SIDigital Animation
Development Work
Here's an overview of tests and work done across the Portfolio unit reflecting my role as a 3D Character Performance Animator.
Finding an Audio and Conceptualising
During the module, I was really struggling with knowing what to animate since I felt a bit lost in my direction. After a discussion with my tutor, we discussed how I could move forward and they made suggestions for next steps:
-
Find an audio I liked and felt had the potential to develop further into the stylised/cartoony animation style I would enjoy,
-
Find a rig (or rigs) I liked and see if it has the capabilities for my idea and the animation style I am specialising in; and,
-
Develop the performance in stages
When searching for an audio, I felt I wanted something tonally dramatic. During my time screenwriting, I found I was more able to place the dramatic elements of my writing and use comedic moments to tie them together. In one of my scripts “Moments by Moments”, I was able to write down the dramatic elements and piece them together with humour to have a neat little flow.
The audio I chose was from a series called “Pantheon” on Amazon Prime Australia/New Zealand, previously on AMC+ and HiDive. The final five minutes of the series finale connected with me. The sequence was about beginning again, and revisiting a moment in time repeatedly. I wasn’t sure which one to choose so I clipped out a various number of sections of that five minutes until I decided on this 15 second clip below.
Finding Rigs
When searching for a rig, I wanted something that felt exaggerated and had the ability to give the character a loose anatomy and clean lines of action with curves instead of hard corners, such as elbows or knees.
For my rigs I chose the Superhero rig from Agora Community and the Thep rig I found online. I felt Thep was a rig I really liked, having a stylised and cartoony type animation that I enjoy. While these two character rigs styles may not match, I wanted that visual difference to really highlight the time between the two; akin to immortality.
Tutor Discussion
I pitched the idea to my tutor, and it was fairly positive. My tutor liked the audio. We discussed certain nuances in the audio: like when the girl says “flicker”, and the emphasis placed on it. They liked it was dramatic, but were aware of my desire to add humour to the situation. He suggested making the humour more “situation-based”. Like these characters keep reincarnating and are sick of it, or are nostalgic for this moment but are starting to get sick of it.
We also discussed the Superhero and Thep rigs. My tutor suggested using Thep rig for both characters, since it will give a consistent character style to the environment and there isn’t another known accessible rig similar to Thep’s style.
Since there are two different voices, I should attempt to find a way to differentiate the male and female character by adding a prop to one of them, like a tie or a flower. This will allow for better visual difference in audience understanding. I’ve often struggled with making stories visually readable, so I will implement this into my work.
Script Notations
To begin my development of the audio, I jotted down the dialogue on my screenwriting software and listened to the audio repeatedly to capture the ideas within the nuance of the dialogue. Blue is dialogue nuances, red is narrative implications, and pink refers to timing notes based on pauses. I then developed four main shots (below) which focused on the staging of the charaters with Thep in mind.
Based on my explorations (left), I also developed my final imagery (below). I spoke to my tutor about this and he told me the composition is focused on the centre of the shots, so I should challenge myself to make the shots more visually interesting.
Jaw Bounce / Staging
I started with the Jaw Bounce and Ins & Outs of both Theps. I do this to give myself a foundation for the lipsync when I would develop the animation further. Based on my plan with camera and character movements, it would be difficult to have a good view of the characters face. I also did general staging of my characters to communicate the story in my thumbnail drawings above, prior to my discussion with my tutor.
Redoing Staging Test
To experiment with a new software and to give myself the freedom to experiment with composition, I did another staging tes in ProCreate Dreams (which just recent released). It was fairly time consuming since I'm proficient in 2D drawings, but it was a way to get my head out of Maya and simplify my vision and ideas outside of my main software.
Reference Footage
For my reference footage, I knew experimenting with lighting could be interesting to explore. I imagined the characters in a pyramid (due to Thep's mummy design) and talking about the aging of their lives in this old, worn and historical location; far from the eyes of modern day humans.
Cameras
Using my Staging Testing in ProCreate Dreams as a reference, I restaged the camera to fit the vision. An issue I recognised was Thep's head being cartoonishly exaggerated proportion in comparison to its body. I had to adjust the cameras so I could get a similar framing while also being able to get a clear silhouette.
Developing Animation
I began developing my animation by creating a strong foundation of poses. I first focused on the parts of the body that were moving: such as the arms and the head. Then I moved forward through the spine and the secondary movements, such as the other arm and neck. Afterwards, I focused on detailing with the eyes and mouth movements; which I then splined and polished as a test. Overall, I wasn't quite happy with it, but I couldn't descern why.
Presentation / Tutor Review
For my FMP Pitch in December, I had to present the work I had done, alongside my initial pitch for my Final Major Project. (For more information, see FMP Pitch) The overall feedback I recieved was that my animation doesn't capture the exaggeration that my role and my reference provides, especially considering my references were excellent resources for exaggeration. Which I was surprised at how obvious that seemed since my focus was on cartoony and exaggerated movement. Since I was so focused on my work, I needed to take a step back from my actual animation and return back with fresh eyes.
Thep Toon Shaders
Since I was going for a stylised animation performance, I took the initial version of Thep (left) and applied Arnold Toon Shaders to give that stylised look (right).
-
I encountered issues with the edge colour. I felt it was too thin, but any thicker would affect the mouth lines and I did not want the lip and inner mouth to be affected.
-
I tried black edges for the character outline but it didn't look right, so I did a more brown/rust-orange colour and it felt better.
-
I realised the eyes did not have a texture, but a ramp effect for colour. Since the eye was easily editabel within Maya itself, I realised that might be more interesting with discerning the two characters with different eye and brow colours.
-
I felt in my experiment (right) the texture lines of the mummy wrapping weren’t being tooned since the lines on the texture were not actually on the character, but I didn't know how to make it stand out more.
Since I was having difficulties, I spoke to my Tutor and received feedback on the Toon Shader experiment.
-
Main issue was that the skydome was showing on the camera viewport, so Thep was on a white background. This was an issue because pure white backgrounds are EXTREMELY unlikely to exist, so my view was being tricked in a way with that. My tutor put the Arnold background to black or grey by reducing the camera view, and suddenly my view of my characters colours and textures were much less appealing but in a way I could recognise and easily fix.
-
The texture lines for the mummy wrapping was fixed by making the weight of the Base Colour (Texture) 1 instead of 0.7 to have the texture accentuated, and also reduce the exposure on the texture colour so it maintains that sort of dusty feeling.
-
For the edges, my tutor focused on Silhouettes in the Ai Toon Shader attributes. This would focus more on the line thickness around that character, while avoiding messing with the lines within. I later used this to match the respective colour of each Thep to give a form a visual continuity in design.
Environment Layout
Since I imagined the environment being a pyramid, I decided to create a rough version of the environment. I had been using a wall asset from LateNightDesigns on Turbosquid for the initial shots. I duplicated and arranged the walls to form a pyramid-like structure and applied an overall Toon Shader to it with a low-saturation yellow-orange. I considered utilising the workshop on Substance Painter for more accurate pyramid/brick textures, but my main focus was the animation so I settled for the all-over colour.
Revisiting Animation - Blocking
After my feedback from Bring Your Own Animation in January, I knew I had to revise my work I did. After the Christmas gap, I had a fresh perspective when revisiting my work and I did some tweaking on the camera staging and blocking. I deleted unnecessary key poses, followed Kieran's critiques from the event, and close attention to my reference. I definitely see improvement since my initial presentation in December.
Revisiting Animation - Breakdowns / Blocking Plus
Breakdowns were fairly efficient. By trying to achieve the attention to subtle movement in storytelling, I ended up focusing on Blocking Plus (the ending stage where I focus on smaller details like eyes and blinks) alongside breakdowns. In the example of Shot 2, I focused on the movements of the fingers and the overshoot and settle.
Splining / Cleanup
I used auto-tangents to spline my work to see how the movement came when not cleaned up (left): the blink in Shot 3 for the man was going to the next key pose which was an issue, and there was a lot of clipping in Shot 4 when the female Thep was moving about. For splining, I started with the biggest control and moved my way to the smallest.
