DVB303 Project | 2D Model Assistant Alyssa Berdar | DVB303



A virtual assistant that is able to connect with consumers in an entertaining way through streaming, promotional videos, and live interactions. This assistant will aesthetically compliment the theme of EBGames, taking inspiration from their already existing mascot, Buck the bunny. This new model will however, be more human-like in order to be able to form a connection with their audience, with future possible implementation into augmented reality.


My project will consist of using technologies such as Live2D and Adobe Character Animator to bring my interpretation of EBGames' rabbit mascot to life. This mascot will be used to increase consumer contact with the brand through the form of informative entertainment, communicating with predominantly an audience who are interested in video games via streaming and videos to promote upcoming releases and review games. I aim to have this model be used to develop informative videos of the company's products that will cater towards those who are unable to visit the stores in person due to disabilities or poor health. This model will be made with the notion of being multipurpose, for example, it has the ability to be recorded for pre-recorded promotional videos, whilst also having the capabilities to perform live.


Inspiration Context | A lot of my inspiration for the model comes from rabbits and carrots. I wanted the design to incorporate a blend of these two aspects, all whilst keeping an humanoid assistant-like appearance.



The goal of this project is to create or further develop a virtual mascot for the gaming retailer EBGames to commemorate the company's 45th anniversary and to increase consumer interest and contact with the brand. The model will utilise emerging technologies such as face-tracking, with the future consideration of being implemented into augmented reality as a digital assistant that answers online shopping queries. As the model is able to be recorded, this will allow for implementation into multiple areas of the company, such as promotional videos and streaming reviews for products.

Key Objectives

  1. Design a character model that aesthetically employs the colour scheme and characteristics of EBgames (Rabbits, carrot points, gaming, blue/red).
  2. Experiment in Live2D/Adobe Character animator, choose which program works best for the expressiveness of the model.
  3. A model that is fully rigged will be developed for the purpose of face-tracking softwares available on both Android and IOS devices.
  4. Consider further implementation into AR for more consumer interaction with the model. (AR assistant that uses auditory and visuals



The deliverables of my project will be promotional material of the mascot which may consist of splash artwork and expression poses, the fully rigged model, and a showcase of the model's capabilities.

  1. Splash Art / Promotional mock up video
  2. Model logo
  3. Concept artwork of model
  4. Rigged model
  5. Instagram Page (for the splash art and logo) / Model Showcase

Timeline and Key Milestones

Major Constraints

In relation to my original pitch of including a physical augmented reality aspect through the use of NFC tags, I concluded that the workload would be unlikely to be achieved within the current time allocated due to external factors such as personal health and commitment to other units. I realised that designing and rigging a model would take up the majority of my time, and I decided that the AR aspect could always be implemented later if I were to pick this project up again. Therefore, I opted to focus primarily on designing the mascot so it is fully rigged for Live2D or Adobe Character Animator and takes advantage of face-tracking technology these programs provide. Most of my constraint with this project is time related, thus the model will have to be simple in order for the rigging process to go smoothly.

Annotated Process Documentation

Phase Zero | Concept Art and Model Personality

Promotional Context | To help advertise the release of their new humanoid mascot, splash and promotional art are made to exhibit the personality of the character. These artworks help promote the mascot and inform the audience of its purpose as a virtual assistant.

Decision Points | During designing the mascot, I decided to take heavy inspiration from EBGames' current rabbit mascot, and their reward system of carrots. These two main features help make the design recognisable when paired with the colour scheme of the brand. With the carrot UI interface on her promotional art, I wanted the leaves to resemble bunny ears to help keep the design aesthetic consistent. Much like the uniform of EBGames, I decided to keep the blue top, black pants combination to convey her as a digital assistant.

Considerations | Due to the time allocated, I wanted to spend less time on this promotional artwork and more time on developing the model for rigging and putting together a showcase. This resulted in a very simple background, but it ended up working well with the artwork of the model as it was not too complex itself.

Dependencies | Developing the promotional artwork for the model allowed for me to come to final character design decisions that will appear on the trackable model. This artwork also acts as an advertisement for the model itself, exhibiting the joyful nature of the character.

Model Logo

Context | For social profiles, a logo was developed to compliment the model and pay homage to it's origins and aesthetic; EBGames. This will help convey the brand identity as the logo is similar to that of EBGames' mascot.

Dependencies | For the Instagram page, I wanted it to have a profile logo that made it recognisable. I took inspiration from the EBGames pre-existing logo, and added a carrot clip to it to pay homage to the model.

Model Phase One | Drawing

Context | After developing the concept artwork, I began sketching out the model. To be time considerate, I needed the model to be as symmetrical as possible to ensure that I could complete rigging within a reasonable timeframe. Phase one of the model is a relatively quick clean sketch of the model with base colouring to test out how she will be put together.

