0

在一般术语和伪代码中,如果墙壁实际上只是一个点正在碰撞的整个正方形的一部分,那么获得沿墙壁滑动的碰撞响应的最佳方式是什么?使用的碰撞测试方法是测试点是否在正方形中。

我应该把正方形分成四条线,然后计算到直线的最短距离,然后将点移回那个距离吗?如果是这样,那么我如何确定碰撞后点最接近正方形的哪条边?

4

1 回答 1

2

通过测试墙壁上的运动矢量来检测碰撞点。如果你了解你的表面(例如你说它是一个盒子的一部分),你也许可以同时测试多个墙壁。

解决方案在 2D 和 3D 之间可能略有不同。我将使用 2D,因为您说的是“方形”而不是“立方体”或“盒子”。

一旦你知道你的点击中的位置,你就取你的运动矢量的剩余部分,将它点在墙壁方向上(从另一个点减去墙上的一个点,然后标准化),然后按这个量缩放墙壁方向。假设没有摩擦,这是平行于墙壁的运动量。

编辑添加了以下代码:

样板:

import math

class Vector2d:
    def __init__(self, x, y):
        self.x = x
        self.y = y

    def __add__(self, rhs):
        return Vector2d(self.x + rhs.x, self.y + rhs.y)

    def __sub__(self, rhs):
        return Vector2d(self.x - rhs.x, self.y - rhs.y)

    def GetScaled(self, scale):
        return Vector2d(self.x * scale, self.y * scale)

    def GetLength(self):
        return math.sqrt((self.x * self.x) + (self.y * self.y))

    def GetNormalized(self):
        return self.GetScaled(1.0 / self.GetLength())

def DotProduct(v0, v1):
    return (v0.x * v1.x) + (v0.y * v1.y)

真正的业务:

class Wall2d:
    def init(self, point0, point1):
        """point0, point1 are Vector2ds"""
        self.p0 = point0
        self.p1 = point1

        # these three steps could be combined to optimize, but
        # for demonstration are left explicit
        self.dir = self.p1 - self.p0
        self.length = self.dir.GetLength()
        self.dir = self.dir.GetNormalized()

        # computing the normal in 3D would require three points
        # on the face and a cross product
        self.normal = Vector2d(self.length.y, -self.length.x)

    def LineSegmentCollides(self, pointStart, pointEnd):
        startDot = DotProduct(pointStart - self.p0, self.normal)
        endDot = DotProduct(pointEnd - self.p0, self.normal)
        if startDot * endDot < 0:
            # the only way a collision can occur is if the start
            # and end are on opposite sides of the wall, so their
            # dot product results will have opposite signs, so
            # the result of the multiplication is negative
            moveVector = pointEnd - pointStart

            # scale the movement vector by the ratio of the move
            # vector on the "start" side versus the total length
            # of the movement in the axis of the normal
            collisionDelta = moveVector.GetScaled(startDot /
                                                  (startDot + endDot))
            collisionPoint = pointStart + collisionDelta

            collisionDot = DotProduct(collisionPoint - self.p0, self.dir)
            if (collisionDot > 0) && (collisionDot < self.length):
                # we've hit the wall between p0 and p1 (other
                # values of collisionDot mean we missed on one
                # end or the other)

                # now, collision response is up to you.  In this
                # case, we'll just zero out the movement in the
                # direction of the wall after the collision
                # (sorry about the poor naming)
                # note that we don't actually care about the actual
                # point of collision here.
                collisionPushBack = moveVector.GetScaled(
                                         endDot / (startDot + endDot))
                endPoint = pointEnd + collisionPushBack

                return True
        return False

我希望这很有用。

于 2010-03-19T17:25:32.037 回答