Lighting Test
When I was ready to put my animation through Arnold Render, I encountered an issue. I realised I referenced Thep using afile which already referenced Thep from a separate file in my sequence, and could not replace it with the original Thep or another Thep file without losing my animation. I attempted to redo the shaders in the Maya scene itself with all the animation, but because it was a reference of a reference, it would not let me change the textures on the characters without importing. I think this will be a fun challenge for the summer, but in the meantime, I playblasted this in Maya with basic lighting. I wasn't able to get the more dramatic lighting with a spotlight due to my issues with Arnold, but I did cheat by using a Polygon cone for the spotlight effect.
Conclusion
Ultimately, I think I made an error with my reference footage since the reference focuses more on subtle movement instead of exaggeration. Personally, I wish I was able to redo the reference to better suit the characters, or used a different rig since Thep doesn't feel right for these very subtle movements and rather should be exaggerated. In addition, my feedback session from Bring Your Own Animation in March made me realise when I was struggling to develop poses, it was because I was focusing on detailed movement rather than the spine for determining key poses. Overall, I am pleased by the efforts I put into this, but I think the errors I made in this experimentation were good lessons to learn and incorporate for the future, especially with the referencing and my ToonShader issues.
Software Experience
An overview of the software used in development, and my experience level with each prior to the module. This is to give a general idea of my skill level and technical perspective prior to unit.
Autodesk Maya 2022
Intermediate animation and modelling proficiency in software, having 2 years of in-depth experience. Confident in animation ability with software prior to unit.
Adobe After Effects & Media Encoder
Intermediate experience with software. Prior experience has been in relation to compiling Maya sequence renders into H.264 format.
UnReal Engine 5
Beginner experience with UnReal Engine prior to unit. Attempted one project transfering Lion animation and using cameras to a degree of success. I have a general understanding of UnReal, but more research would be required.
Blender 3..6 / 4.0
No prior experience with Blender.
Adobe Substance 3D Painter
No prior experience with Adobe 3D Substance Painter.
During Summer for Portfolio Module, I started a variety of animations but was unable to finish them due to personal reasons, timing, and issues that popped up. I decided to put my initial development of these projects here since I decided to start afresh during the semester. After the Portfolio Unit, I may revisit them; but it is important to show the work that I developed, regardless of how finished it may be.
1 Hour Animation - Jump
On X (formerly Twitter), I follow 1 Hour Animation account which posts an animation prompt for each week and the artist can use the week to create an animation based on the prompt. I attempted the "Jump" prompt using a Tom the Cat rig.
1 Hour Animation - Curious
Continuing my following of 1 Hour Animation, I used the Tom rig to follow the "Curious" prompt. I recorded my own reference and developed a further understanding of timing and developing golden poses (the main storytelling poses).
Wolf & Dog - Lipsync Animation
To experiment beyond my comfort with biped animation, I decided to challenge myself with a quadruped performance test. Using an audio from the 11 Second Club, I did the jaw bounce for the Wolf rig.
An Inspector Calls - Performance Animation
I wanted to experiment with an older sound, so I found an audio from "An Inspector Calls" (1954) showcasing a conflict between an unempathetic man and empathetic woman. I did the initial jaw bounce on Ray and Tina rigs and recorded reference for the animation.
Bedford Falls - Previs Test
For a Previs Test, I had an idea of a familial conflict surrounding marriage based on a song called "Bedford Falls" using a dialogue from the film "It's a Wonderful Life" (1948). I showed a previs pass to my tutor, who informed me Previs focuses more on cameras and action rather than acting performance so it may be better suited for me to explore that instead in previs.
I entered the Portfolio module confident in my blocking abilities, but not for my splining and polishing skills. I wanted to do a test improving my splining and polishing skills, so I took my old performance animation from one of my previous modules to an audio from "Moon Girl and Devil Dinosaur" before the Portfolio module. The below videos are the final spline I submitted (left) and final stepped (right).
