Dynamic – depending on the lighting, either white/black outline will be present.
2. Black filter:
Makes everything black and white
The second part of the post-process was completed early on by David; this only left the outline. Originally, John was tasked with this, but I ended up taking over, as they didn’t have access to Unreal over Easter.
Henry sent a couple shaders and I did some supplemental research into unreal shader settings and using shader nodes in unreal to find a solution to the outline. I tested these in game; however, none of the outlines did a white and black ‘dynamic’ outline.
Henry made a shader which was perfect but unfortunately in the wrong Unreal version. However, I was able to recreate it following his example as a guide. (details further down)
https://www.youtube.com/watch?v=NtPq6NDWNyQ
As I recreated the shader, I ran into some issues. Despite understanding how the shader worked, I lacked a lot of basic Unreal knowledge, like how to delete node links.
I read Unreal Engine’s guide to familiarise myself with the correct language and UI.
In the correct version of Unreal, the shader’s material functions were missing. The main issue was that certain nodes had different names or contained other custom material functions that had to be remade. I roughly figured out the use of every missing node and used the resources I found to create replacements.
Material functions
Material Functions for shader:
Pixel_Depth: Calculates the distance from the pixel to the camera.
Get_Occluded: adjusts visibility of outline based on occlusion.
BlurSampleOffsets: Calculates the outline by multiplying the object and offsetting it in each axis.
CombineSamples: I couldn’t figure out its purpose but found that changing the input values affected the background. I believe it converts values for 3D rendering on a 2D screen.
get_subject_depth: Affects the continuity of the outline
proper Subject_depth values
Wrong values
Edge_Detection: Adjusts line thickness
Edge_detection_1: Unsure of its function
Final Material
I managed to recreate the material using the newly remade material functions.
Final Results
I’m really happy with the final shader. I learnt a lot about post-processing in Unreal, and although I relied heavily on Henry to build this shader, I’m still proud I managed to reverse engineer it and understand most of its components.
Cutscene
A cutscene was planned for the game to introduce the setting and plot; however, we unfortunately ran out of time and could not complete it. The following was the beginning stages of the storyboard I made for it.
Feedback received from the style guide presentation:
Blender shaders can’t be transferred to Unreal .
Henry offered to help us with the shaders in Unreal Engine.
The models would likely need really exaggerated geometry to achieve the angular style.
Models:
Modelling
I followed a polymodeling pipeline, as we discussed having a low-poly style.
First attempt: I created the facial features from a subdivided cube; however, it felt primitive and difficult to develop and add detail to. I used a Blender shader to replicate the final black and white effect.
I researched other methods. This was another way to go about polymodelling, which I recognised. You build the loops first, then join everything after.
Second attempt:
Improved the facial details
Eye shapes were more accurate.
Additional topology gave more freedom for facial animation.
Used more extreme geometry, e.g., cheekbones, nose and chin shadows. (This creates built-in shadows which recreate the 2D style from all angles)
The shading sometimes looked off despite the topology looking good – to fix this, some vertices had to be angled weirdly. I asked for advice, as I felt the topology was starting to look crazy, I was told that good topology is whatever achieves the correct shading with the least amount of geometry possible; only N-gons are considered objectively bad. I developed really weird geometry but it functioned well in different lightings, so I stuck with it.
I recognised the model needed textures for details like the eyes and beauty mark. The eyes were a big issue, as they appeared pitch black but needed to have some white and shape to them.
Ex. Weird shading
I identified an issue in the shadows caused by the topology:
At poles, shadows would blur or extend oddly.
This also happened in quads if one vertex was pushed too far out.
I could work around this by:
putting poles in inconspicuous areas .
readjusting pushed-out vertices.
To-Do & Notes for solving issues with in Unreal.
I followed the same method as the face, building each limb separately, then connecting them at the end.
redid the hair.
connected all topology
Redid the skirt to have built-in pleats so it’s affected by the shader like the face.
hands
I made blocky hands that fit the angular style; however, when I looked at the topology, I realised that it wouldn’t work well for animation, so I chose to redo it entirely.
Using a reference, I remade the hand. The number of loops could be reduced to emphasise the angular style; however, the topology now works for animation.
Connected the feet and hands to the body.
tested the model in different lighting.
Unreal Test:
flipped some normals.
Textures
My rough attempt at the textures; I used Substance Painter rather than Krita for more control.
I decided where I should put the textures/lines.
The textures looked blurry; needed an increase in the resolution for the hair.
I ‘drew on’ the eyes, as the original method didn’t work; however, it didn’t look good.
Second attempt:
UVs were overlapped in some areas, which made black lines appear randomly on the model.
My notes mentioned using a white emissive layer for permanent highlights that aren’t affected by the shader.
I redid the mesh and UVs, as they were affecting the resolution of the textures too much.
The main change was the eyes– I used a technique I learnt for making Vtuber models: make the eye a flat surface and create the pupil and eyelashes on top as planes.
Final model with textures
added white strips to the skirt so it could have highlights
new eyes
higher resolution hair
fixed UVs
Rigging
I used the robot tutorial to build the model’s rig. Coincidentally, all the previous models I made in other modules were serpent-shaped, so this was my first time making a human rig.
I made all the controls and adjusted the weight painting.
I wanted to push myself and try making a face rig too; I looked up a couple of videos to become more familiar with the general concept; the second one helped the most as it slowly built up in complexity.
I used the second video’s 3rd example. The deformation of the face isn’t great, which limits the exaggeration possible, but these basic functions work:
open mouth
eye tracking
eyebrows
To improve this I would:
add more mouth space
define lips more
Be more specific with weight painting.
Made a skirt rig for extra control.
rigged model in unreal
Fix: Removed weightpainting on IK wrist control.
Animation
I researched idle animations to understand the general requirements for these animations. I noted the following:
Looping actions
‘Small’ or contained actions.
Simple
The main action I chose for the animation is a ‘FootTap’, which works as a simple loop; however, the game team requested a smoking action and a few randomisable actions. So I created a couple of different movements based around smoking that the secretary could do at the same time.
Smoking references
I researched depictions of smoking in film noir and animation and combined them with the basic foot tap aniamtion I made.
As requested the movements were split into multiple actions so they could be randomised in game.
Although I was tasked with making the secretary’s model and animations, we needed the background made to test textures and contrast, so I pivoted and worked on that.
First attempt:
tried to make built-in blinds in the room, but they looked half-finished.
Final room
Redid the room:
Added a door.
New blinds.
Textures
To create the textures, I brought the UVs into Krita and handpainted them. I marked the UVs in Blender so I could line up the pattern properly.
After testing it in Unreal, I decided to set the wall texture to a white emissive so it wouldn’t be affected by the shader to help the scene’s clarity.
Flooring
Hand-painted textures in krita.
Blinds
Feedback from the lecturer:
Redo the blinds
make them thinner and more.
Original Blinds
I improved the blinds by:
bettering the topology
Making the string mechanism accurate.
added a wood texture using the Krita technique.
Added a slight curve to the individual strips for more realism and interest.
I was struggling to understand how to UV map in the new Blender version. This video explained the new unwrap options.
To optimise UVs for gameplay, I overlapped individual strips so less memory is used on textures.
Textures
Textured in krita.
The orientation of the top section was wrong.
Fixed the orientation
resized the textures
However, it still looked bad.
The texture was too harsh but felt closer to the style of the game.
These had a similar issue: the texture looked like static when applied to the model.
I chose a subtler texture that had most of the detail around the edges to avoid the static feeling.
Upped the contrast.
Resized it a bit.
Textures I made: walls, blinds, flooring, and secretary.
I’m happy with the results; the new technique saved some time. Despite only using black and white, the scene doesn’t feel overwhelming.
Fireplace
I made a list of all the items left to do. Although I still had to do the animation for the secretary, I felt helping the team with the textures was more important with playtesting coming up. I chose the fireplace, as it stood out, making the room look unfinished.
I did the Krita-Blender texturing. matching the orientation of the brick texture to the labelled UVs.
I’m happy with the amount of work I completed. Although the quality of my work could’ve been improved, mainly the animations, I couldn’t spend time redoing or perfecting anything as other work needed to be done. Furthermore, I unexpectedly had to take on the post-processing, so I had to manage my time to get everything done.
I suggested a film noir inspiration for the visuals, focusing primarily on achieving a pure black-and-white aesthetic. Rachel advised us to research certain media for the story development.
development
Following feedback, we reduced the scale of the game to one location.
I organised the animation team.
Made a list of all the work we needed to complete.
Sent examples of previous years for reference to help everyone understand what we were working towards at this stage.
We didn’t have any set jobs; however, the general work distribution for this stage was:
After researching noir films and the yellow room, I came up with a couple of fitting ideas for the game.
Naming the Game
The reddit thread offered insight into creating a detective game. Someone cited a thesis on players’ agency in detective games; the thesis concluded that players experiencing choice when collecting information and utilising instrumentation will give them a sense of agency and make gameplay enjoyable. I used this when coming up with ideas for the ‘mini-mini games’.
