ALIGN

Inspiration

There must be a point in time where we do not have any creative inspiration, or we simply have limited design knowledge. All those sophisticated design applications equipped with wide-ranging tools and brushes are challenging to use for noobs like us. The free design applications bombard us with an array of templates, but none to our liking. Hence, we brewed this idea of making an intent-based automated design tool. The design process essentially is composed of 5 different components: ask, imagine, plan, create, and improve. Our goal is to eliminate the creative thinking processes that consume the majority of the budget (cost and time).

What it does

ALign is an intent-based design tool that helps users to automate the process of design without requiring a creative appetite or any design knowledge.

(Demo of the application is in the link below)

How we built it

This application is built upon Interactive Genetic Algorithm that aims to iteratively progress towards an optimal solution. This algorithm is inspired by the Darwinian Evolution Theory, where only the "fittest" individuals survive and evolve over time. A Genetic Algorithm is an iterative process that relies on the user’s feedback. Hence, the optimization in this algorithm is to search for a design that caters to the user’s preferences and likings (which are the feedbacks).

In this application, the algorithm generates two components which are Procedural Art and Text Information. The Text Information is obtained from the preset of data that the users should input. The design attributes such as colour shades, sparsity, etc are parameterised with the genetic algorithm in order to optimise the output.

The final product is a web application using Next.js framework (based on React.js library), hosted at Vercel. The backend is a set of serverless functions on top of Next.js API routes. We also used Firebase Cloud Storage to upload/download generated PDFs and Firebase Cloud Firestore to store current design information used by the genetic algorithms.

Challenges we ran into

Programmatically generate an appealing design while adhering to design theory.
The embedding of the canvas have incompatibility with the Vercel’s (node-js based) serverless instance.

Accomplishments that we're proud of

Implementing Interactive Genetic Algorithms in various business use cases.

What we learned

Algorithms can be implemented in various use-cases and automate on complex issues

What's next for ALIGN

Diversified product portfolio (eg: banners, calendars, presentations, infographics)
Incorporating more constraints (eg: preset questions and more editing options) to generate a more narrowed result
Cater to a wider range of demographic
Incorporate more design attributes into the genetic algorithms to generate enumerate more design possibilities.