Based on the feedback I recieved, this was my best piece on my showreel, but it required additional polishing passes to develop further. My research from recruiters gave advice on returning to old animation pieces and improving on them. So I did exactly that and attempted Arnold Rendering passes to beauty-ify the animation so its pleasing to look at. The point of doing this is to: a) improve on my splining and polishing skills; and, b) making a piece showreel-ready with lighting and textures.
Tweaking Aspects
The first thing I had an issue with was the costume in the back. It felt too far away and not on the same eye level as the characters. It's a simple fix, but makes the joke much more effective. Initially I think I was worried that the costume in the background would distract from the silhouettes of the character, but this was a decision where narrative overrides technicality.
Auto Tangents - No Cleanup
Since I wasn't happy with the splining and to improve on it, I decided to undo all the splining I did and turn the animation back to stepped in the graph editor.
I couldn't remember what I had/hadn't tweaked with and so by seeing the animation as it was, I would be able to notice where I exaggerated and where worked better.
Cleanup - Shot 01
I started by dividing my animation into each shot so I could focus my attention to a specific part of it at a time. Shot 1 had a few issues I resolved in the splining:
-
When the Client says "So-?" and leans back, there's a glitching effect of the arch back and not getting a smooth movement.
-
When the Client says "What do you think?", the movement for the waist control did not feel like a clear swish.
-
When the Manager speaks with his arms crossed, his arms move around as if they were alive instead of being grounded in the arm cross.
-
When Client says "Yeah!", the movement feels very linear. Instead of a flowing movement, it hits the three points and settles.
Polishing - Shot 02
Shot 2 was a lot less intensive since it was a dirty two shot, so the main focus was on cleanup for Manager with some clean up on Client.
-
When Manager says "No-o!", the movement of the arms felt robotic and didn't really capture the feeling of a sassy and firm "No".
-
When Client says "What?!", Manager's shoulder flickers a bit.
-
The switch between the "No" and the pointing to the "outfit", the movement doesn't feel as efficient as I would like.
-
When Manager points to the outfit, the movement hits the pose and feels very static. I wanted it to feel more alive, so I added overshoot and settle.
Polishing - Shot 03
By contrast, Shot 3 is a more complex shot than Shot 2. Combining a walk and a character-holding interaction, there was a lot to focus on. I started by cleaning up Manager character's animation first, since Client's cleanup would depend on Manager's.
-
The poses I keyframed had some general tweaks to be done, like hip controls or the footroll.
-
When Manager leans back for "No-o", the flow to the exaggerated lean back is very abruptly linear and not a clear transition to the overshoot.
-
The walk for Manager has certain tweaks, like when he lands on his right foot after the exaggerations, he leans forwards far too much and felt like his chest was broken for a moment. Breaking Joints is a fundamental aspects (see "Research") but this was not one of those instances to apply it.
-
At the end, the Manager's left arm swings from side to side as he grabs Client. It didn't feel right, so using Follow Through and Overlap to convey Manager has made his arm go limp and focusing on the one around Client.
Continued to Client (orange). For this, I divided this shot into two parts: first half when Client spins to the costume and back, and second half where the Client gets pulled into a hug which requires its own focus because of the secondary character action.
-
In the first half, there was a lot of unnecessary movement that felt like it was popping, so I corrected that. I say that easily, but it took a long time to utilise the difference between stillness and not moving at all.
-
I didn't like the holding pose in the first half, so I improved that a bit more for emotional readability.
-
The second half was a lot easier to clean up since it focused more on smooth movement and I could really visualise the arc of the hand as the arms swing.
-
The arm swing and settle felt stiff, so I offset the rotations for the clavicle, elbow, and wrists (like a chain) to give the natural movement and settle.
Arnold Render Tests
While my animation is ready to be shown to animators, it is important to beautify my work in order to make it something visually appealing to look at. Based on the workshops we experimented with Arnold Renderer, I thought I should experiment with skin and sub-surface scattering given the number of characters in the scene to go from doll-like to feeling actually human. I thought to experiment with Toon Shaders, but since I experimented with that with "The Eternal Question" I thought I'd challenge myself with skin.