Mini-mini-games ideas
Style guide & research
I researched industry-level style guides and noted what I liked and didn’t like. I found that a lot weren’t clear enough at first glance and felt like a game of spot the difference because of unclear labelling. I used lines and arrows to avoid this issue. I did like the comparative drawings for the ‘yes’s’ and ‘no’s’ and how the Batman guide showed how highlights and shadows function; this became a huge factor in the post-processing stage.
I believe the target audience would be 20-25, as the comic style attracts teens/young adults.
Noir has an older age demographic, but I don’t believe that would impact the main demographic, as film noir has mainly been used as a reference/influence.
The combination of noir with graphic style caters the game towards a younger audience rather than the usual 30+ noir targets.
The aesthetic might attract creative people that appreciate the reference and unique artistic direction.
Film Noir research
https://www.youtube.com/watch?v=jsmVL7SDp5Y
Noir cinematography research:
Noir’s black-and-white moody aesthetic was born out of necessity due to lacking finances and inexperience with cinematography (leading to minimal lights available and inexperienced lighting sets properly).
John Alton stood as one of the greatest cinematographers. “He had the guts to use little lights.” His skill perfectly captures the noir style– a purposeful minimalist, low-light appearance utilising lighting to create contrast and focus in a scene.
Technical guide
technical guide Research
I updated the team on my research. I did additional research for the technical guide, as I wanted to ensure the style I was building was achievable in 3D.
I found 3 main methods to contrast in the style.
Creating an alpha texture for the shadows, which would create a cross-hatching effect.
A black outline– using shader nodes, it’s possible to create a black outline.
Cell-shading– a colour ramp converts the realistic shader into a cell-shading style.
These techniques set realistic expectations and also showed how the models could look. Although it helped me confirm the style we wanted, it couldn’t, as Blender shaders don’t transfer to Unreal.
Technical guide discussion
I recommended using the Asaro head as a reference for the models to achieve the angular planes of the face.
The red arrow points to an idea that later developed into the style’s main factor for converting from 2D into 3D.
Comic book inker, known for his work in many well-known comics such as Spider-Man, Deadpool and more.
This video explained his technique for inking and demonstrated how inkers create visual contrast and clarity despite only utilising 2 colours.
I learnt of multiple techniques, like the ‘Halo’– this developed into the ‘dynamic’ outline that became a central feature of our game’s style. The outline would change colours so that it always creates contrast between the background and object, similarly to the Halo’s effect.
Other aritsts referenced
I used all the notes and analysis from noir and the ‘graphic style’ and techniques I learnt (from inking, shader nodes, etc.) to make it feel more authentic and create more effective contrast to create the game’s aesthetic and main concept art:
We were nearing the first deadline, and Aoife was sick, so I did the secretary’s design.
Secretary design:
Angular shapes.
Straight lines- Follows the set style but also creates a striking silhouette hinting at her hidden harsh nature.
Referenced 1940’s style for women.
Limited to black and white.
Carefully considered colour placement to create visual interest and contrast. Used a ‘sandwich’ method, alternating between predominantly white and black colour blocks to create contrast.
Environment design
enviroment & era Discussion
We were running out of time, so I also did the room textures.
I first placed her in a black room and took note of any legibility issues.
Black areas blend into the background, losing the silhouette.
Plain black floor makes her look like she’s floating.
Alternating black and white ‘W’ pattern.
The pattern with the wooden texture creates visual noise to contrast the solid colours of the characters.
I reference time-accurate wallpapers, but the patterns were too overwhelming for this style. It also increases the risk of tangent lines, and the white becomes overwhelming and distracting.
tried a loose and sketchy design
Vertical lines created confusion around the skirt’s silhouette.
Used diagonal lines to avoid many tangents with the character’s silhouettes
added horizontal lines to create depth, making the characters pop out.
marked the wall to help create a sense of scale within the scene.
Problems
The Style
We received conflicting feedback on the style; some lecturers saw potential, and others thought it was unachievable.
The primary issue was the limited colour palette made it difficult to see; however, having done research into the technical aspect and comic books, I knew creating visual clarity was possible.
Unreal Post-Process– everyone was inexperienced; John attempted to create the outline; however, it didn’t work.
Henry helped guide us; he explained the post-process volume UI, shader nodes and more.
Using textures to support the shader.
With this I understood foundational concepts for making the outline and working with Unreal post-processing volumes; for this reason I ended up doing a large majority of the post-processing.
Sound Design
Updated the sound designer on the game.
some of the game elements were removed following lecturer’s feedback.
Although I don’t have much knowledge on sound and music, Sophie asked for direction on the general style of the game.
I used my research into noir films to make up a visual and auditory identity. I made a guide that detailed the exact styles and sounds that matched the era and visuals.
Chosen Article: Vanishing Point: Spatial Composition and the Virtual Camera
Choosing the Article: I suggested choosing something that was relevant to our course, specifically our ‘creative industries’ module, as our classmates likely be more interested if they found the content relevant to their work. To reinforce this, I also suggested using our own uni works as examples, as that would be a unique element that would make it interesting and relatable.
We intended to split the article’s sections between everyone so we’d have a compilation of notes. I took 2 sections, introduction and the mediated and unmediated camera, as there were 6 sections; I created in-depth notes for 3 sections and made a short summary for myself for the remaining one (Athens did the notes on the other 2) as the note-taking was incredibly time-consuming.
Analysis and notes:
Additional research
To gauge if the article we chose was reliable, I looked into other articles.
Virtual 3D Camera Composition from Frame Constraints.
The article’s stated:
the virtual camera still has the same limitations as the physical camera
every shot has a ‘message.’
Other constraints of the virtual camera, such as constraints of software or constraints set by storyboard artists and models.
It felt as though Jones was underestimating the necessity of the established visual language. Jones’ also never took into account the rest of the animation pipeline. Despite the strong arguments and extensive research of this article, it was published almost a decade ago and given how much technology has progressed using this article as sole proof of the virtual cameras limitations would be bias and likely completely inaccurate.
Three Varieties of Realism in Computer Graphics:
Overall quite unhelpful as it covered virtual scenes not the virtual camera
the content of the virtual scene limits the camera- While not the same as the limitations of the virtual camera, one cannot exist without the other, and so I do believe this is still relevant to some extent
outdated
Directing for Cinematic Virtual Reality: how the traditional film director’s craft applies to immersive environments and notions of presence:
mentioned exactly what Jones stated in his article but with virtual reality
users will gravitate towards a medium that allows them to choose their own view and perspective
My Slides:
I found a presentation template so the visuals of the presentation were all uniform. I then made guides and outlines based on my notes and research on the module handbook and presentation the lecturer gave us in class.
I originally was only meant to do the conclusion slide; however, as someone did not show up, I also covered the reflection slides, using my own notes, research, and script. I made the template for these slides, and on the day of the presentation, I edited it to ensure it covered all points.
Having read the entire article in-depth and doing additional research I felt confident about doing these slides. In the presentation I outlined Jones’s point, why I disagreed with it, and what I believed was more accurate.
Reflection
Although the article used very complicated language, some of which was so overly specific to this article that I couldn’t find definitions online, I found it very interesting once I understood the core concepts. We had some issues with time management and two members did no work. However, Tegan and Athens did a great job on their slides, we managed to deliver a coherent and informative presentation. I made a guide, outline to the article/marking rubric and template for the presentation which I think was essential to our success as it kept everyone focused on specific points and made sure everything said fed into other parts of the presentation.
The point I found most interesting about this article was its description of cinematography as a language and the camera acting almost as an interpreter between director/animator and audience. The visual language has been built throughout the years of cinema using a physical camera, which now still affects modern filmmakers, and the virtual camera as features of the physical camera caused by its physical limitations and mechanics has become a staple in the visual language and is remade and applied to the virtual camera, which is free of those physical constraints.
I think the research and article I did will be really helpful to me in my animations not just as it has broadened my knowledge on the history of animation/film but because it changed my perspective on what the camera is for. I admittedly had a very ignorant opinion that the camera was purely an apparatus designed to capture an image and scene, so I never respected it much; however, the author completely changed my mind on this as I realised the camera is the way we communicate with the audience, and by not using the camera as an element, I was completely degrading my own work and making it harder to understand.
I mostly agree with Jones. The camera isn’t an apparatus however I don’t believe the virtual camera actually is infinitely flexible. The article established there’s a preexisting cinematic language that the camera uses to communicate the narrative with its audience, in my opinion we need to accept the camera not as a tool like history defines or even an element as this article does but rather as a combination of the two, there are certain cinematic rules that the camera demands as breaking these would mean speaking to your audience in a visual language they don’t speak. Meaning the ‘infinite flexibility’ is merely an illusion, as there are still rules we must follow to be able to communicate with our audience. However, I agree that the virtual camera is incredibly flexible, and if these rules are broken with intention, that itself can be a form of communication, like silence in a movie. When the camera stops speaking the viewer’s language, it opens a scene to greater interpretation and new depth, but that can only be achieved by respecting the established visual language.
I took two different videos and analysed the acting and movement using both as I saw fit. My main observations were a combination of extreme disgust and anger; I liked the first video, it leaned heavier into the disgust in some parts, even adding in some fake half-smiles.
