Tomebound
Group Project
The Collaborative Group Client Project aims to teach students the basic principles of their chosen field of study. Also, students can enhance their critical thinking and specialism skills by meeting the needs of the end user.
​Introduction
Client: Rob Blofield
Project: Action/Adventure Puzzle Game called ‘Tomebound'
Story
The boss fight is over, the credits have rolled. After an arduous campaign to acquire the greatest magical tome ever made, two wizards successfully retrieve an enchanted book from the top floor of the tallest tower, only to find it instantly curses them…
One adventurer can only move forwards and backwards, and the other from side to side. It seems like this journey back to the inn is going to be a lengthy and challenging walk.
My Role in This Project
Why Did I Choose This Project?
I selected the 'Tomebound' project from various clients for several reasons. Firstly, it offered a specialised focus on specific technical skills, such as sculpting and animation in Blender, that I am keen on enhancing. Additionally, the substantial number of students in the 'Tomebound’ project adds a collaborative element, providing opportunities for diverse perspectives and knowledge exchange. This not only enriches the experience but also fosters a sense of shared accomplishment. These factors combined make 'Tomebound' the ideal project for me.
Character Design & Animation
In this project, I served as a 3D artist to acquire skills in character design and animation. With this intention, I volunteered to create an enemy character aimed at obstructing the progress of the players.
Reference
Firstly, I initiated the creation of concept art using Microsoft Bing Image Creator. By modifying the prompt, I generated a variety of AI concept arts. (All the images above were generated using Microsoft Bing Image Creator). I analysed which prompts were closer to the ideal design using Miro board to further improve the generated outcomes. Within the generation process, a character design featuring a tail that appears to be melting was discovered. Due to my specific preference for this element, I deemed reptiles to be a suitable choice for the design of the character.
Upon researching reptiles for these reasons, I uncovered the mythical creature known as a 'salamander.' While the term 'salamander' refers to an amphibian that exists, it also bears significance in Greek mythology. The mythical salamander is commonly portrayed with a form similar to that of a lizard and is frequently associated with fire, sometimes known as a 'fire lizard. Considering that our game features a fantasy theme with elements of lava, I deemed this concept to be appropriate. After obtaining approval from the client, I proceeded with the design generation modelled after reptiles.
 |  |  |
|---|---|---|
 |
Drawing from AI-generated illustrations, I sought feedback from both the client and peers. After meticulously selecting four artworks that garnered positive responses (Images above), I determined to create an enemy character with a three-stage form, using these chosen artworks as specific reference points. The appearance of the first form is modelled after lizards with spines on their back composed of flames, while the second form is inspired by axolotls with flame gills. Regarding the third form, based on feedback from the client, I chose to utilise the AI-generated image as its concept art. However, as the client suggested using only two forms in the game, we opted to incorporate the axolotl on multiple levels and introduce a third design for the boss character. This semester, my focus was on creating the character model for the second form, drawing inspiration from axolotls.
1st Form
2nd Form
3rd Form
Subsequently, I refined the second form through redrawing to establish a more intricate design. Included in the settings at that time are the following:
-
Gills composed of flames
-
Fangs
-
A glare
-
Melting through evolution.
-
A burnt/soot texture with lava glowing inside.
-
The characteristic of being defeated by the player using a bucket to pour water on the enemy.
I began collecting reference images of axolotls, aiming to gather various angles, including front, sides, and top views. While incorporating features such as external gills inspired by axolotls, I gathered reference images primarily focused on the leopard gecko. This was undertaken to incorporate more game-like attributes and bestow upon the character both an enemy-like appearance and a charming appeal that players would find endearing.
<330,086 Triangles>
To familiarise myself with sculpting in Blender, which was a tool relatively new to me at the time, I undertook the creation of the enemy 3D model as a practice exercise. At this stage, lacking a comprehensive understanding of re-topology and texturing methods, I finalised the model with a substantial number of polygons, employing solely the base colour in Blender for texturing. Also, I attempted the animation of the tail, but it did not meet expectations. Nevertheless, this experience contributed to my acquisition of fundamental knowledge in inserting keyframes.
At that time, I had believed that models, such as the head and body, must always have connected meshes. Consequently, when creating models, I spent more time than necessary, resulting in a prolonged production process. However, through inquiries with the lecturer, I came to understand that each part does not always have to have connected meshes and examples in games like 'Animal Crossing' that utilise separate objects as a single model. This newfound understanding significantly improved efficiency and enjoyment in the modelling process, fostering increased proactivity in addressing uncertainties by asking questions to the lecturer compared to previous projects. Also, based on the initial enemy model, I eagerly gathered feedback from peers and pinpointed areas for improvement, including an excessively frightening appearance and disproportionately large limbs.
Modelling
In the re-creation of the axolotl model, since this was my first attempt at modelling a creature and I initially planned to use the model as the final version without sculpting or re-meshing, I made a conscious effort to ensure the meshes were smoother than usual throughout the production process.
Although I initially began the modelling process by expanding faces (Images above), I encountered challenges in maintaining the proportion of the body. Consequently, I shifted to a blocking-out approach using rough shapes (Image on the left side).
While each part was still separated at this stage, here is the completed model. Initially intending to model chameleon-like hind legs, I made adjustments to reduce polygon count, ultimately referencing the hind legs of an axolotl.
