To figure out which objects can be reconstructed using photogrammetry, you need to understand its limitations. Photogrammetry works by comparing and connecting images based on common features. Therefore, objects that don't look the same in different images won't be reconstructed accurately. Here are some key factors to consider:
Object Texture and Detail: Objects with diverse textures and clear, distinguishable features are ideal for photogrammetry. Smooth or reflective surfaces can cause issues as they may not provide enough unique points to be identified. For example, colorful and matt objects typically give significantly better results than glass objects, which may not be reconstructed at all. Use a scanning spray and add different color elements to your objects if possible.
Movement: The object and the environment it is located in should remain static during the capture process. Any movement can result in misalignment and errors in the reconstructed model.
Lighting Conditions: Shadows, glare, or changing light conditions can distort the appearance of the object features, making it difficult to match common points accurately. Even without consistent lighting, you may end up with an aligned and reconstructed model, but it is still highly recommended to have good lighting conditions. The best time for photogrammetry is on an overcast day.
Shape Complexity: An object's shape is important, and those with many complex details and hidden parts may be challenging to reconstruct. Capturing from multiple angles to cover all parts of the object is crucial.
Number of Images: A sufficient number of overlapping images from different angles is required to create a detailed model. Insufficient coverage can lead to gaps or incomplete reconstructions. You can take up to 250 images in RealityScan, and we recommend getting close to that number.
Background Features: The environment the object is located in can also be helpful, so it is recommended that you apply the above-mentioned recommendations to it.
If you want to learn more about photogrammetry and scanning, check out the Introduction to Photogrammetry course.