I took pictures of a plush at a table to decide the staging of the scene, then got my mum to act out the first part so I could have a sitting reference, using the other videos for the rest. The whole scene is a combination of these images, videos, and my analysis, plus my own acting as I worked.
Development
Workflow:
find references for the emotions (along with using a mirror by my desk)
block out main body
block out face starting with chin
switch from constant to spline
fix glitches
add pauses and intensity back
lighting + details
I read through “The Animator’s Survival Kit”. Although I planned to do 3D a lot of concepts were still applicable.
This video was great for the technicalities of lip syncs and working out the specific mouth shapes and how they flowed into each other.
Although the focus of this module is on body mechanics, I wanted to push myself to create a mood and compelling ambience that supported the audio and emotion from the acting.
As it was the first day, I was just setting up the scene and prepping for the blockout. I ran into a lot of problems setting up the rig and working with the different versions of Blender, mainly missing controls, broken models, and more. I was only able to fix it by using this video, Blender 3.6 (which a lecturer okayed), and a couple of Blender support threads.
I looked up some videos to help me visualise the process better and have a better idea of the movement I’m working towards.
I tried to be very thoughtful with my blockout, taking into account posing, staging, and arcs. This did result in the head having too many keyframes at one point, as I wanted to mark the arc properly, but it just made it jumpy and awkward. This was something Rachel also pointed out and just advised deleting frames one by one and checking the change to see if it looked better. I really tried to make the blockout as good as possible, as my previous animation had been done directly on Benzeir, and they felt really overwhelming, and doing a blockout was not only recommended by the lecturers but also felt like a good way to stay organised and not get overwhelmed.
To make the blockout, I used my references and analysis but also my own intuition and a mirror, as there were some moments I saw potential to make better and more impactful than I had originally accounted for, for example, the bar stool falling as Rain stands up, Rain swinging her arm as she stands up, and the camera shake from the point.
I feel camera shake actually made a huge difference; I was quite insecure about this decision initially, as I find camera work really challenging, but after consulting with Rachel, I realised how simple it was to implement, and that small addition really improved the cinematography, which made me feel more confident about something that has previously been a huge struggle and also motivated me to improve the lighting, seeing how big a difference using good cinematography makes.
I felt the blockout was still lacking some emotion, so I looked back to the animators survival guide and found that I wasn’t utilising the eyes effectively, especially the upper eyelid, which was completely still; however, I was a little unsure if it looked like that because it was just a blockout, so I held off on making changes.
After switching to spline I realised the issue wasn’t fixed, and more problems arose from the switch to spline.
Because the glitching was so distracting, I decided to tackle that first. I fixed keyframes that went from positive to negative, which were the cause of the glitches, and I added supporting frames to bring back the strength from the blockout and convey that anger better. I also started animating the hair, which was challenging in its own way. I found myself overexaggerating it a lot, which distracted from the main animation; however, I wanted to have a lot of drag and follow-through to convey the speed and strength of her movement, which matches her anger. I should have found reference for this; however, I’ve had a lot of trouble in the past finding references for ponytails for some reason, and I wasn’t able to replicate that cartoony whip myself.
I added blinking mainly during the head turns and lowered the top eyelid and raise the lower one to create a squint to convey a feeling of pain.
I finally figured out how to render the video with audio; however, the audio didn’t sync up properly. I also added in the background at this point (previous steps’ renders were done after the background was finished). I used two lights: one blue one to the back right, which framed her in shadow to convey an underlying sense of sadness, and a softer yellow light at the front to light up the face better as it felt it was too dark and you couldn’t see her facial expression properly.
I finally got the audio synced up properly and added additional lighting to the bar to create three-point lighting, and the lighting didn’t feel dynamic and interesting enough, and the ambience wasn’t as effective as I wanted.
I chose to use 3-point lighting as it accentuates the height, width, and depth of the subject, making them stand out in the shot, and as in the previous lighting attempts, she blended into the background, or her facial expressions weren’t properly visible from the heavy shadows.
My fill light is the yellow soft lighting at the front, which helped with visibility; the key light is the strong blue lighting, and the backlight is the bar lighting, which gives Rain a soft rim light, making her stand out.
I tried to add the final part that got cut off but it messed up my timing so I decided to keep the previous version as my final.
Final Thoughts
Overall I think I improved a lot, especially in 3D; focusing on the blockout made the process a lot easier and better. But I’m especially proud of my improved cinematography; it’s my first time using 3-point lighting purposely and really paying attention to background and ambience to support the animation. I could’ve added more weight to the movements to better convey the anger, mainly in the hair, as looking back it feels a bit too smooth, which contrasts too much with the harsh motions of the arms. I’m glad I put in a lot more effort into my research and analysis, as it really paid off in both assignments and meant I didn’t waste any time and had fewer steps. I’m glad I took advice from the lecturers and implemented that into my work, though I’m hoping to have a little bit more confidence in my own decisions when my intuition tells me it’s the right choice, as it did with the camera.
