Code Logic Question

[EmperorJon]

This one has me stumped.
Yes, I know, it sounds stupidly easy, and it's ovbiously a problem that has been long solved.


Assuming one has an integer coordinate grid, how does one make a 2 dimensional object move gradually to a specific coordinate, when it's NOT moving in one of the 8 cardinal directions?

Currently my 2 attempts have led to 2 different outcomes; forcing the object to randomly fly around in cardinal directions, and causing it to beeline for the point it's told to go for... but then reach a point where the precision causes all problems and it just hovers backwards and forwards. (Seems to be going "Move one step. Now that's too far, move one back. Ok, try Y now. Nope, and X? Nope. Ok, I'll just keep doing this.)

[Gazz]

You have 2 vectors. Take the higher ABS(scalar) of those.

Your object moves in [scalar] steps = n.

x.vector = ( n * x.vector.total ) / scalar
y.vector = ( n * y.vector.total ) / scalar

If the integer overflow doesn't get you, that's it.
With X3, you kinda get used to integer math. =)

[EmperorJon]

I did something every similar to that and it didn't seem to work...

[InFlamesForEver]

Oh god Gazz. That makes my knowledge of coding (visual basic lol) quiver.
Although, in computing, we seem to do less coding more theory, I hate theory...

[Samuel Creshal]

[Gazz]

[EmperorJon]

[Chips]

Without a code example, kinda hard to help. No idea what you've done so fa

However, I believe the word you need is:

Matrices (Matrix)


http://en.wikipedia.org/wiki/Translation_%28geometry%29
http://wally.cs.iupui.edu/n351/3D/matrix.html

You can define the points of your object in the form of a matrix... as well as the function you wish to apply to it (translation, scalar, rotation) as another matrix. A bit of matrix math (multiplication) and you get out your new coordinates...

[EmperorJon]

Yeah, it'll end up being taped together with matrices I dare say...


Gazz, hopefully an example on your part will clear things up for me.


Say I'm at (0,0) and I want to move to (30,17).
I need to do this in steps, however I can't stop at a point that isn't an integer.
The numbers just don't work. There's no integer common factor for them.
Surely this means, unless I want to do it all in one jump, I'm require to stop somewhere that's "close enough" and introduce inaccuracies some how?

EDIT: Well, scrapped it all and I'm going to bed. Too tired for coding.

[Gazz]

Well, then 30 steps.

x.vector = ( n * 30 ) / 30
y.vector = ( n * 17 ) / 30

It doesn't matter if it's integer or not. There is no iteration "somewhere around" the destination because you can only arrive at the exact destination. You have no choice.


Well, depending on your floating point math you may end up at something like (29,999999 / 16,9999999) but that, too, can be avoided if you care and if it's mission critical.
To do that you'd not be adding rounded values together but instead calculate the entire vector for every step. That way you eventually have maxScalar / maxScalar and no rounding error whatsoever.

[Tsar_of_Cows]

[berth]

With the major caveat that I know piss-all about coding and from a purely conceptual, coordinate geometry-type approach..

could you just go from (0,0) to (30,0) and then to (30,17)? Or are you tied to travelling in a straight line?

[InFlamesForEver]

[EmperorJon]

Visual Basic is not for making games!


Anyway, thanks for that, it's very very similar to something I tried earlier, basically by saying "Put it rounded at the next coordinate" then checking if it's 1 away fromt he destination we can cheat it, I think.

[Samuel Creshal]