Detection Settings

Do you want to use Simple of Complex collision when detecting?

Usually we want volumes that are simplified representations of the space, so simple is often preferable.  Complex can result in unnecessarily complicated or jagged volumes as the detection will pick up more detail. 

However your game content may benefit from Complex depending on how it's set up and if there are gaps or inaccuracies in the simple collision. 

There is a slight CPU cost when detecting (i.e. piloting/scanning) but it's fairly negligible. The bigger question is around the nature of your content and what you want to detect.

While simple collision is often better, it's effectiveness is reliant on the author of the art assets to have made good collision without gaps or holes. It's often not critical for art to have air-tight collision which means Volumator could 'see through' the geometry or walls, as in the below example

In other cases, some art doesn't even have simple collision, forcing you to have to use complex if you want to detect it. Of course, in some cases this could be a good thing and you might not want to detect it. This is where it will depend on your art assets.

We can further filter the type of geometry we detect with this setting. There are three choices:

• Visibility

• Camera

• Objects (world static + word dynamic)

Visibility is straight forward in that what you can see is usually what is detected, although this means transparent objects like glass will NOT be detected as you can see through them

Camera is similar to Visibility, but is what the camera uses to not go through objects, so detects more objects. Transparent objects like glass will be detected, although other cosmetic objects may also be undesirably detected, such as foliage and cabling

Objects instead will only detect world static and world dynamic objects, which is likely to filter out other cosmetic objects but it'll depend on how your assets are set up. It  has greater CPU cost during detection as we run multiple scans, returning the first that actually has blocking collision enabled. Objects can work well in edge cases, such as when there are invisible blocking volumes that you don't want to detect. If you're getting detection where you wouldn't expect to and you've already tried switching between simple and complex, try using Object detection too.

Even with Detect Complex enabled, if there are actual physical gaps in your geometry then detection can sneak into undesirable areas. 

In these cases, you can widen the detection scan, so rather than a tiny line trace, we instead fire a sort of horizontal 'bar' instead. This can stop detection from sneaking past small gaps in the geometry and ensuring that detection does a better job of staying within the expected area.

The Detection Width determines if we use this 'bar' approach, and how wide it should be in Unreal units. With a value of 0, we don't use a bar and just do simpler line traces. If a non-zero value is used the bar-based system is used.

The downside of a non-zero Detection Width is that corner detection will not be as good. Volumator does additional passes in an attempt to find the corners of areas, but this is much harder to do when using this bar shape as the bar tends to hit more stuff while trying to find the corners. Usually if you find you need to use Detection Width to stop your volumes reaching to other areas, the less accurate corner detection is absolutely a reasonable compromise, but it's worth being aware of.

Typically if your geometry is good, ideally watertight, then you wont need to use non-zero Detection Widths, but when you're dealing with problematic geometry, it can be the difference between usable results and garbage.

To elaborate on the corner detection and what to look out for when using Detection Width, when a newly detected location is found, we try to wriggle our way to the corner. 

This is done by carrying out traces along the dot product of the hit object's normal, then we do a 2nd trace along the dot product of the result of that line trace. The goal is to get as far into the corner as possible and it does a good job. You don't really need to understand how it works, but this 'wriggling' is the reason Volumator is particularly adept at finding holes in geometry 😅

Anyway, when Detection Width is non-zero, it makes it considerably harder for the detection to find a corner as it inevitably hits something before reaching the corner, thanks to the wider scan. The end result is that some exact corners may not be detected with Detection Width increased and the corner may be cut off diagonally, as shown below. This is most evident on right angle corners like the example shown. You may need to expand the volume to compensate for the lost corners.

One of the key detection variables. This defines the density of the scanning data. Each  time something new is detected, we check to see if it's further away than existing data. This is the threshold it must breach, in Unreal units. E.g. 75 means a newly detected point must be at least 75 Unreal units from the next closest point. 

By reducing this value, we will get more scanning data for more accuracy. It increases CPU cost when generating after scanning and increases the memory footprint of the Data stored. 

If you area scanning really large areas, especially if they don't need to be hyper accurate, it can be helpful to increase this value to keep processing reasonable.  In extreme cases, Unreal can believe there is an infinite loop due to the amount of data being processed. You will see error information if this happens to inform you.

If you area scanning very detailed areas, especially with drastic changes along the perimeter that much be accurately reflected, reducing the Min Point Distance will allow Volumator to do a better job of capturing the area.

It's worth noting that Min Point Distance and Resolution (in Generation Settings) work in tandem. If you want hyper accurate volumes, try reducing both this and Resolution. If you want hyper coarse volumes for some reason, increase both. 

That said, the default Min Point Distance should work in most cases and it's usually only worth changing Min Point Distance in extreme cases. Also, as a reminder, the scanning data as defined by this setting is frozen or locked after detection. Resolution on the other hand is good to tweak after the fact as it's a Generation setting that works with the frozen scanned data.

While building, Volumator will determine if the area it's detected is way too large for the current settings. If this happens, it'll update the Min Point Distance and Resolution settings to something more appropriate. This is to stop Volumator getting overwhelmed with data or cause your PC to become sluggish. 

Scanning is carried out with sphere traces, this value is the distance we scan outwards. Increasing may result in less scans being required as greater areas can be covered quickly, although CPU use increases rapidly with this value. 

If using Piloting to scan exterior areas without geometry, thus using the pilot location as part of the scan data, it can sometimes be beneficial to reduce this value, so scanning doesn't inadvertently pick up distant areas you don't care about. 

The number of traces that fire out each scan. These fire in a ring around you and are equally spaced. Reducing this can aid CPU usage while scanning. Increasing may aid scanning accuracy in extreme cases, but is usually not necessary, especially if using a drone. 

Often flying around for longer (either piloting or drone) results in better data as it's more important to get good coverage through position rather than angle.

It's possible to set number of rays to zero. This is more useful for piloting than drone, as piloting with zero rays will allow you to create a volume representing only where you have flown the camera. With a zero ray drone, you'll likely get a jagged and compressed volume that's not terribly useful, but maybe there's a use case for this.

The frequency of scans in seconds. Tweaking can aid CPU if your computer is struggling.

Do you want to detect height while detecting or just use the current Volume Height variable amount in Generation Settings?

The height of the volume is set by the Volume height variable in Generation Settings. Howeve, you'll probably want to ensure this variable matches the actual height of  the detection area right? With this detection bool enabled, while detecting the pilot or drone will contually take height readings and when finished it'll take the maximum high and low points, figure out the height that encapsulates it all and set Volumator's Volume Height variable to this amount.

Will enable opacity on any barriers in the level during detection. This is on by default as it helps reinforce what areas are blocked and what you may have missed.  Once scanning is complete, opacity is returned to it's previous setting. See the Barrier section for more information on barriers and their opacity.