I did research and compiled points to keep in mind while animating so I could better recreate the mechanics of movement. My class work getting deleted, so to support my work I did animation analysis and studies instead.
Notes from video:
Force: strength exerted on an object. The more force a person exerts on an object with less reaction the more heavy it will appear, If I want to make the weight believable I have to make the person struggle and exude a lot of force.
weight & balance
Pose- Poses are the main way to communicate weight. To make sure I did this properly I recorded myself and used many references and analysed and chose the best poses that conveyed the struggle best.
Centre of gravity: A weight supported by the centre of gravity of the person will always appear heavier.
Anticipation, arcs & follow-through: the heavier an object, the more build-up for the lift will be needed. I think this was the only point I struggled with as I there were some part my arcs were clear and there was nice secondary action but other areas were overlooked.
I liked the heavy weight in this animation and the use of the whole body to manage it. I also strongly believe that storytelling is an essential part of animation, so I try to include character and appeal in all my animations.
I added my own twist by making the person lift a fat cat. Despite this presenting a couple of challenges, overall I think it actually resulted in a stronger final piece with more technical skill, as I wouldn’t have considered some details if I had gone with a more basic approach.
Challenges:
Because a cat is an organic weight, its shape changes and the weight shifts any time it’s moved. To create a convincing animation, the person’s handling of the mass had to mirror that, which meant a lot of movement and shifting weight from one side to another.
Perspective: I opted for a simpler perspective to focus on the movement; however, that actually meant I had to do a lot of foreshortening, making perspective a challenge.
I showed Aodahn my progress for feedback, and the response was mixed but encouraging. The beginning worked, but as I suspected, the actual lift and walk backwards didn’t, it was awkward and relatively boring. Additionally, although I added the back steps to play with perspective and dimension to make the animation more visually intriguing, it actually made it harder to see the movement and cat, so the humor and appeal were lost.
The red drawings on the images above show the feedback given to me. There were 2 suggestions on how I could improve:
Flipping it forward
Swinging it sidewards.
I found it difficult to picture which would be most effective, so I recorded myself attempting both lifts. I used a couple of videos on lifting rocks so I could better replicate the weight of the cat and the struggle to lift it. I ended up going with the swinging sidewards, as it allowed me to explore more animation principles, mainly exaggeration, appeal, staging, and arcs.
exaggeration—although I could exaggerate a forward swing due to the angle, it wouldn’t look as good unless I pushed it into cartoony territory, which I don’t tend to like
Staging—this was the main factor in my decision, as the first option has the cat and lift happen in the same spot; the cat would end up blocking the person the whole time. The other option also does that somewhat but to a lesser extent.
Appeal—Although appeal is somewhat subjective, I felt the side swing was ridiculous but in a funny and impactful way, which made it more appealing.
Arcs—again, simply due to the staging, a forward swing makes it much more difficult to show the arcs in the lift.
https://syncsketch.com/sketch/NzhhYjVkZjQ5/
The SyncSketch link has the full video I used with the frame-by-frame analysis. These are a couple of standout frames that highlight my process.
I started by finding the key poses and marking them on the top left with kp-; I also marked some big shapes I felt would be helpful.
I then used a red pen to mark each change from the previous key pose; I looked for shifts in weight, elbows and knees moving in, the head coming down, etc.
Analysed the bag and the straps; this was helpful for depicting the follow-through of the cat’s legs and tail.
This the most time-consuming step ; however, it made animating very simple. I didn’t have many steps in the animating phase, as this analysis blocked out my key poses and everything for me.
My analysis and research meant the animating stage was fast and simple. I followed my notes and completed the person exactly as intended. Although the drawing is rough, I also think this gif best highlights the subtle movements of the person. There was a little bit of a size loss as I was working intertagibly from pose to pose and straight ahead, but I managed to fix it later on.
I asked for feedback on it, there weren’t any major issues it just needed cleaner lines; however, as the other module’s hand-in was approaching, I was told to focus on that. I was really happy with this animation and the feedback was really encouraging, although even then I still felt there was room for improvment. Mainly the cat missing some follow-through, and some additional inbetweening at the start would polish it further, but on the other hand, I wasn’t sure I had time, so I focused on the cleanup as I was advised.
I cleaned up the line and added the cat; I personally prefer the rough lines, but it does make it harder to see the shapes. I tried to add more squash and stretch in this stage along with other principles of animation. I exaggerated the drag on the cat as it swings to play on the humor a little and also adjusted the person’s arms so they would fit around the huge cat, as I hadn’t accounted for its size in the original.
