3D-rendering of a modern office plaza. Focus on the balance between realism and extreme low-poly, for the purposes of game optimization
This project is based on an AI-generated image of a plaza like this, provided by the client. I had to take this image and create a 3D-setting from it, while adapting any logical errors from the generation process. The ramp leading down to the lower level is of my own design as this angle was not covered by the AI-image, and adheres to building requirements for disabled users; This would be a requirement should this plaza be built in real life, and I wanted to add it as an example of inclusive design.
This is a situation where AI has a purpose; The client felt they could provide a visual and concrete example for this project, rather than relying on collecting reference images of other architectural projects to communicate their desires. This however increased workload in the model-building process, as I constantly had to find solutions for mathematically impossible architecture while keeping it visually identical to the reference. AI also understands many of the elements required to be on a building such as this, but without any clear understanding of why these elements need to exist or how they are used - much of the work became thinking about how people move and use such a plaza, and adapting the AI-results to this. It must also be noted that this worked as well as it did because this is not a real building (despite being based on dimensions from real skyscrapers such as the Willis Tower in Chicago), and I could freely adapt the size and shape of the plaza without any restriction on the shape of the building or property. In an actual architectural project, I would have to make larger changes in adapting the AI-image result with the requirements of the building itself, turning the AI-image into a reference for the general vibe and appearance more so than an actual architectural guideline.
This means that implementation of AI into this process fits best in the early planning phases of an architectural project, and fills the same role as other reference pictures. The advantages AI grants compared with collecting online reference images is that the client could more clearly show what they wanted, and felt more ownership of the concept. I leave it to the reader whether these advantages are worth it, as my goal here is to objectively present my experience working with AI during this project