Sora vs Photoshop AI: A Real Test for Building Detection and Masking

The Experiment Setup

I decided to test two popular AI systems for image segmentation: Sora with Make Sense AI and Photoshop's built-in AI tools. My goal was simple - see which one performs better when detecting and masking buildings in low-quality images.

Test Image: A low-quality building photo from Aurecon Group's website

Challenge: The image had poor resolution, visible pixels, and blurry details - perfect for testing AI capabilities.

Round 1: Testing Sora with Make Sense AI

What I Did:

• Used Sora system combined with Make Sense Alpha

• Applied mask segmentation to identify building structures

• Analyzed the results for accuracy and detail

Sora's Performance:

• Good: Successfully identified major building shapes

• Good: Separated sky from buildings reasonably well

• Excellent: Automatically detected small details like windows, tables, and furniture

• Excellent: Found tiny objects that I hadn't even noticed in the original image

• Limitation: Struggled with individual building separation

• Limitation: Created some rough edges around complex structures

Overall Sora Result: Decent performance for general object detection, but lacked precision for architectural elements.

Round 2: Testing Photoshop's AI Tools

What I Used:

• Object Selection Tool with Select Details feature

• Cloud processing enabled for better accuracy

• Remove Background tool for comparison

Photoshop's Performance:

• Excellent: Incredibly precise building edge detection

• Excellent: Successfully separated individual buildings

• Excellent: Recognized architectural details like windows and facades

• Excellent: Clean, professional-quality masks

• Excellent: Handled complex overlapping structures perfectly

Overall Photoshop Result: Superior accuracy and professional-level precision.

Head-to-Head Comparison

Building Edge Detection

• Sora: Acceptable but sometimes rough

• Photoshop: Extremely precise, even with poor image quality

• Winner: Photoshop

Individual Building Separation

• Sora: Struggled to distinguish between adjacent buildings

• Photoshop: Clearly separated each building structure

• Winner: Photoshop

Architectural Detail Recognition

• Sora: Missed fine details like windows and building features

• Photoshop: Recognized and preserved architectural elements

• Winner: Photoshop

Small Object Detection

• Sora: Excellent at automatically finding tiny details like windows, furniture, and small objects

• Photoshop: Requires manual selection for very small details, but gives precise control

• Winner: Sora (for automatic detection), Photoshop (for precision control)

Mask Quality

• Sora: Functional but required manual cleanup

• Photoshop: Professional-quality masks ready for immediate use

• Winner: Photoshop

Processing Speed

• Sora: Moderate processing time

• Photoshop: Very fast, especially with cloud processing

• Winner: Photoshop

Why Each System Has Its Strengths

Sora's Advantages: Sora excels at automatic detection of small objects and details. It can find windows, furniture, small decorative elements, and even tiny objects that human eyes might miss. This makes it excellent for comprehensive scene analysis without manual intervention.

Photoshop's Advantages: While Photoshop may require manual selection for very small details, it provides superior control and precision. Once you select an area, the masking quality is professional-grade and the edge detection is extremely accurate.

The Trade-off:

• Sora: Better for automatic comprehensive detection

• Photoshop: Better for precise, controllable results

Why Photoshop Still Wins for Professional Work

Specialized Training: Photoshop's AI has been specifically trained on millions of architectural images, giving it superior understanding of building structures.

Advanced Algorithms: Adobe's Object Selection tool uses sophisticated edge detection algorithms designed for precise professional work.

Cloud Processing Power: When enabled, Photoshop uses powerful cloud-based AI that can handle complex architectural details better.

Continuous Updates: Adobe regularly improves their AI models based on real-world professional usage.

The Surprising Discovery

What I Expected: Both systems to perform similarly on this basic task.

What I Found: Each system has distinct strengths. Sora impressed me with its ability to automatically detect tiny details I hadn't even noticed - windows, small furniture, and decorative elements were identified without any manual input. However, Photoshop's precision for building edges and overall mask quality was significantly superior.

The Reality: It's not just about one being "better" - it's about different approaches to AI-powered image analysis.

Practical Implications

For Architects and Designers: Photoshop's superior building detection makes it the better choice for architectural visualization work.

For General Users: While Sora is excellent for many AI tasks, Photoshop remains the gold standard for precise image segmentation.

For Professional Work: The quality difference is significant enough to impact final results in professional projects.

Conclusion

This test revealed that specialized AI tools often outperform general-purpose systems in their specific domains. While Sora with Make Sense AI is impressive for general object detection, Photoshop's AI excels specifically at image editing tasks like building detection and masking.

The Takeaway: Choose your AI tools based on your specific needs. For precise image editing and architectural work, Photoshop's specialized AI currently leads the field.

My Recommendation:

• Use Sora when you need comprehensive automatic detection of all image elements, including tiny details you might miss

• Use Photoshop when you need professional-grade precision and control over your final results

• For architectural work specifically, Photoshop's superior building detection and masking quality make it the better choice despite requiring more manual work for small details

Photoshop Masking

Sora Masking