This week we’re comparing the three AI platforms we’ve tested over the past month. Each of the tools are strong in its own way, none of them are strong at exactly the same thing.
ArchMaster: one building, told consistently
For our latest test, we took a single office building through three spaces. An exterior 3D model screenshot, an interior, and a courtyard. Two different style prompts to see whether ArchMaster could make them all read as one consistent building.
Exterior (Before vs. After):
Courtyard (Before vs. After):
Interior (Before vs. After):
The interior rendering tool was the standout. Reference images work best when their perspective roughly matches your base models and when all your reference images share one consistent style. Mixing styles across references is where accuracy breaks down. A prompt layered on top of the reference image sharpens accuracy further, which is useful for holding a project’s visual identity across multiple spaces.
Reference Image:
Generated Image:
Gendo: one canvas, edited live
Gendo splits into two tools. Edit for geometry-level changes and Swap material for precise surface swaps. Both live on the same canvas.
Every version sits side by side for direct comparison. You can jump straight from tool to tool without starting over. Run a result through the Enhance tool, for instance, to quickly and easily push it closer to what you actually pictured.
Xfigura: one image, different directions
X figura’s node canvas turns a single image into a real exploration tool. Upload or generate an interior image, you can then use the image-to-image tool to achieve an AI interpretation of the exterior, or any other features of the same space. All from one starting point.
Other than that, video generation and 3D models also stood out to us where we were able to produce short animations through a space from various visual directions such as ‘panning’ or ‘rotating’. WorldLabs produced results strikingly close to real-world street level imagery and the platform’s ability to plug into different external ai models means that you can chase genuinely different visual styles without leaving the canvas.
So which one?
Consistency across a whole project: ArchMaster. Fast, side-by-side iteration on a single image: Gendo. Exploring one concept across many forms: Xfigura.
Have a great week, and as always, feel free to email with your thoughts or questions.
Best Regards,
Charie
AAL INNOVATION Team