An AI Experiment
In this experiment I will be using AI at different parts in the design process. I wish to create a living room space where the ceiling curves gently down towards the center of the room. I will be using AI and Rhino/GH to try an achieve this deisgn aim. But what I hope to find out is when is it best to introduce AI.
The two images below are of my attempt to capture the room shape I am interested in trying to create. The first image is of the Rhino/GH model and then I used Gemini AI to create a realistic verson of it.
Following this first attempt, I then experimented with AI to see if given a simple sketch what sort of shape and patterns it will create for a room like the one I was after it could create. See below the prompt and the result from Gemini AI.
This second set of images is where I used AI at the start. The design concept was created by Gemini AI then I used Rhino/GH to create a model of it.
My conclusion from this experiment is that when there is a constraint on the AI, such as in this case the ceiling should curve down towards the middle and then up at the sides. The AI did not come up with anything a lot more complicated than what I was able to generate in Rhino/GH.
Below are images of the two models, when starting first in Rhino GH, I was able to build a larger model because the idea I had worked throughout the entire building.
When starting with Gemini AI I only built one room because the AI only provided the idea for the single room. So the rest of the model is not built up. The ideas from AI need a lot more input from the humans to generate an entire building model.
https://skfb.ly/pNnnV – Link to Rhino GH to Gemini – Model
https://skfb.ly/pNn6n – Link to Gemini to Rhino GH – Model
Please like and subscribe. I will be doing a next blog post on a process entirely by AI, using AI to generate the brief, concept, model and final images.
Thanks, Sim
SJM Studio Leeds 