The Promise and Reality of AI in Architecture

As AI dominates headlines with promises to revolutionize every industry, we at Webber + Studio took a more measured approach: testing these tools ourselves to separate hype from practical value. 

After a year of systematic exploration led by Managing Principal Thomas Tornbjerg, we’ve discovered that while AI won’t replace architects or transform design overnight, it offers specific advantages that are already enhancing our practice.

The Search for Better Visualization Tools

Our journey into AI began with a fundamental challenge: creating compelling visualizations quickly and cost-effectively during early design phases. 

Traditional architectural rendering is time-intensive and expensive, often requiring dedicated staff or outsourcing. Yet showing clients preliminary designs in basic SketchUp models can feel insufficient.

“A persistent problem in my 30 years of practice has been quick rendering,” Tornbjerg explains. “The software is expensive, there’s never enough time, and there are always issues with materials, lighting, or landscape. Getting good renderings has never been easy.”

Early Experiments

Our initial foray used SketchUp Diffusion, a free plugin that applies AI enhancement to basic SketchUp models. The results were intriguing, particularly in how it handled lighting and atmosphere. Consider this interior transformation:

While the AI-enhanced versions captured compelling lighting and materiality, the tool lacked precise control. It would often invent architectural elements unpredictably or misinterpret spaces, making it unsuitable for client presentations.

Finding the Right Tool

This led us to explore more sophisticated options, particularly Prome AI, which offers greater control over the rendering process. The breakthrough came in understanding how to prepare our models for AI enhancement:

In this example, you can see how a basic SketchUp model (top left) transforms when paired with a carefully chosen reference image (top right). The final rendering (bottom) demonstrates how AI can enhance materiality and environmental context while maintaining architectural accuracy.

Interestingly, the AI changed the shipping container branding from “MAERSK” and “P&O” to “MASPEK” – highlighting one of its key limitations. While the tool excels at visual transformation, it can struggle with precise text reproduction, reminding us that human oversight remains essential.

How It Actually Works

Through extensive testing, we’ve developed a refined process that maximizes AI’s strengths while mitigating its limitations:

Preparing the Model

The key insight was realizing that AI tools read contrast more effectively than linework. This led to a counterintuitive approach where we:

1. Create clean SketchUp models with minimal complexity
2. Remove traditional architectural lines
3. Use contrast to define spaces and elements

Look at this example:

The unusual red sky in the model isn’t for aesthetics – it helps the AI clearly distinguish interior and exterior spaces through contrast. This seemingly odd technique produces more reliable results than traditional architectural line drawings.

From Basic to Beautiful

The transformation can be dramatic:

These images show our complete workflow – from basic SketchUp model to reference image to final AI-enhanced rendering. Note how the AI tool maintains the architectural integrity while adding:

• Natural lighting and shadows
• Material depth and texture
• Landscape integration
• Atmospheric qualities

The final result achieves in hours what might take days with traditional rendering software.

Understanding the Limitations

While these results are impressive, it’s crucial to understand what AI can and cannot do. Here’s a successful example that also reveals the technology’s boundaries:

This rendering demonstrates AI’s strengths in handling:

• Complex glazing and reflections
• Integration with natural elements
• Lighting and atmosphere
• Material textures and finishes

However, achieving this result required careful preparation and understanding of the tool’s limitations. The AI didn’t design this solution – it enhanced our design through sophisticated visualization.

Looking Forward

As we integrate these tools into our practice, we’re seeing immediate benefits for both our firm and our clients. Once preliminary designs are complete, clients can now see photorealistic visualizations of their project within hours rather than waiting days for traditional renderings. This quick turnaround enables more dynamic collaboration – we can explore different materials, lighting conditions, or landscape options in near real-time during design meetings.

For clients, this means more informed decision-making earlier in the process. You can better understand how your future home or building will look and feel in its environment, how light will play through spaces at different times of day, or how various material choices might affect the overall aesthetic. This enhanced visualization capability not only saves time and money but also helps ensure the final built project aligns with your vision.

While AI won’t replace architectural expertise anytime soon, it’s becoming another valuable tool in our arsenal—one that helps us communicate our ideas more effectively and focus more time on what matters most: thoughtful, site-specific design that meets our clients’ needs.

Additional Resources

For architects and firms interested in exploring AI tools, here are some current platforms worth investigating:

Architectural Visualization

• Prome.ai – Advanced AI rendering with strong architectural focus
• Neolocus.ai – Specialized in architectural visualization
• ArchitectRender – Comprehensive rendering solution
• SketchUp Diffusion – Free plugin for basic AI enhancement
• Arko.ai – Architectural rendering with plugin integration
• ArchiVinci – AI-powered visualization platform

Planning and Development

• Maket.ai – AI-assisted architectural planning and development

3D Modeling

• Hyper3D.ai – AI-enhanced 3D model generation

General AI Art Tools

• Midjourney – While not architecture-specific, useful for conceptual exploration

Note: AI tools are evolving rapidly. We recommend thoroughly testing any platform before incorporating it into professional workflows.

This post draws from Thomas Tornbjerg’s presentation “Initial Forays into AI for Architecture” and reflects Webber + Studio’s ongoing exploration of emerging technologies in architectural practice.