Blog

In the realm of 3D scanning, photogrammetry stands out as a transformative method, offering a common and accessible approach to reconstructing three-dimensional environments without the need for additional hardware. Unlike methods such as LiDAR or depth sensors, photogrammetry harnesses the power of images to map out intricate 3D models.

Traditionally, users have relied on cloud computing to process the extensive data required for accurate 3D scans. However, a paradigm shift is underway. With advancements in mobile technology, the emergence of real-time photogrammetry on mobile devices is rewriting the rules, enabling users to perform 3D scanning right on their devices without the necessity of external computing power, and receive instant results. In this blog, we will delve into the nuances of photogrammetry, highlighting the distinctive realm of real-time on-device computing and shedding light on its advantages, limitations, and the groundbreaking possibilities it presents.

When and why to choose on-device computing? 

On-device computing in photogrammetry offers distinct advantages over traditional cloud computing, with considerations varying based on different use cases. Let’s explore these computing methods across five key aspects:

1. Instant 3D Results:

/ On-Device Computing: The most significant advantage is the real-time processing capability, allowing users to instantly visualize reconstructed models on their devices.

/ Cloud Computing: In contrast, cloud processing involves uploading images and waiting for remote data processing, introducing delays, especially with large datasets.

2. Privacy and Security:

/ On-Device Computing: Data processing occurs locally, enhancing privacy and security. This is crucial for applications dealing with sensitive data, such as human body scans.

/ Cloud Computing: While cloud service providers implement robust security measures, uploading data raises privacy considerations for some users.

3. Offline Capability:

/ On-Device Computing: Users can perform 3D scanning even without internet connectivity, ideal for fieldwork, outdoor activities, or situations with unreliable internet access.

/ Cloud Computing: Dependence on the cloud poses challenges in scenarios without internet access, limiting accessibility and flexibility.

4. Reduced Latency:

/ On-Device Computing: Local processing reduces latency, providing a more responsive user experience crucial for applications requiring quick decision-making or model interaction.

/ Cloud Computing: Uploading, processing, and receiving results introduces latency, which may be acceptable for some applications but lacks the immediacy of on-device processing.

5. Cost Efficiency:

/ On-Device Computing: Costs associated with cloud services can be avoided, as on-device processing eliminates the need for extensive server resources.

/ Cloud Computing: Costs for cloud computing can accumulate over time, making it a significant consideration based on the volume and frequency of 3D scanning.

Understanding these benefits allows users to make informed decisions aligned with their specific needs, considering factors such as speed, privacy, accessibility, and cost efficiency.

Drawbacks: A trade-off between speed and quality

While on-device computing in photogrammetry offers remarkable advantages, it’s essential to acknowledge a potential drawback – the inherent trade-off between processing speed and output quality. Here are key considerations to bear in mind:

/ Limited Computational Resources: Mobile devices, despite rapid advancements, may still lag in processing power compared to cloud systems. Finite computational resources on mobile devices may restrict the complexity of scenes or the number of images processed in real-time, affecting the level of detail in reconstructed models.

/ Sensor Limitations: The quality of scans heavily relies on mobile device sensors, such as cameras, which may have limitations in capturing data compared to specialized 3D scanning devices. This can influence the overall quality and fidelity of reconstructed models.

/ Data Storage and Management: Real-time processing generates substantial data. Mobile devices may face challenges in efficiently storing and managing large datasets, potentially leading to limitations on the size and complexity of 3D models that can be processed in real-time. 

/ Environmental Conditions: A common limitation in photogrammetry is sensitivity to environmental conditions, including lighting variations and the presence of reflective surfaces. These factors may impact the accuracy and reliability of the scanning process, especially in challenging or dynamic environments.

On-device processing excels in delivering instant results, enabling users to rapidly visualize 3D scans. However, this speed may come at the expense of achieving the highest level of precision and detail in the reconstructed models. On the other hand, cloud computing, with its robust computational capabilities, can deliver exceptionally high-quality outputs but often requires more time due to the data transfer and remote processing involved. Users must carefully consider their priorities: whether swift, on-the-go results are paramount or if sacrificing some speed for top-tier precision and detail is acceptable for their specific applications. This trade-off represents a key consideration in the decision-making process when selecting the most suitable computing method for photogrammetric endeavors.

How to determine the scale of a model reconstructed using photogrammetry?

Photogrammetry, being rooted in its purely image-based approach, offers accessible 3D scanning but may present challenges in terms of scaling when compared to other sensor-based methods. Scaling, in simpler terms, refers to determining the real-world size of the captured model. While photogrammetry consistently ensures the creation of an accurate 3D model, it’s essential to note that the scale of the model cannot be derived from images captured by a single camera alone. In practical terms, this means that any absolute distance measurements taken from the model lack real-world significance without a reference to the actual size. To determine the scale of a model reconstructed through photogrammetry, various methods can be employed:

/ LiDAR (Light Detection and Ranging): LiDAR technology measures distances by emitting laser pulses and calculating the time it takes for the pulse to return. Mobile devices equipped with LiDAR sensors, such as iPhone Pro Series, can provide highly accurate depth measurements, allowing for precise scaling of the model.

/ Stereo Camera: Stereo vision involves using two cameras to capture an image from slightly different perspectives, simulating the way human vision works. By analyzing the disparities between corresponding points in the images, the depth and scale of the scene can be calculated.

/ Reference Object: Placing a known reference object of a known size within the scanning area can help establish the scale of the model. By measuring the reference object’s dimensions and comparing them to the scanned model, you can accurately determine the scale.

/ IMU Sensors：The IMU (Inertial Measurement Unit) sensors, such as accelerometers, on the mobile device play a role in estimating the distance the camera has moved between capturing images, potentially aiding in scale estimation. The main drawback is that it can be arbitrarily bad, depending on the phone’s sensors and especially the actual motion the user engages in during capturing. Due to the potential variability associated with these factors, relying on IMU sensors is often avoided as a primary source for scaling.

Each of these methods has its advantages and considerations, and the choice depends on the specific application, available hardware, and desired level of accuracy. Understanding these scaling methods allows users to tailor their approach based on the unique requirements of their photogrammetric endeavors.

Summary

In conclusion, the choice of on-device computing in mobile real-time photogrammetry offers immediacy and reduces privacy concerns, but not without the trade-off between speed and quality. Navigating the complexities of determining scale highlights various methods with unique considerations. 

Want to try out this cool tech? Scroll down and book a demo now for more information. We also have a great news: Astriscan, our first public 3D scanning app for real-time mobile photogrammetry, will be released soon for iOS. Experience the fusion of innovation and accessibility by signing up for the Astriscan release today. Be the first to know and test it!