D-scan specifics

hu, no ?

A program does not “have to”. and actually, “should not”.

The program you are considering is very, very, very bad.

1 Like

I’ve been precisely saying the distance scan method described earlier would be spectacularly inefficient. I’m not sure where we are disagreeing here.

The volume of the area scanned is easier, and simply compares the objects’ coordinates with the coordinates contained within the scan.

Are we in disagreement there?

Somehow your action “d-scan” has to define a volume that is checked against the coordinates of each object.

We’ve only talked aobut how that check is made, and I think we both rule out the idea that it would immitate an actual distance scan.

Yes it does have to, you have to enumerate the state of each object with regards to the action performed.

For instance you D-scan an area, is that object contained in that area, true or false? That process requires checks against each objects’ coordinate placement.

Unless you’re suggesting there’s another set of data the object has that the d-scan trigger checks against?

that can be done without a “check for the value in each location.”

why do you need to define an area ?

You’re telling me that you do action: d-scan and you will know the state of all the objects therefore returning a value to the user, without checking their location?

Because the scan only returns results based on where you set the scan, the volume the scan checks.

No, I tell you exactly what I wrote in the post I wrote. Nothing more.
You don’t need to “check for value in each location”.

BTW you sentence makes no sense at all. That’s why no, I don’t mean “no sense at all”.

Not based on the volume.
You don’t need to define an area.
You are making it unnecessarily complicated.

So lets be clear : the user is not making the DSCAN. The server is.
When a user makes a dscan in a direction and with a given angle and distance, the server must ONLY return the list of items that are in the given direction ± the angle, and in the distance required.

The server already knows ALL the items in space. So all it NEEDS to do is to iterate over them and check if their position meet the conditions specified by the user. Anything more is a waste of time.

more formally, once you have the point for the user, a point to make the direction, you can iterate over each entity in space and

  1. check if the distance² from user to the entity poistion is <= distance² specified by user
  2. find the distance da from the entity to the directional vector using eg http://mathworld.wolfram.com/Point-LineDistance3-Dimensional.html , and from this you can check if the ratio da²<=sin²(angle)*d² (that is if the entity is within the cone of direction+angle)
1 Like

Let me rephrase because I probably wrote it poorly and we are just misquoting each other.

Every object in system has to have a coordinate reference of where it is in the system.

Your d-scan has to define an area of the scan, and refrence that object’s coordinates to see if it is within the scan volume.

I don’t see what you’re getting at:

When you say this, this proves my point in the following way.

When your action “D-scan” is executed, the server has to find what items are in it, by referencing the objects in system and their coordinates.

The d-scan function needs to enumerate this somehow.

no, the “volume” you are talking about is not necessary.

I updated the previous post with a part of the formula to check if an entity is withing dscan range. There is no need to have any notion of volume.

Your notion of volume/area is a waste of time. Forget it. There is no volume. no area. Just the user position, a vector of direction (that is a second point), an angle, a max distance.
With this you can tell if an entity is within range or not.

the server already has the list of entities in the system. Enumeration is not an issue. Filter is.

1 Like

Grr, what you keep missing is that you have to do this for every object in the system.

Somehow you’re going to have to have a true/false value whether its:

Check object > compare to distance-angle > is the distance-angle true for that object?

Or if it is:

Check volume for coordinates valid > are the objects’ coordinates within the valid coordiates?

The volume is a lie.

You are making the things more complicated than they actually are.

1 Like

You have to check against something to validate the objects’ location within or without the scan parameters.

something is a filter.

You don’t need a volume to make a filter.

the filter is defined by (distance, direction, user position,angle). period.

the 10x10x10 volume makes no sense here.

Indeed something is a filter, I have suggested that a non-volume filter is more intensive computationally.

more intensive than WHAT ?
What is a non-volume filter ?
again : YOUR VOLUME MAKES NO SENSE.
even in real life, a volume is described by its limit.

The only case where it’s worth it to represent a system as an aggregate of points is when there are very complex relations between those points, for example nuclear/thermal/fluid modeling.
In most cases you don’t need to have a discrete modeling of space and instead can use continuous effects, that is the “volume” is modeled as forces applying to the interesting elements of the system. So your whole system is only represented as a very little number of items to manage (that’s why it’s hollow)

What in the actual hell is this thread? How is it still going?

You point the camera, which points the d-scan.
You set an arc, whether it’s 360 or 180 or 5 degrees.
You set a distance from yourself, whether it’s 14 au or 1 au.
Then you hit scan.

If there is something within the arc, within that distance, then it shows. It’s going to be rounded at the edge, b/c that’s how radius works. Straight line from the origin out to the spherical edge of chosen distance. This is clearly obvious with a 360 scan b/c you get to see the entire sphere. When you reduce the arc, you’re seeing portions of the sphere, but it’s still rounded out there at the edge.

If it didn’t, then the flat face of the cone would either be further than designated at the edges, or the center would be closer than designated.

2 Likes

My emphasis.

You claimed the calculation needs to be done for every point in space, which would be inefficient, I claim the calculation needs to be for every ship in space.

So something like this:

def dscan(myship, direction):
    for ship in system.all_ships:
        d = dist(myship, ship)
        a = angle(direction, ship)
        if (d < maxd and a < maxa): dscan_report.add(ship)

Where the dist and angle functions are as I defined above. This implicitly defines a rounded cone in space. Ships within this cone will be shown on dscan.

Obviously I’m guessing how CCP has coded this. Possibly depending on how the system data is structured, you could use some heuristics to reduce the amount of objects you have to check.

I cant prove its a cone, but it is. I use it to find targets at scannable locations without combat probes. You just have to remember to aline the cone with the ship. I’m very rusty at the tactic, but it looks like I’m going to have to brush up on it now. :grinning:

https://forums-archive.eveonline.com/topic/432114/

Make one, drop cans.

The end of the cone can not be flat as it would curve at the upmost range scan is possible at.

Thats about as basic as it gets but in reality what your gonna end up doing is pressing v a lot and just reducing the angle; looking at the cone probably isnt as efficient but each to their own. :slight_smile:

You gotta prove this one.

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.