Step By Step Guide (cloned)
When working in natural marker mode the single most important step to ensure a good tracking experience is to make sure the camera bar is positioned in a way to maximise the available tracking features in an environment.
The direction it should face will depend on the environment the camera is in and the type of movements the camera is conducting. Some general points to be aware of though:
- When operating outside, in nearly all cases it is beneficial to tilt the camera bar away from the sky Ideally aim the camera bar at an area that has fixed features and less moving objects/people
- The camera bar will not track well when it is in very close proximity to a wall or floor (1m and closer). For instance, if planning on using the camera very close to the floor, then making sure the camera bar does not face down will help
- When working inside, aiming the camera bar up slightly to favour the ceiling is usually beneficial
Once you've decided on a direction for the camera bar to face it is important to calculate the camera bar to main, this can be done by the WebUI or FullUI.
In some situations it may be beneficial to apply an ND filter to the camera bar. This can help when using the camera bar outside in bright environments.
Do not use the ND filter when using the system in dark or low light areas
The ND filter prevents the camera bar from being over exposed.
To operate in natural tracking mode, Enable the Natural Feature tracking core in the preferences. It is recommended to turn off the Circular feature core.
The density value defines how many tracking points the system will attempt to acquire when the camera is moving. It's important to understand that this can't be used to "find" tracking points in areas where there simply isn't a feature to track.
The default value is 20. The value can be increased to the 25-28 range if slightly more tracking points are required, but it is highly unusual to raise the number beyond this. The system is likely to drop frames if trying to track too large a pointcloud at once
These values can be used to control the depth range that the system will use to detect tracking points from. Generally leaving the values at their default levels will cover nearly all scenarios, but finetuning the values here can help in certain situations
When operating in Natural tracking mode, the system will start tracking as soon as the system is initialised. It's important to bear in mind however that whenever the system starts tracking, the camera will need to be aligned in the space.
This can either be done manually inside the Full UI or by using the alignment wizards in the WebUI
Using the WebUI, choose an alignment method that suits your scenario.
Typically when using Natural tracking, due to it's quick setup time and adhoc nature, manually selecting the zero point is often preferred. It's important to realise though that while this is a convenient and fast way of setting the zero point, it can require some finetuning to ensure it's accuracy
Instructions for how to set a zero point in the Full UI are available here:
It is possible to turn on both the natural and circular marker core at the same time. It is typically not something we recommend but can be beneficial in certain scenarios.
If you want to finetune the position of the zero point in an already existing alignment, choose the "align world" option in the actions menu.
Then use the finetune controls to adjust the position of the zero point. When finished, save the changes and choose to either create a new alignment (will keep the original) or overwrite it
Natural tracking can work in a set and forget style workflow. It is important to realise that it is dependent on the stability of the natural tracking points that the system uses. Unless you have an environment that has very distinctive and fixed natural tracking features, we would typically recommend that Marker mode is used instead.
If you do decide to use saved alignments in natural mode, then we recommend the use of the "fiducial" markers. Information about using these can be found in the section linked here: