and intimacy a close up creates for the cinema audience, their acting was still about big gestures and emphaiszing movements to make it easier for the audience to udnerstand and follow. This style then transformed into the more low-key and naturalistic acting we see in modern cinema – although these naturalistic acting techniques already developed in the 1920s, for a while rivalling with the expressionistic style.
Studying a performance as carefully thought out and designed as the example extract from The Artist is a great source of information about body language and how it is perceived. The rubbing of her hands, the facial expression changes as she enters the room or the tilting of the head, allowing us to follow her emotions even as she turns her back to us later in the shot are great ways to find alternative ways of delivering emotion and learning the foundations of body language. Small moments, like her entering the room and holding the tension even with her smile dropping, only ending the moment and acting impulse with the rolling back of her head made me understand body language on a new level. (Hopefully I can upload some screen grabs to follow)
This case study made it clearer that my area of interest is comparing different acting techniques and hopefully improving my abilities to create and structure a believable performance (definitely a current weakness of mine).
For now I feel a bit in a rut with my work, I think I might have to define clearer learning outcomes and work towards more specific goals. My next step therefore will be to learn about different acting techniques and hopefully doing several small tests to see how far I can implement them into my animation, as well as understand exactly what differences there are between the different approaches.
My aim in Specialist Study One is to improve different areas of my character animation in preparation for my final year project / Specialist Study Two. I decided to animate several smaller pieces (as much as time will allow, but if necessary I will only complete this one first brief but ensure it is polished.) starting off with a sneeze.
My main focus will be to improve my poses, losen them up and make the action clear, easy to read and consistent. I hope that with a stronger foundation and ground work I will have an easier time animating with better results and have more freedom in the polishing stage to really enhance the performance instead of constantly fixing up mistakes.
To losen up my poses I want to explore many different acting possibilities and work with exaggeration, and also make sure to keep strong line of action, arcs and clear silhouettes in mind. I researched several illustrations and short animations about sneezing to get a feel for the process and good poses:
I haven’t produced my own reference yet, but here are some shots of youtube:
For this project I teamed up with Huni, a 3D animator from the L6. His final year project was a short comedic piece about a drunk person getting in trouble when encountering a policeman and being asked to walk a straight line.
The most important aspect of the animation I had to complete was getting the scene and characters across. Since Huni had picked two slightly different versions of the same rig – allowing for visual unity – I had to clearly distinguish the characters and set the conflict (drunk person meets authority).
The policeman was especially easy to work with, allowing for stereotypical “authority” body language. With minimal movement and clear, sharp poses I was able to animate him in a very short amount of time, following mostly Huni’s prepared poses. With only little rotation of the upper body and slight weight shifts I could make the rig feel alive even during times of no real movement.
In great contrast to that, the drunk character had to constantly be moving and noticeably shifting his balance, unable to keep himself still and upright. Additionally, Huni focussed on a cartoony, exaggerated style of animation, so there was a lot of need for broad movements and stereotypical “drunk” poses.
Having never animated a drunk person it took me a while to fully understand the concept behind movements, but through observation of real life footage and several animation tests I understood several principles in drunk movement.
Real life reference was actually less useful here since most of it seemed faked and gave away very little about body mechanics under alcohol influence, so I got most of my clues from two examples of good animation:
And one bad example:
There is obviously a great amount of follow through in the floppy arms. The chest usually leads the movement in an almost unconscious way, with the rest of the body trying to counter balance. The legs are in a constant attempt to keep the balance, leading to small and frequent steps, with the feet often landing at awkward and twisted angles due to the lack of concentration in stepping. The Animator’s Survival Kit compares walking to “controlled falling” – drunk walking obviously loses a lot of the control and turns more into a constant fall with last minute catch.
Steps don’t necessarily have to be forward or backwards but are sideways and as tendency very non-directional and simply to balance out the upper body. The weight is always on the leading leg: Walking means balancing on one leg for a short time while the other leg moves forward. Drunk people are incapable of the balancing act. their upper body moves and their legs follow, the leading leg is always placed in response to a previous action, to catch the weight, therefore a drunk person won’t lift or move a leg without weight on it.
The bad example features a lot of off balance poses and an overall stiff spine. Especially the steps are off, lacking weight shifts and good timing. The arms trying to reach for support and misjudging the distance however is a nice touch.
Originally I planned for a lot of facial animation for the comedic effect – with the rigs in use this was obviously impossible. But having to focus on the body movements this much was a good exercise.
I first planned some poses and animated a linetest in 2D. I’m more fluent in this medium and it allows for quicker and more expressive sketches. I also find it more intuitive to adjust the timing this way.
I then posed out the models in maya, following the linetests quite closely.
Especially for the drunk person I could and should have adjusted the timing more, getting more changes between slow and fast. The built up to some of the falls is too slow – the character is almost sinking into his fall, making it seem less like an actual fall and loss of balance.
The first weight change for the police officer feels very forced and awkward. Looking back the amount of shift, as well as the way the foot plants back on the ground should have been adjusted: With a little less movement in the hips the move would have been subtler (even though too subtle would have defied the purpose. The idea was to losen the rig up for a moment to make the action more interesting). If the foot had planted differently to how it lifted the movement would have felt more natural and possibly more subtle.
The entire process of actually creating the infographic took me about 2 days. I’m quite grateful for the Illustrator practise I got through creating all the little assets and word blocks. Having After Effects and Illustrator open simultaneously and quickly updating files from one to the other is a real blessing for typography and it took me several attempts to get the size and resolution right for all the close ups. Speaking of which, I think the typography part of reasons for stress is the weakest part of this animation. It was fun to try typography and I think I’ve gotten a better feel for it as I was doing it, however the sequence lacks some overall connection. While I licked the ideas I had for each reason individually there should have been some way of showing them all together. I should have probably tried to composit them all into one pleasing text block that could have been revealed in a zoom out (only briefly so it doesn’t feel like repitition). Next time.
The “#1 reason” part was actually created second and I managed much better with that part.
All the transitions were done mostly straight forward without much planning. I came up with most of them as I went along – one transition always seemed to trigger the next. I’m surprisingly pleased with how it came out. The trickiest part was possibly the little shadow monster in the “Final Year” bit – originally I had planned to insert an animal sound (grrrwl) to exaggerate the image, but I decided that it might make the rest of the film sound very quiet unless mixed well. If I had given every transition its own sound effect the animation might have become a bit hectic and almost slapstick like, so I left it.
The music was found during a research session for free background music. At first I was worried if it might be too upbeat considering there is some talk about suicide in the animation, but I decided that it goes well with the very fast and energetic feel of it all. Also, I didn’t take the topic too serious, so dramatic music would have been inappropriate.
The shadow monster was also pure horror to composit, because for some reason I couldn’t make the pre_comp transparent, even though I had a white background, set to multiply. I ended up exporting the shadow move on its own on white background, rendering the entire video without background in Alpha RGB, then putting it on top of the background, followed by just the shadow move, layers set to multiply. Pretty simple in itself, but since I had to get there by trial and error it took a lot of time.
The final film still has some minor timing issues (for example the odd way thw two statstic groups crowd in the corner in the beginning) – I blocked in first timing very roughly and often got mixed up with the scale of things, so it took several attempts of pre-rendering and changing individual bits. Sometimes by fixing one part, another one got messed up, so these last ones are kept in because fixing them would be too risky for too little result (now the animation has character. ehem.)
In the last info-bit about the positive effects of meditation the countign numbers were supposed to show 7 weekdays and go up as the word “daily” is spelt out, then combine into an 8 for the 8 weeks. I liked the potential but I feel like I messed it up a little. I am not very comfortable with generated text yet so spelling out “daily” was a real challenge for me, and I didn’t exactly know how to best turn the 1-7 into 8. My original thought was rather hand-drawn 2-D inspired ideas of the 8 turning into a thing, light line, that zig-zags over the screen briefly before darting into the 8 position and unfolding again, but I am BY FAR not good enough in controlling masks to do anything even close to that. I experimented with some scaling of the 8, but finally I decided that the simplest solution was the best for what I could achieve. I still wish the whole counting-down idea would have become clearer. I wanted to have the numbers not only appear but actually count up almost like a digital clock, but again, I didn’t really know how to do it while keeping a 3D layer with other effects on it, so again I settled for whatever would get close to the idea.
When I created the pie charts and made a slightly transparent background I liked the textured feel they had, so I experimented a lot with an extra texture layer set to lighten or multiply,, only affecting the foreground (through masking), but the benefits of occasional nice texture in moments where a lot of solids were on the screen didn’t outweigh the disadvantage that the little groups of people became faded and part of the background. The whole thing felt flat and less engaging, so after a few experiments I decided against it left it as it is now.
I probably learned most about transitions and text effects in this, even though there was one effect (the wind-sweeps-letters-away text effect) that I would have liked to utilize for the “relax” at the very end but I couldn’t find any tutorials on it, so unfortunately it stayed a pretty lame fade. I moved it upwards to still convey some feeling of easy and lightness, but it’s as stiff as can be. Next time.