Decision Points | As the first model sketch looked a little broken, I kept the design and refined it more in the final concept. Referring to my inspiration board, I made the bunny ears look a little more interesting by giving them a curve instead of them being straight.

Considerations | Some things I had to consider were the layer order, especially because when importing the file to Photoshop and then Live2D, the order can either make the rigging process more difficult or easier. I kept everything in order by separating them into labelled folders.

Dependencies | This separated model artwork is required as Adobe Character Animator and Live2D both use 2D resources when it comes to rigging.

Alternatives | The software used to develop this model was Clip Studio Paint, which allows for files to be exported into PhotoShop documents. I considered using Photoshop, however during the creation of the model it felt a bit clunky with my style, and I did not feel confident enough to develop all 80+ layers in Photoshop in case it crashed.

Model Phase Two | Final Separated Art

Layer separation, final model outcome, and model parts

Context | Here I begin separating the layers so there is a left and right side to the model, this will allow for more dynamic rigging. For example, when the character turns to the right, there will be some 3 dimensional aspect to the 2D character model, giving it more life.

Alternatives | When researching how to develop my character model, there were some methods that painted the character all on one single layer and then cut out the pieces for rigging after the drawing was complete, and some methods that went through each layer one by one to avoid having to separate and fill in empty spaces. I chose to do the latter as it seemed more time efficient for this project.

Challenges and Solutions | One of the predominant issues I ran into whilst rigging was layer order. I noticed that the back of the hair went in front of the back of her collar, which was solved by quickly changing the layer order to bring it behind her collar. I did not have an issue with any other segment as it was only this minor layer order issue that I had.

Model Phase Three | Rigging for Face-tracking

Head Rigging

Body Rigging

Rigging Context | Using the 'glue' tool Live2D provides, I glue the joints of the body parts together to help create consistent movement. Here I go through each art mesh to create these joints to make the body rigging process easier.

Considerations | Gluing parts together is a struggle in itself, so I have to consider the art mesh as a whole rather than as two separate pieces.

Dependencies | The movement of the body of the model is necessary to making the model feel more lively, and contributes to the overall movement of the model. This is important as when it comes to face tracking, we do not just want only the head to move, otherwise it will look odd when the whole model is done.

Alternatives | Adobe Character Animator was considered for rigging the model, although through researching the software, it was evident that it works better with models that are solely 2D. It did not allow for me to manipulate individual meshes and create a 3D feel with 2D assets like Live2D permits.

User Feedback | I had my mutual, Jennifer Ngo, test out the face tracking part of the model every now and then and provide feedback on what to improve on. The key things I took away from her feedback were to make the head and hair meshes curve more to provide a more 3D feel, as well as make her eyes a little more expressive. This is evident in the final model showcase.

Model Showcase | Adobe After Effects and Premiere

Splash Animation for Final Behance Video

Video Context | Through using After Effects, I was able to develop a quick introduction using the splash art I made at the beginning of the project.

Tools and Resources | The splash art animation was edited in Adobe After Effects, and then imported into Adobe Premiere where the final showcase was being developed.

Decision Points | Rather than having the full artwork move onto the screen, I decided to cut up the pieces and animate them separately in After Effects.

Dependencies | The splash art helps add more to the final showcase video of the model, helping the audience understand the development from sketch to model creation.

Challenges and Solutions | The main challenge I faced was mostly evident in animating the splash art for the final showcase video, during this process I struggled to position the elements of the artwork properly and it took a while for me to find what worked well. The solution was mostly trial and error until the segments all lined up correctly.

User Feedback | I consulted in mutuals undertaking film courses at Griffith to provide pointers in how to go about making the splash art come to life. They responded by suggesting to make the assistant's bunny holograms rotate and change positions to make it seem as if they are 'emerging'.

Instagram Profile for EBuni Model

Instagram Context | To help advertise this model and its purpose, it will be marketed on Instagram to garner an audience.

Considerations | After creating the model, I wondered where I would exhibit my project. I thought social media would work well as it will allow for the model to connect with the target audience.

Alternatives | An alternative I considered was Twitter, but it lacked the layout that I wanted for my multiple artworks, I also really like carousel posts, so Instagram was the obvious choice.

Behance Portfolio for Final video Showcase

Future Considerations | Possibility of Being Incorporated as an AR Assistant

Model Iterations

For the model itself, there were a lot of improvements that could have been made if there was more time allocated to the rigging process. I believe the virtual assistant, EBuni, is a little bit more stiff in her movement than I would like her to be. Future iterations of the model could consist of more movement and bouncy hair physics to compensate, as well as more emphasised turning when she moves from side to side.

Implementation into Augmented Reality

As the model is ready to be implemented into multiple technologies such as virtual and augmented reality, I believe this could be a future consideration to help bring the character model into real life scenarios. However, as this is a model that uses face tracking, I do not think the technology is up to the standard of being able to handle a live 2D model well enough for it to keep its form.