I utilised the creation method demonstrated in class for the lizard character (Image on the left side), particularly refining the nose and mouth, and to attain the desired proportion, I drew inspiration from the character 'Salamander' in the movie 'Frozen.’
Next, I joined body parts to create a single object. At this stage, connecting the parts required them to have the same number of edges. It was challenging to align the edge counts using loop cuts and the knife tool. Following this, I proceeded with UV unwrapping and created a prototype of the material using Substance Painter. While I applied a scale material to the back and introduced a reddish tint underneath to convey the elevated internal temperature, upon importing the texture into Blender and rendering the result, I identified certain areas for improvement. These include observations that the colour appears excessively dark, and the scale patterns are relatively small and challenging to discern.
In the original plan, this model was intended for use as the final version. However, upon joining the parts, the flow of meshes became disrupted, leading to texture inconsistencies. As a result, a decision was made to reconstruct the model as a low-poly version, leading to subsequent engagement in sculpting and retopology. In sculpting, I increased the polygon count through remeshing to capture intricate details such as horns and fangs that were not fully expressed in the initial model, and based on this, I transitioned to creating the low-poly model by retopology.
​Retopology
 |  |
|---|---|
 |
Initially, I utilised the shrinkwrap modifier to form components like the eyes and nose, wrapping each mesh individually. Subsequently, I proceeded to model the occipital and jaw. When modelling the face, I aimed for a minimal polygon count while being conscious of detailing aspects like fangs and an angry expression.
The modelling of the joints in the legs was inspired by the modelling of human finger joints with careful attention to ensure smooth bending when adding animations.
While my initial plan was for the axolotl model to have hind legs that were dragged and did not resemble the legs of reptiles, I was dissatisfied with the outcome. Consequently, I revisited the high-poly model, reworked the legs, and improved their appearance through retopology.
<6,044 Triangles>
This is the finalised low-poly model. Compared to the initial model, the mesh has been cleaned up, and the polygon count has been reduced to 6044. Considering the client's specified limit of a maximum of 20,000 triangles, I believe the polygon count is satisfactory.
Texturing
After meticulous UV unwrapping, to minimise seam visibility, I imported the low-poly model into Substance Painter and proceeded with the baking process. I conducted a thorough examination of the textures by consistently importing them into Blender after each update to ensure accurate representation through base colour and normal maps. The aspect I particularly focused on was the eyes, as I consider them crucial for conveying the creature's emotions. Additionally, I enhanced the prominence of the eyes and mouth by darkening the edges.
Final Outcome
Bucket
As mentioned in the axolotl's setting, when the player pours water on the axolotl using a bucket, it will be defeated. Alongside the axolotl model, I also designed the bucket model. Using a similar design process to that of the axolotl, I initiated it by generating concept art using Microsoft Bing Image Creator. I curated 10 images from the generated AI art, focusing on distinctive elements like a wooden bucket, handle, and metallic rim. Subsequently, I referenced them as a basis for creating the low-poly model of the bucket.
​
This is the low-poly model. Among the reference images, I paid special attention to the metal fittings parts. To create a realistic appearance of being securely fastened with metal fittings, I took care to minimise gaps between each part. Additionally, I elevated the handles slightly above the metal fittings to prevent any awkward appearance in case animation is added to the handles in the future.
​
Following that, I sculpted exclusively on the body part to depict a wood pattern. To create the impression of jointed wood, I delineated seams between each of the corners.
Subsequently, I employed a decimate modifier to reduce the polygon count of the handle and the metal fittings. As a result, the polygon count of the handles became 452 triangles from 1348.
The model was imported into Substance Painter, and I baked the texture of the high-poly model, applying it to the low-poly model. In Substance Painter, I experimented with various functions to deepen my understanding of the application. For instance, I applied not only smart materials but also dust filters. Additionally, I used a water drop filter to heighten the realism of the model.
Here is the rendered animation in Blender. While I am satisfied with the model to a certain extent, there are some points for improvement.
Firstly, the metal part appears excessively thick, giving the impression of imparting a visual weightiness incongruent, and it fails to align with the desired wooden ambience. Secondly, it will suppress the overly clean impression of the metal part and give it a more weathered and worn appearance by applying the bevel modifier. Thirdly, while the game only utilises albedo, normal, and emission textures, the textures also include displacement and metallic textures.
Considering the client's specific colour palette as well, it is necessary to fix the metal part of the model and redo the textures.
Future Developments
While I completed the modelling and texturing of the axolotl enemy character this semester, I remain dissatisfied with the current texture and the game creation will continue in the following semester. To prepare for rigging and applying animations in the future, I created an animation to clarify the movement of the limbs, flames, and tail. Although the current animation has the same movement for both left and right limbs, I intend to refine and create more captivating and realistic game-like animations by studying references including the image provided by the clients and observing the movements of actual animals.
Reflection
Through this project, I had the opportunity to explore and utilise various applications such as Blender and Substance Painter. While it initially required significant time to become proficient in these tools, the development of these skills has significantly bolstered my confidence and passion for 3D modelling and texturing. Beyond honing technical skills, this project facilitated communication with both peers and clients, fostering a sense of responsibility for time management and motivation to seek and implement feedback. Despite encountering challenges, such as a lack of familiarity with certain applications, proactive inquiry and problem-solving have been instrumental in overcoming these obstacles. I recognise that facing such challenges is an integral part of my ongoing academic journey, and I am committed to continuous learning to enhance my skills and confidence further.