Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Gamerlord]

Get Visual Studio Community, yeah.

I just tried Visual Studio Community and the same damn KB2999226 thing happened.

[Rose]

Get windows 10?

[Gatleos]

Spoiler (click to show/hide)

So I'm trying to detect collisions between the two convex polygons at the top left. The magenta polygon is the Minkowski difference of the other polygons (derived by inverting polygon 2 and merging the list of edges, sorted by angle). If the polygon (translated by the relative positions of the input polygons) crosses over the origin (cyan cross), that means they are intersecting.

I got the shape completely right, but for some reason it doesn't line up correctly with the origin. I tried computing the centroid and offsetting by that, which almost works, but it's still off by a bit somehow:

Spoiler (click to show/hide)

The cross should be right over that closest vertex here, but for some reason the polygon is shifted slightly to the left. I just can't seem to find a formula that offsets it for the right amount for any pair of polygons.

Any ideas?

[MagmaMcFry]

Spoiler (click to show/hide)

So I'm trying to detect collisions between the two convex polygons at the top left. The magenta polygon is the Minkowski difference of the other polygons (derived by inverting polygon 2 and merging the list of edges, sorted by angle). If the polygon (translated by the relative positions of the input polygons) crosses over the origin (cyan cross), that means they are intersecting.

I got the shape completely right, but for some reason it doesn't line up correctly with the origin. I tried computing the centroid and offsetting by that, which almost works, but it's still off by a bit somehow:

Spoiler (click to show/hide)

The cross should be right over that closest vertex here, but for some reason the polygon is shifted slightly to the left. I just can't seem to find a formula that offsets it for the right amount for any pair of polygons.

Any ideas?

The Minkowski difference A - B is the set of all pointwise differences {a - b : a in A, b in B}. A vertex of the difference must be the difference of two vertices. For example, if a is the topmost vertex of A, and b is the bottommost vertex of B, then a-b is the topmost vertex of A-B. However not all differences of vertices are vertices of the difference. Make sure that the vertices you're using for aligment actually correspond to each other.

[TheBiggerFish]

Get Visual Studio Community, yeah.

I just tried Visual Studio Community and the same damn KB2999226 thing happened.

That's not good.

Have you tried searching it?

[Gatleos]

The Minkowski difference A - B is the set of all pointwise differences {a - b : a in A, b in B}. A vertex of the difference must be the difference of two vertices. For example, if a is the topmost vertex of A, and b is the bottommost vertex of B, then a-b is the topmost vertex of A-B. However not all differences of vertices are vertices of the difference. Make sure that the vertices you're using for aligment actually correspond to each other.

Yeah, that's the problem. I start with convex polygons whose edges are sorted so that the edge with the smallest theta comes first, and I merge the lists. Each vertex is just offset from the previous one, so the only offset I need to compute is the first one. So right now, the first vertex I add is at the origin. That point could either be from polygon A or B, so which points should I be subtracting to get the offset for it?

[lethosor]

I found a decent, extensive set of tutorials, but they require Visual Windows.

I'd be wary of any tutorials that require a specific IDE. If they're tutorials for using a specific IDE, that's fine, but if they're tutorials for C++ in general that "require" Visual Studio, either they really don't (but are just using VS to demonstrate), or they're using VS-specific stuff that will cause you massive trouble if you ever work with another environment.

[MagmaMcFry]

The Minkowski difference A - B is the set of all pointwise differences {a - b : a in A, b in B}. A vertex of the difference must be the difference of two vertices. For example, if a is the topmost vertex of A, and b is the bottommost vertex of B, then a-b is the topmost vertex of A-B. However not all differences of vertices are vertices of the difference. Make sure that the vertices you're using for aligment actually correspond to each other.

Yeah, that's the problem. I start with convex polygons whose edges are sorted so that the edge with the smallest theta comes first, and I merge the lists. Each vertex is just offset from the previous one, so the only offset I need to compute is the first one. So right now, the first vertex I add is at the origin. That point could either be from polygon A or B, so which points should I be subtracting to get the offset for it?

If you're sorting the edges by unsigned bearing (0° to 360°), then the first point of A-B is a-b, where a is the furthest vertex of A in the 270° direction, and b the is furthest vertex of B in the 90° direction. If there are multiple furthest vertices, then choose from them such that a is as far as possible along the 180° direction, and b is as far as possible along the 0° direction.

If you're sorting by signed bearing (-180° to 180°), then you should choose them the other way around.

[Reelya]

I found a decent, extensive set of tutorials, but they require Visual Windows.

I'd be wary of any tutorials that require a specific IDE. If they're tutorials for using a specific IDE, that's fine, but if they're tutorials for C++ in general that "require" Visual Studio, either they really don't (but are just using VS to demonstrate), or they're using VS-specific stuff that will cause you massive trouble if you ever work with another environment.

Or the things they teach you will end up (if not already) being out of date or incomplete about the features of that IDE.

[Gatleos]

I just looped through every possible combination of a-b, none of them offsets the polygon for the correct amount. Thanks for the help anyway, I'll try to think of something else.

[3man75]

Anyone ever have an issue where getting (using an actual getter for an object) gets you something, like an int let's say, but when you test to see if that int is 0 or below it fails to register?

Here's a code snippet btw

Spoiler (click to show/hide)

Code: [Select]

while (playerOne.getHealthPoints() >= 0 || playerTwo.getHealthPoints() >= 0){
System.out.println("1. Bite 2. Slash 3. Flee");
int userChoice = inputSystem.nextInt();
if (userChoice == 3){
break;
}
else if (userChoice == 1 || userChoice == 2){
playerTwo.enemyDamage(1);
}
System.out.println(playerOne.toString());//shows player pokemon
System.out.println(playerTwo.toString());//shows enemy pokemon
}


[Reelya]

Your while loop looks suspect.

In English, your while loop means "keep looping while player 1 has some hitpoints, or player 2 has some hitpoints". i.e. the battle won't end until both are dead.

you really want the behavior "keep looping while player 1 has hitpoints, and player 2 has hitpoints", i.e. keep fighting as long as both players are able to fight.

It's a subtle difference in logic but makes all the difference. Fights don't keep going until both sides are dead, they keep going while both side are alive. This is a key skill: being able to interpret your code as English, then logically follow how different values will affect the flow of the program.

[3man75]

I honestly feel ashamed..*sigh*

Thank you Reelya. I

[Reelya]

Don't be. The and/or thing happens to a lot of people. Mainly because it's often opposite to how people think of and/or in normal conversation.

Imagine "The petshop sells cats and dogs" vs "The petshop sells cats or dogs". Here, both "and" and "or" probably mean "and" (has-cats && has-dogs). Then consider "you can have cats and dogs" vs "you can have cats or dogs". Here, "and" is close to logical OR (you can get both or either), and "or" is close to XOR (you can get one but not the other).

Originally this did confuse me, but I learned to read out if-statements as English and just got used to what sense or/and were in.

[Putnam]

Here, both "and" and "or" probably mean "and" (has-cats && has-dogs)

unless, of course, "cats" and "dogs" are both represented as different bits in some 8-bit data structure, in which case you actually want cats || dogs

(e.g. if cats==00000001 and dogs==00000010, both would be cats||dogs==00000011)