A bowling ball wouldn't curve significantly unless at very high speed and moderate to high spin, perhaps if it could be shot out of some type of high pressure cannon.
Something like a hollow wiffle ball without the holes could be thrown to produce a slight curve.
Table tennis balls are very light for their size and curve easily. The table tennis paddles use a very sticky and elastic rubber surface, allowing a lot of speed and spin on the balls, so they can curve quite a bit, including truly rising from backspin. Example video with mostly top spin and side spin related curves:
http://rcgldr.net/real/tt2.wmv
The best description I could find is from this archived article:
The more recent studies agree that the magnus force results from the asymmetric distortion of the boundary layer displacement thickness caused by the combined spinning and flow past the spherer. In the case of a sphere(or cylinder), the so-called whirlpool, or more accurately the circulation, does not consist of air set into rotation by friction with a spinning object. Actually an object such as a sphere or a cylinder can impart a spinning motion to only a very thin layer next to the surface. The motion imparted to this layer affects the manner in which the flow separates from the surface in the rear. Boundary layer separation is delayed on the side of the spinning object that is moving in the same direction as the free stream flow, while the separation occurs prematurely on the side moving against the free stream flow. The wake then shifts toward the side moving against the free stream flow. As a result, flow past the object is deflected, and the resulting change in momentum flux causes a force in the opposite direction(upwards in the case shown in figure 1).
http://web.archive.org/web/20071018203238/http://www.geocities.com/k_achutarao/MAGNUS/magnus.html
The key here is the boundary layer separation at the rear of a ball.
I was wrong about golf balls and corrected my previous post. In the case of a golf ball, because of the higher speeds, the laminar flow would separate from the ball on the front side of a smooth ball resulting in a larger wake that creates a type of "wake form drag". The dimples trip the boundary layer into a higher energy laminar flow, which can follow a curved surface better than a lower energy laminar flow. There's increased "skin drag", but the turbulent flow reattaches and follows the curve of the ball to the back side of the ball, resulting in a smaller wake, and the decrease in "wake form drag" more than offsets the increase in "skin friction drag".
http://www.knetgolf.com/GolfBallDimp.aspx
Looking at that golf ball article, if the dimples are too small, then you get more lift, but also more drag. The wake would be larger, and displaced enough so that the Magnus effect is also larger, but the large wake produces more drag. At the other extreme, if the dimples are too deep, you get less lift and less drag, but the total distance becomes less than optimal dimple size.
The competing issues here are how large the wake is and by how much that wake is deflected, and this is affected by the smoothness, speed, size, and spin of the ball. If the ball is moving slower, the laminar flow remains attached longer, eventually remaining attached on the back side of a ball, which changes the situation allowing smoother ball to curve more, which is the case for table tennis balls.
In the case of table tennis balls, the ball tends to curve less at higher speeds, then curve more as the speed decreases due to the relatively high amount of drag. My guess is that most of the curving takes place when the speed of the ball decreases to a point where the laminar flow remains attached until reaching the back side of the ball, which would explain why smoother table tennis balls tend to curve more. However, it could be related to the fact the the smoother balls slow down sooner, so you see the transition into the region where the curving becomes more noticable sooner. This Wiki article has some diagrams of the path of a table tennis ball with backspin and topspin examples:
http://en.wikipedia.org/wiki/Table_tennis#Effects_of_spin
So the smoother versus rougher surface versus Magnus Effect response depends on the speed and nature of the ball. According to some web sites, within a certain speed and spin range, and with a smooth ball, it can end up up with laminar flow on one side, and turbulent flow on the other, resulting in a reverse Magnus effect with the ball curving the "wrong" way.
Getting back to the original question, a smooth ball can curve, from this wiki article:
For a smooth ball with spin ratio of 0.5 to 4.5, typical lift coefficients range from 0.2 to 0.6.
http://en.wikipedia.org/wiki/Magnus_effect#Calculation_of_lift_force