Mantis Vision F5-B, Руководство пользователя

Страница: 29 / 126

F5 3D Imaging System User Manual
29

At the same time, hovering or covering the object very slowly results in
very large files of unnecessary overlapping data that may decrease
model accuracy and require more work (processing time) to achieve
the desired result

Plan the Shot
Before beginning to record a scene, visually look around the area. Plan the
positions of symmetrical/planar patches, obstacles and obstructed objects in
the scene. Then, determine the best path for the Imager to cover the scene
from all sides (meaning, the fly-through path of the Imager).

Symmetry
Avoid recording sections of empty walls and tabletops (flat surfaces), and
strive to capture frames that contain several angles. Frames containing
minimal 3D geometry are very hard to combine accurately later.
Smaller amounts of 3D geometry within the recorded frames make it more
difficult in later stages for the software to determine the correct alignment
between frames during the stitching process.

Scene Size
The Imager was designed to work optimally for objects between the size of a
computer mouse and a small room, meaning approximately 3m x 3m x 3m
(10’ x 10’ x 10’). Such scenes may be acquired in up to a minute. As scenes
become larger, they become more complicated to stitch and may require
more user intervention to produce accurate 3D models.

Dynamic Scenes
Individual 3D video frames (point-cloud video) of objects and people in
motion, may be viewed and measured. However, only static scenes may be
stitched (aligned) into a consistent (Euclidean) 3D point-cloud model.

Materials
Some objects are made of materials that are difficult to sample, such as:

Wires:
Thin wires and nets (where holes are wider than the threads) are
difficult to capture when the width is less than approximately 6 mm
(approximately ¼ of an inch). When capturing such objects, take longer
shots and vary the recording angles frequently.