CODEWITHSUNDEEP
geometrycollision-detectionphysics
Armen Tsirunyan
Armen Tsirunyan

Reputation: 133072

Arkanoid Collision Detection... Again

I have read dozens of questions here on SO (and not only) regarding arkanoid collision detection, namely, moving circle collision against a stationery rectangle, but all of them ask how to detect a collision or how to detect which side of the rectangle the ball hits. My question is a bit different - it concerns the calculation of the new speed direction in case when the ball hits the angle of the rectangle.

For simplicity's sake let's assume that Vx >= 0 and Vy <= 0, that is, the ball is coming from the below from the left upward and rightward and also suppose I know it's gonna hit the lower side of the rectangle. The green arrow shows the approximate direction of the ball and the blue dot shows the first point on the line containing the lower side of the rectangle that the ball hits. If that point lies strictly within the lower side of the rectangle then all is trivial - just change Vy to -Vy. However when that point lies outside the lower side it means that the first point of the rectangle that the ball will touch is going to be its lower-left corner, in which case I don't think that changing the Vy to -Vy is correct. I think that the new velocity angle must be dependent on the distance of the blue point to the corner. Also I think that not only Vy but also Vx must change (preserving, possibly, the length of the V vector). pic

So, how do we calculate the new Vx and Vy when we hit an angle? If you know any good links that address this question I'd be delighted to know them. Also note that I am more interested in the absolute physical model of this rather than easy-to-code optimized approximations. You can assume there is no rotation involved. Thank you very much in advance

Upvotes: 4

Views: 2402

Answers (2)

Ian Turner
Ian Turner

Reputation: 1413

A simpliciation that is often made when modelling rock blocks in rock mechanics is to assume that the corner of the rectangle has a small radius. The calculation is then one of two curved surfaces contacting. This approach tends to give more consistent behaviour than modelling corners as right angles an so is preferred where consistent results are needed.

Upvotes: 1

Beta
Beta

Reputation: 99134

You know how to bounce off a horizontal wall. Do you know how to bounce off a wall that is at some other angle?

When the circle hits the wall, it makes contact at a single point. That single point is all the circle "knows" about the wall; the location of that point gives you enough information to calculate the new V. When the circle hits the corner, it makes contact at a single point (i.e. the corner) so it bounces just as if it hit a wall at that point.

Is that enough to go on, or would you like some math? (And if so, how comfortable are you with vector algebra?)

Upvotes: 1

Related Questions