Points can easily be generated along a Bezier curve or spline using parametric values (provided by the function bezier), however these points are not evenly spaced along the curve. Points generated by parametric values will be closer together in regions with the highest curvature and furthest apart in regions that approach a straight line. This function provides three different functionalities for generating points along a Bezier curve or spline that are more evenly spaced than those generated using parametric values. The 'evenly_spaced' method generates n approximately evenly spaced points along a Bezier, optimizing point position according to specified convergence criteria. The 'max_dist' method places points along a Bezier such that the distance between consecutive points does not exceed a specified Euclidean distance (max.dist). And the 'adjoining' method generates points along a Bezier as a series of integers and is intended for use with pixel coordinates.
The input of the control points p is identical to bezier and can be a vector, matrix or list (see Details in bezier). As with bezier, when control points are input as a list and the number of control points differs for different dimensions, the degree will be elevated so that all dimensions are of uniform degree (see elevateBezierDegree). t1 and t2 are parametric values along the Bezier curve or spline between which points will be generated. The default values for t1 and t2 are the start and end points of the Bezier curve or spline, respectively. For a Bezier spline, if t2 is not specified, it is calculated based on the number of control points and the degree (deg). When using pointsOnBezier for Bezier splines, deg must be specified or else the points will be treated as a single Bezier curve.
If n is non-NULL, pointsOnBezier generates n evenly spaced points along a Bezier curve or spline. This requires accurate approximation of Bezier arc length. An initial estimation of the total arc length between t1 and t2 is made (using bezierArcLength) to determine the interval at which points should be placed to equally subdivide the curve. optim is used to find the optimal position of each point, calling bezierArcLength via compareBezierArcLength, such that the arc length between points is nearly equal to this interval. When positioning each point, the arc length is estimated from t1 (rather than from the previous point) so that errors are not compounded. As a consequence of repeated calls to optim and bezierArcLength, this functionality can be rather slow.
The parameters ending in min.slope are convergence criteria passed to bezierArcLength. The parameters relative.min.slope and absolute.min.slope are the criteria used in the initial arc length estimation, while sub.relative.min.slope and sub.absolute.min.slope are the criteria used to estimate arc length in placing each point along the curve. Larger convergence criteria values will cause pointsOnBezier to run faster but at lower accuracy. For a complete description of the convergence criteria, see Details in bezierArcLength.
pointsOnBezier runs an alternative routine when max.dist is non-NULL. In this case, n and the convergence criteria are ignored, bezierArcLength is not called and pointsOnBezier generates points along a Bezier such that the distance between consecutive points does not exceed the specified Euclidean distance max.dist. The parameter max.dist.factor is a factor that is used to iteratively increase the parametric value to reach the next point without exceeding max.dist. The lower max.dist.factor is, the closer the interpoint Euclidean distance will be to max.dist but the longer pointsOnBezier will take to run (see Examples). If max.dist does not evenly divide the total arc length between t1 and t2, the interval between the second-to-last point and the end point may not be close to max.dist. If max.dist evenly divides the arc length, if max.dist.factor is low and if max.dist is small, the points will be more evenly spaced than with parametric point generation.
When method is 'adjoining', pointsOnBezier will generate points as integers at adjoining positions along the Bezier curve or spline. The arc length is first measured (very roughly) to approximate the first parametric interval at which to find adjoining points. The function adds this initial interval to t1 and finds the position of the next Bezier point, rounded to the nearest integer. The interval is decreased or increased depending on whether the point is too distant or too near until the next point adjoins the previous point. Adjoining is defined as two points whose positions (as integers) differ by one in either or both coordinates. Thus, 2D points adjoining on the diagonal are considered adjoining. For instance, the points [3,5] would be adjoining with [4,5], [3,4] and [4,6] but not [1,3]. The function continues to iterate through the parametric values up to t2, generating points adjoining to the previous point. This method is intended for use with pixel coordinates, such as when Bezier control points are used to trace Bezier curves and splines on an image. Unlike the previous two methods, most of the generated points will not fall exactly on the Bezier curve since they are rounded to the nearest integer. This method currently only works with curves or splines in two dimensions.
In the case of Bezier splines, note that borders between spline segments are not respected and arc lengths are calculated across spline segments. In order to generate points within spline segments, pointsOnBezier should be called separately for each segment.