After I did my first Pass, I realised there was an issue with the subsurface scattering and lighting: the teeth and fingers yellow/blue-ish and made the character looked sickly. At first I thought it was because of the subsurface scattering, but it was actually because of the AiSkyDome I was using causing an overemphasis on this. By reducing the intensity of the SkyDome and using three directional lights for three point lighting (see "Development Research - LookDev Workshops")
Conclusion
Overall, I'm really happy with how this came out. I definitely do think this is one of my best pieces yet and I'm glad I revised through it. I think I finally understand the graph editor somewhat, which I think is a big improvement from before the Portfolio module where I didn't know how to do it. I do think my lighting/rendering requires more attention for a fully polished showreel, but I am glad I at least tried.
In my Company Research, I found out that Mikros Animation uses Maya, Blender and UnReal Engine on in their pipeline. Since Blender is developing itself as an important tool for animation, not quite replacing Maya, I thought it would be good to learn and experiment with the tool for an animation performance. Since my tutors had little to no experience with Blender, I would mostly be self taught in how to use the software.
Choosing an Audio
In my research for my Dissertation, I was watching reaction videos to the videogame "Life is Strange" (2015). During this, there was an interesting audio clip from a reactor (Not-So-Average-Fangirl) switching from suspicion of a character to excitement of "Belgian Waffles" being on the menu and one of the options the character could eat. I think the switch in emotion during this clip made this audio very interesting to explore, especially since it was six seconds, so a lot shorter than my Maya audios which I was fine with since it would ease the pressure of the amount of animation I would have to do in Blender.
Researching Rigs
When I was searching for rigs to use, I came across Blender's official website which has a number of rigs I can use. I was interested in exploring the Cat or Dog rigs, however I recognised the rigs were paid and used more animalistic features which I wasn't familiar with; since I was experimenting with an entirely new software, I wanted to animate a character more human-like since that's what I am familar with.
Since the voice was fairly feminine, I decided to begin animating with the Victoria (right) but I encountered an issue with Keying Set (this will come back later, don't worry about it) so I stopped using Victoria and went with Ellie (left) for both characters.
Reference Footage
For my reference, I knew the context of the scene meant two characters were in it, and there was a surprise of Belgian Waffles. My main focus would be animating one character instead of two, so I imagined the characters at a diner or some kind of breakfast restaurant. Using the paper in my hand, I felt as though it could make for interesting dynamic movement.
Maya Preferences
During my development, I found Blender difficult to navigate since the controls felt so unfamiliar after using Maya for so long. I did some exploration of the preferences and found a way to have the interface like Maya by going Edit > Preferences > Viewport > Themes > Maya. This made future work and navigation so much easier during development of my project.
Setting up a Project
Audio Setup: Doing this in Maya is far more intuitive to me, but I figured out how to do this. To put an audio in Blender to animate with, I went to the Video Sequencer Editor and inserted the audio. Then go back to the Dope Sheet, Playback > Audio > Scrubbing (ticked) to have the audio play at the same time.
Reference Footage in Viewport: In Maya, an image plane needs to be created and using reference footage with image sequence. In Blender, you just need to drag and drop an MP4 file on the viewport and place it where you want.
Grabbing Controls: Blender has a variety of tabs based on what specalism the user wants to use. I went to the Animation tab to animate, but I couldn't select individual controls. I realised that there are different modes: Object Mode and Pose Mode. Object Mode allows you to choose an object in the animation tab and move it globally in the space, similar to the global control in Maya. Pose Mode lets you choose controls on a rig to pose your characters or objects a specific way. I was on Object Mode for most of my initial experimentation, but once I figured that out, it was easier to select them.
Blocking - Key Poses
To set key poses in Blender, I used the Timeline and the audio automatically synced. At a certain point, the Insert Keyframes button didn't work, so I had to duplicate keys and place them where I wanted to insert a new key pose. It was a very inefficient workflow, but I managed to do a lot of key poses to get the initial blocking pass.
