Could AI generate accurate CGIs for the property industry?
For artificial intelligence to succeed in producing computer-generated imagery (CGI) for the property industry, it needs to pass the ultimate test. The challenge is to combine the brain's left and right sides at the correct time. The left side of your brain deals with logical decisions, and the right deals with imagination and creative input.
Suppose there is a fixed, precise design in 2D, a first since the design process is usually a moving target. To convert 2D plans and elevations into a 3D model, you typically need to import the 2D drawings into software and then use the software's tools to extrude, loft, or otherwise manipulate the geometry to create a 3D model: SketchUp, Autodesk Revit, 3D Studio Max, Rhino, and ArchiCAD can do this but need the human input to interpret design information and clash detection. Clash detection has advanced but still needs the prompt to decide which surface, material or component takes priority in the design.
The issue might become more manageable for an existing structure where a 3d scan generates a point cloud of the building, and each room and service work within predetermined XYZ coordinates.
The trend towards modular construction makes the AI challenge easier; for example, a typical bathroom has modular sanitaryware, tiling and a lighting grid.
If AI can successfully build the 3d model of a room, it then needs to determine and apply the textures and finishes to prepare for the rendering. The finishes and materials must match product specifications, colours, textures, reflectance, the lighting design, and how the daylight or artificial light interacts with the space. However, advancements in texture and material recognition and machine learning algorithms can enable the AI to match product specifications, colours, and textures with greater accuracy and speed.
Then, the rendering process would begin. Several passes and 'auto correct' might result in a perfect design rendition, applying several moods to test the colour saturation and atmosphere and even placing people into the space.
It sounds like a mammoth task, but no more than in the late 1980s when realistic rendering was a massive challenge. Steve Jobs Pixar's solution to this challenge was to develop a new rendering technique called ray tracing, which simulates the behaviour of light as it interacts with objects in a 3D scene. Ray tracing involves tracing the path of light rays as they bounce off surfaces and interact with other objects, ultimately producing a highly realistic image.
However, ray tracing was computationally intensive and required massive processing power to render simple scenes. This presented a significant challenge for Pixar, as they needed to develop new hardware and software solutions to handle the demands of ray tracing.
Conclusion
AI could generate an accurate internal CGI of a new room on balance. The classic chicken and egg in the design process may mean even more design revisions, but these may be a touch of a button and substantially reduced labour costs in the redesign. Never underestimate the value of a good, creative designer worth their weight in gold.
Overall, while challenges are still to be overcome, the potential benefits of AI in the property world are significant. In the meantime, we will continue to produce the revised renders overnight and service our clients from our studios in London and China.