Featureless Object Scan, though effectly combats the limitations for Photo Scan, still draws from certain limitations and can be common reasons for generation failure:
- Object out of frame during capture: the algorithm relies on the object being centered and framed at all times during a Featureless Object capture. As many of the objects scanned are reflective/transparent, it's crucial to capture the entire object with it being framed throughout, which helps KIRI to distinguish it from the background.
- Blurriness/Rapid movement: the footage needs to be steady throughout the capture. Blurry footage is often caused by rapid movement, which effectively captures nothing. Be sure to move in a slow and steady manner throughout the recording process.
- Object movement during capture: with Featureless Object Scan, it's essential to keep the object stationary throughout the capture to allow the machine learning algorithm to calibrate its precise location and shape.
- Fully transparent/reflective object: as innovative as Featureless Object Scan is, fully transparent surfaces are still difficult for the algorithm to identify, for example a glass cube. Fully reflective surfaces such as a mirror is also difficult as it reflects all the lights being casted on its surface, which can be very confusing for the algorithm.
