Technology Approach and Roadmap
Over the next 5 years we are going to have to open up our thinking in completely new ways in order to move humanity forward.
monavaile is driven to be part of the digital revolution and enable a native digital future where digital economies of scale are able to serve new possibilities and unlock new opportunities for humanity.
Never in the history of mankind have we had so many advancements emerge in parallel and then collide to enable such huge innovation and create previously unfathomable opportunity. The progression from Atari in-game graphics, like Pong in the 1970s, that displayed a simple pixelated black and white 2D animation, to the hyper detailed and dynamic graphic 3D virtual worlds that we see today is simply breathtaking. The question that seems to get a lot of people thinking is what happens when these worlds become so real and so advanced that we can no longer tell the difference?
If living in a 3D virtual or digital world could enable us to experience things differently, experience things better, and allow us to overcome so many of the physical limitations that we currently face, then wouldn't that be extremely powerful and beneficially impactful for our long term future as a society? The ability that these virtual worlds would allow, in terms of being able to separate our physical location from where we are present is transformative - we could vicariously sample our dreams, give more power to our imagination, and create experiences that perhaps couldn't have been possible previously.
These new digital economies have the potential to unlock trillions of dollars worth of new value. However, in order for this to occur, we need to be able to enable seamless immersive digital experiences that are highly customizable, engaging and enable new dynamic ways for people to self-express. These concepts of individualism, identity and self-expression are rooted as some of our most core values as humans, and ultimately, form our human nature. While the Internet has given us the ability to connect with others and re-create parts of our identities, the impersonalization of the screen-to-screen interaction doesn't hold the same level of impact that a true digital world interaction would enable, in terms of truly unlocking communication, collaboration and self-expression in the information age , enriching humanity and people's experiences with the world around them.
Fashion brings with it an emotional experience that is highly personal to the wearer and integral to our identity and self-expression. Being able to translate this same experience in a digital environment with digital clothing is absolutely vital to being able to scale online worlds that can create meaningful memories, emotional experiences and new connections and relationships for the user. Our identities are moving targets that never truly reach an endpoint - and creating these emotional experiences through our clothing unlocks inspiration for the wearer to be able to feel good, express their personality, be authentic to who they are or maybe try out a completely new identity that makes them stand out and feel powerful.
Building on the recent advancements in AI and cloth simulation, we are looking at tech around optimising 3D models of garments across multiple vertices, resulting in visually realistic simulation, that is accurate to the material properties, and allows for greater interoperability when it comes to being able to take these designs for use on different avatar geometries and sizes in different virtual environments.
Our goal is to develop and maintain an exclusive digital-only inventory across garment geometry, materials, and textures, which can then be selected and seamlessly plugged into various games and virtual environments.
Compatibility of digital assets, especially cloth, across different game engines and physics simulators is a challenging problem - and to this day, no standard protocol exists that can guarantee any interoperability. In part, this problem exists because there has not been a major push to standardize cloth simulation. However, this is only part of the story. Cloth modeling is an intricate and labour intensive process. From designing a cloth to then bringing it to life in a virtual environment requires a series of complex steps that demand special attention and hours of labour intensive manual work. One must carefully consider animation, avatar shape and geometry, interaction with the environment, lighting conditions, whilst also adhering to enforced computational constraints. As such, building tools and systems that can help streamline this process will bring about significant industry transformation, and something that we aim to solve with our initiative.
We are proposing a model similar to automotive manufacturing. The production of mass manufactured vehicles is highly automated but has limited customization options. On the other hand, boutique vehicles such as a Lamborghini or Bentley can offer extreme hand-crafted customizations for higher prices.
To scale up digital cloth manufacturing, we must first simplify the process drastically so that it can then be highly automated using AI and Machine Learning tools. As such, the primary focus should be on modularity where each module performs a very specific task and interconnecting such modules completes the pipeline. Depending on the customization options, a cloth design may require any different number of modules to manufacture. Automating across as many modules as possible will eventually result in a full production line toolkit for interoperable and realistically simulated and visually appealing cloth for both pre-compute and simulation stages. Digital fashion designers then have a streamlined, interchangeable and highly efficient channel for taking their realistic designs directly into these environments and enabling direct collaboration with game developers.
We are looking at optimisation around some of the most important aspects when it comes to translating clothing into a real time game environment, from establishing stronger hierarchies for colliding meshes when it comes to mixing and matching different clothing combinations to easier customisation for different body proportions.
We believe in enabling a future of limitless design possibilities for creating and building more realistic metaverses where brands and designers can distribute their fashion to an entirely new customer base of players and interactors through new business models like merchandising for digitally-in-game performed concerts or shows. Amongst this, garments can be bought and sold into different outfits that are fully interoperable and real time motion responsive across different virtual environments.
Our longer term technological roadmap is heavily geared towards taking into consideration the recent advancements around cloud infrastructure and the early support by major game engines for infrastructure like Stadia and GeForce Now. It is now possible to run deep learning models on specialized hardware such as GPUs and TPUs and so improve the fidelity of cloth simulation using post-processing techniques such as style transfer GANs and image and video restoration for fixing bugs such as overshading, clipping etc. We see the potential of transferability for some of the approaches and processes used in deep fakes. Taking certain techniques from ML models like autoencoders could provide a unique use case for both cloth visualization and simulation enhancement, especially due to the fact that cloth rendering must be performed on a per player instance, as the perspective of each player differs significantly, and therefore leveraging the power of AI assisted post-processing techniques to our pipeline can help to add additional fidelity as well as fix visible bugs built on in the pre-processing and simulation stages. Materials such as wool and glints are notoriously difficult to simulate - style transfer GANs may be used to add such details. Similarly, it is possible to use various image restoration techniques to fix rendering bugs such as overshading and clipping, and additional details such as cloth creases and wrinkles can be added based on the draw distance as part of the post processing step.
We are also tailoring our project's focus and approach, at least initially, towards multiplayer and metaverse type games, where greater emphasis is placed on the creation and customizability of different identities, and so therefore more emphasis on realistic cloth designs and the ability to give player control when it comes to mixing and matching of outfits. This also feeds into our broader vision of properly enabling Designer X Developer collaboration. We see huge fundamental and scalable value in properly establishing these networks, which will act to enable both the fashion industry in exploring new technological use cases and also the gaming developer community in being able to fluently translate fashion creativity into the virtual worlds of tomorrow.
There is truly amazing work being done, and we are at a time where there is no reason why anyone can't try out any and all ideas.
The technology opportunity is enormous and will unlock a new wave of creativity for enabling true digital self expression in these future worlds and digital economies.
See what monavaile is enabling right now across the digital fashion industry with our latest NFT focused initiative: DIGITALAX.
Copyright © 2020 monavaile