This is the final animation. I cleaned up overlapping lines and fixed some other issues. Although I’m very happy with the person’s movement and feel it convincingly shows weight, given a little more time, I would try to add a little more animation to the cat, as there are some parts where the body is completely stationary, which breaks the illusion of this organic dead weight whose weight is constantly shifting, making it impossible to lift.
I decided to make a couple changes last minute as they were really bothering me
added more movement to the cats tail and corrected some awkward movement
added squash and stretch to the head where it was previously still
edited the hands a little so they weren’t as jittery
Final thoughts
Overall, I think I was a lot more organised with in this assignment, and the research and analysis of references I made were invaluable. However, I didn’t like the line weight, which actually made me prefer the rough version to the clean one. I also realised that I found myself gravitating towards straight-ahead animation more than pose to pose; thankfully, because of my planning, it didn’t cause any major issue in this animation; however, it was a problem for the cat, which suffered a considerable loss in size throughout the animation.. I feel I was able to successfully convey the weight and struggle of managing an organic mass.
Overall I really enjoyed this assignment, though there were a lot of challenges, and the constant switching between programs was difficult to deal with. As I worked through all the steps, I began understanding the effects from one section to another.
I think I’ve got a clearer understanding of the individual steps in the workflow and their importance and the tools utlised. To create a successful model, researching and understanding our 2D design was incredibly important. I looked into real-world references, animals, and other people’s interpretations of godly, snake-like creatures to inform my decisions, which I believe my design is successful, and despite not looking like a specific animal or mythological creature, it’s grounded in reality, and its features are familiar to the viewer, making it believable and realistic despite some of my choices purposefully going against realism. I also tried to pay close attention to the design principles. I found that creating a simpler design was actually quite challenging, as I couldn’t rely on many features for visual interest; instead, I tried to work with just triangles and cylinders to keep the shape language consistent and limited. I made this choice as I feel that by limiting the recurring shapes, the intention with the design becomes a lot clearer. Elongated shapes make the creature feel slender and elegant while the triangles convey the danger and power. Contrast was also another principle I relied on; I also took into account the background and tried to make the creature both blend and stand out in its environment.
I found my notes from class and the PowerPoints to be my most useful resource, especially when making the concept art and when sculpting; however, I also used YouTube, Reddit, Google, Quora, and just about everything else too to find solutions to issues I kept encountering. Although I looked at many sources for help and research, I realised I didn’t take note of quite a few, so when writing my blog, I couldn’t reference them; this is something I’ll definitely try to fix next time, as it greatly hinders my blog.
I did extensive research into my creature and tried to bring in storytelling subtly through the design; the tentacle shows it fights krakens, and it also has multiple recent and old injuries showing it’s quite old and strong. The tail is a lunate tail, which is characteristic of fast predatory fish and some sharks. I tried to communicate the creature’s speed and power through the tail. The fin bones also create a claw-like shape, reiterating the danger. Because of the simplicity of the design, I couldn’t add too many additional features or textures without ruining the intended effect, so I tried to use the design itself to convey the creature’s story and purpose.
Finally, I’m very happy with my creature; it told the story I wanted, and its function is clear in its design. I’m also a lot more comfortable with all the programs and actually understand how each one affects the other, which I previously didn’t get at all. I also think I greatly improved my skills in each sector of the pipeline, mainly creating the textures in Substance Painter and creating a realistically believable sculpt in Zbrush. My goal for next time would be improving the retopology and UV mapping as they lacked. However, overall I’m happy with the result and that the decisions that led to it had thought and research behind them, which took into consideration the feedback from the lecturer and my own research and judgment.
Given the creature will be translucent, I had to look into both the muscular and skeletal structure of whales, snakes, and other similar creatures to reference in my block out. Although the block out we did in class was very different from my creature, it helped me find a work flow that worked well. I looked to snakes’ skeletal structure for my creature’s spine but used whales for the fins and muscles. I noticed that all the references didn’t have defined muscles, though the shoulders had some notable definition to them, which I applied to my own work.
Due to the sheer amount of individual parts that make up the creature, it meant each step became incredibly time-consuming. Retopology for the spine was especially challenging; despite using ZRemesher, I found myself behind because of how long it took to do each individual vertebra. Given what I know now, I would take note of these repetitive steps and avoid applying array modifiers until the original object is completed. I also encountered this issue with applying seams/UV mapping and applying textures in Unreal Engine.
