AI-assisted parametric design

AI-assisted sightline design and analysis

The reactions to AI within architectural design vary widely. Some resist it, concerned that automatically generated content is difficult to control. Others embrace it too enthusiastically, relying heavily on AI to produce material that may ultimately compromise design quality. Both positions are understandable—yet both overlook an essential truth: when used thoughtfully, AI is not a replacement for professional judgement, but a remarkably capable assistant that can elevate both quality and efficiency.

Over the past months, I have been experimenting with ways to integrate AI meaningfully into the theatre-design workflow. The video linked here shows recent progress in automating and visualising sightline design, one of the most time-consuming tasks in performance-space planning.

Sightlines are not just a visual metric; they are deeply connected to acoustics. A seat with an unobstructed view often enjoys an unobstructed direct sound path. Improving one tends to improve the other. By accelerating sightline studies, we simultaneously open the door to faster and richer acoustic evaluation.

To support this, ChatGPT has been used to generate custom Python components that automate several key elements:

• constructing a seating rake using the C-value sightline method.

• dynamically placing seats along this rake,

• analysing stage visibility and local obstructions, and

• generating seat-by-seat viewpoint images.

Automatically generates a seating rake based on the C-value method and positions the seats accurately along it with given seating spacing

Analyses stage visibility and identifies elements that obstruct the view. In this example: green indicates no blockage, yellow represents 5% blockage, and orange indicates 10%.

Automatically takes viewpoint photographs at the selected seating locations, providing a clear visual understanding of sightlines.

These tools keep the designer firmly in control, while dramatically reducing the effort and repetition usually required. AI does not replace the logic of the method—it simply makes the coding and iteration far more accessible.

To enhance communication with clients and collaborators, I explored AI-based visualisation tools as well. Nano Banana was used to transform an abstract viewpoint presentation into a realistic image. This enables early-stage 3D model studies to remain computationally light while still producing visuals that express sophisticated spatial intent. Abstract geometry can be replaced with realistic elements effortlessly, making design experimentation fluid and immediate.

Nano Banana transforms the abstract presentation into a realistic image, while preserving the exact positions of the audience and stage as defined in the original model.

Sora was then used to convert a still image into a short video sequence—an engaging way to convey atmosphere and design intent, even if the current example shows that the technology is still maturing. Its potential, however, is unmistakable.

I am always interested in how others are incorporating AI into their work. If you would like to discuss how these methods might support your own projects—or if you would like to make use of my parametric and acoustic design expertise—please feel free to contact me.

---

Some of our recent projects include:

• Several theatre projects

• An office fit-out at Maxim Office Park

• A recording studio in Leith Edinburgh

• A dance studio in Edinburgh

• A multi-story residential building in Edinburgh

• A few houses in Scotland

• Dust extraction fan noise assessment in Musselburgh

• Noise impact assessment for a model flying park in Scotland

• Sound insulation testing for some apartments in Scotland