"Keying set failed to insert keyframe"
This was probably the biggest issue I encountered in the entirety of my time in Blender. When I key my animation at times, I recieved this error and I was unable to key new frames. A 'Keying Set' is a set created to be able to key multiple poses at once, much like a Quick Select Set. When I was experimenting with Victoria this issue came up immediately, so I assumed it was an issue with the rig itself. So when I switched over to Ellie, I realised this was a general interface issue.
I attempted to resolve this by deleting the Keying set, but regardless of deleting it, it seemed to randomly return back. I've noticed this is a general issue that recurs again and again. I asked a variety of animation discord communities and online resources (see "Self Promotion - Social Media") but I recieved no response. So unfortunately, I have no answer for this or how to resolve it.
Blocking - Redoing Key Poses
After my visit to Bring Your Own Animation in March, I realised I was following the key poses of the hand instead of the spine, which was really lovely to find out why my animation wasn't working as well as I wanted. So I redid the key poses of the animation, focusing on the spine instead of the hand and it looks a lot better than before.
Blocking - Breakdowns
Since a lot of my poses required cleanup, I decided to do it simultaneously with my breakdowns. My first cleanup was on the spine. I did lose a fair amount of exaggeration in it compared to my previous developments, but it was the price to pay for efficient movement. The next major cleanup were the eyes. The eye controls aren't parented to the head, so I had to manually position the eyes each time.
Splining / Cleanup
Splining in Blender is similar to Maya, but there aren't any easy shortcut keys for it. To do so, I went to the Graph Editor, then Key > Interpolation Mode > Bezier for Blender's version of auto-tangents. To switch back to stepped, it is Key > Interpolation Mode > Constant. I was mainly worried since I was having a hard time with using keyframes in Blender, so I thought it would work better to spline and then animate with the spline poses for more efficient development and cleanup.
Cleanup in Blender is really interesting. The graph editor is fairly similar to Maya's, but there were some interface differences my subconcious was trying to fight against: the key's tangents appear regardless if you have selected the key or not, since they were similar to the naked eye, sometimes when I'm trying to ease in/out a key I would select a tangent instead and vice versa.
In Maya, selecting a channel in the Graph Editor would isolate all the keys and graphs for that specific channel. In Blender, the channel you choose gets highlighted, but all the other graphs are still there. I had to toggle the eye button for each channel to isolate an clean up each one. It did take me a few tries at cleanup to get used to this, but once I did, it felt like smooth sailing.
Overall though, splining wasn't much different from Maya, and using Ollie Trotman's advice on choosing the biggest control and cleaning that up first made it more bearable. At some points, I didn't have enough keyframes for a proper movement, so I duplicated the controls for an easier cleanup.
Experimenting with Blender Lighting Styles
Blender has a variety of built-in different lighting styles in their animation render, so I decided to experiment with the renders/playblasts to see which one I liked better.
Studio Lighting
This is the equivilent of playblasting in Maya, where there's no lights and the focus of the shot is focusing on animation and movement. I used this for all my playblasts and viewing my animation's movement.
MatCap Lighting
This is Material Capture lighting, where material show their real-time attributes being lit apprpriate to their texture: specular, matte, metalness, etc., all appear as their attributes assigned appear. I liked this one the best for my animation since it doesn't show detailing like Studio Lighting, but nothing is lost in it like my experimentation with Flat Lighting.
Flat Lighting
This flattens the textures so they appear more toon-like. While I do like the Toon attributes, I do think this is focused more on clear silhouettes and cartoony exaggeration which fits my role, but not this specific piece. I am interested in experimenting with this lighting in the future with clearer poses, but I think a lot of the detail in animation gets lost in when lit this way.
Conclusion
Overall, I'm really glad I experimented with Blender. I think the software was really interesting, but my unfamiliarity with the software was the main thing holding me back along with my confusion surrounding the means of interface. The Maya presets helped a lot, but I definitely think I need more time with the software to develop my animation skills with it; similar to how a traditional artist needs time to adapt to the digital medium.