Overall, though, I feel I’m a lot more comfortable now with ZBrush than I was before, and I’m very happy with the arm bones, especially as I think I captured that hard bone look quite well. I used a combination of HPolish and Dam-Standard brush to define the planes of the bones, I learnt this from a workshop I attended.
Detailing
I began defining the muscles and details.
I kept it simple as I knew I wanted to do a lot with the lighting and opacity, so a lot of the finer details wouldn’t be visible, as well as the referenced I used didn’t have heavily defined muscles; however, it actually still felt like it made a big difference and turned it from a cylinder with a head to an actual creature.
details added: a chunk of the right shoulder being bitten off, a tentacle that was ripped off a giant kraken during a fight, and various cuts and scrapes on the frills, including some circular scars that are normally seen on sperm whales from fighting with giant squids, as this behaviour is very similar to my creatures. I thought this would be a nice detail despite it being quite subtle.
Retopology / UVs
The first time I did retopology, I relied solely on Zremesher, and at first glance it looked alright; however, Henry pointed out that what I believed to be loops were in fact spirals. I redid the retopology using Zguide to dictate where the loops were supposed to go. This was a lot more successful, and although it still had quite a few faults, I felt confident it wouldn’t impact the quality of the final render.
I cleaned up the topology as some polygons were missing and there were some random floating vertices. I had to redo the seams a couple of times as I wasn’t sure where to put them initially, which caused some parts to overlap.
The last few semesters I really disliked Substance Painter and found it to be one of my biggest weaknesses, so I was really concerned about this step, as this creature’s selling point is its bioluminescence and translucence. Given the pressure, I looked up a lot of videos and references on texturing and images on jellyfish to try to create the result I wanted in substance. I also found it important to put effort into the tentacle, as it’s one of the few areas that isn’t blue, so it would naturally draw a bit of attention, so the quality had to be good.
Here I had to deviate from my original concept art, despite the strength of the 2D being its simple elegance. When I tried to recreate it in 3D, it felt boring and uninteresting; the white and yellow highlights repeated throughout the design that balanced it and created contrast were lacking. I looked to one of my previous designs that utilised yellow circles that mimicked eyes; I really liked the contrast, but the design was too large, so I made it smaller and added a repeating pattern throughout the head frill to create balance in yellow for some contrast.
The body was also too simple in 3D; without hard lines, the body lacked definition. I really struggled with this problem. I tried adding a rougher texture similar to whale skin and then a scaly crocodile one, but the roughness clashed with the elegant, sleek design. I fixed the issue by looking back at my moodboard and referencing the deep-sea creatures; although their bodies tend to be transparent with one main color, there actually tends to be some dimension to the skin with more subtle colors and light. This prompted me to add a textured, luminous texture on the body.
I added an opaque flesh material hidden behind a black mask that is only revealed in the recent wounds, although it would be more realistic for the revealed flesh to be transparent; that would make those details virtually invisible and insignificant. This detail helped break up the texture a bit and added visual intrigue upon closer inspection.
The bones, like the concept art, also have bright white highlights. I found that the realistic lighting felt too monotonous, and when made that white color, it was incredibly jarring due to how bright it was. Again I contemplated the effectiveness of the design and what I need to be successful, and I decided that replicating the flatter, stylized highlights actually benefited the design greatly.
Rig/Animation
The weight painting didn’t work, and the bones weren’t parenting to each other properly. I manually assigned the vertices to the bones to fix it.
I kept the rig quite simple, as the pose and general shape of the creature don’t require a complex rig.
I used shape keys to animate the movement; although animation wasn’t required, I felt the movement was a huge part of the visuals of this creature.
(fin movement in blender)
Unreal Engine
This is the template I found, the effects and assets were exactly what I was looking for however the atmosphere was the complete opposite.
I downloaded the template and began by removing the sealife, as I wanted the setting to look more like the deep sea; less animal life was important. I reduced the strength of the god rays, bounced light, and darkened the color of the lighting and fog. I continued tweaking the colours and changed the angles and more. I wanted to change the size of the assets and locations of them more; however, because of my computer’s limitations, I couldn’t.
Unfortunately, because my creature used a couple of channels in Substance Painter that aren’t the default, the class tutorials didn’t work for me, so I had to do a lot of research and trial and error with settings. Although I was able to find a satisfactory setup for the texture nodes, I don’t think it looks as good as it does in Substance Painter. I also encountered many issues with the animation not transferring properly, and it even resulted in my mesh breaking, so I decided to leave it. I rendered a few different images, changing the brightness of the lighting and changing the camera lens and focus as the images were coming out blurry; however, I think that the blurriness is actually from a combination of the light and water physics. I tried to clean it up a bit in krita bringing back some sharpness and changing up the contrast of the background.
