We seem to disagree on a lot of things, which is fine since everyone has their own ways of doing things and there really is no "right" way. I do enjoy sharing and comparing methods, so don't take this as any kind of slight that we do disagree. I do want to share why I don't use the methods you like and why I do like the more tedious method of generating additional geometry to drive toolpaths. If I had the "Flow" option you have in HSM Pro, I would be using that over any method available in Fusion for roundovers, but we don't have that one yet (I'm hopeful it will be part of Fusion Ultimate when they finally start to delineate between levels).
I have not been able to produce a finish with 3d contour that is as good as the Morphed Spiral option, in my opinion, even when using touch surfaces. I did figure out that you can get by without adding an offset boundary if you use the "Tool Outside of Boundary" option, but my feelings on that option are that I'd rather create a boundary that I control than leave it more arbitrary. Old habits die hard, so my preference will probably always be to use a defined boundary instead of that setting in particular. It does work on the example shown (first image below), and the same method with Morphed Spiral does "work" on both the same part (second image) as well as the 3d example I used earlier (third image shown), but you can see that the drive surface concept produces a better result on the complex 3d roundover than using the "touch/avoid" combined with "tool outside boundary" method, which resulted in extra jumps up and down and a disjointed toolpath compared to a completely contiguous toolpath when using the drive surface instead (image in previous post).
The other problem I have with the 3d Contour option is that the stepovers are not uniform at all, with or without the "Machine Shallow Areas", and the toolpath never reaches to intersection between the top surface and the start of the roundover no matter how low I set the stepdown value that controls that (fourth image below has .01mm stepdown and .1mm step over, but gets no closer to the intersection than previously). In theory, there isn't much difference in Z level between where the 3d contour is starting and the Z level where that intersection occurs, but the stepover from that point is large enough that I'd expect a noticeable scallop in the finish when using a 1/2" ballend mill. 3d Contour does not seem to take that into account when it's determining where to start the toolpath, only how much the geometry has stepped down from flat to determine that. That places much more importance on how you finish the top surface in determining the actual finish quality at that intersection. If you use a bit face mill, or you use a parallel strategy with a cap over the opening in the example below (first image is one of 3d Contour), you could help reduce the impact of the way the 3d Contour operation works, but you can clearly see that the Morphed Spiral example (2nd image) provides cutting all the way up to the intersection between the top and the start of the roundover, leaving no chance that the finish will be substandard.
When you then move on to the roundover on a 3d surface edge, as in my previous example, you can see that 3d Contour produces a very undesirable toolpath (fifth image below). By it's nature, 3d Contour will machine at a Z-level, then step up to the next Z-level, etc. On any 3d roundover that follows a complex edge, this will just not work. As you can see in the image, the result would be poorly finished due to so many entry/exits and require perhaps 3 times the machine time as compared to the Morphed Spiral version.
If I've missed ways to improve these results, I'd love to learn more about the settings in HSM/Fusion. To my knowledge, I've produced about as good of a result as each strategy is capable of for the given example, and my personal evaluation is that Morphed Spiral is the go-to tool for roundovers based on the results I've been able to produce on these sample parts. Obviously, real world situations vary and some conditions will favor different methods than others. I do think these are both pretty fair examples of the kinds of roundovers I run into on a regular basis and my experience is that it works the same on most parts I cut as shown here. For example, if you want to cut a roundover on the perimeter of a Strat-style guitar, that roundover comes out closely to the same as the complex 3d roundover example I've shown here. I can get a neraly continuous motion around the guitar by linking the roundover thought the neck pocket and using a drive surface around the entire guitar as well. I have had no success machining that kind of complex roundover using any other method, though as I said earlier I'd much prefer to have the "Flow" option available which would completely eliminate the need for a drive surface.
As a side note, I have created a feature suggestion for Fusion that I'd love people here to go vote for. It is a modification to the Scallop strategy to work in a spiral manner, which would eliminate the links between passes and provide a similar continuous path that Morphed Spiral provides, with the exception that the step over is always the same instead of morphing. I've also suggested that if it's easier to develop, the same option could be made in Morphed Spiral by creating a "Suppress Morphing" option so that the stepover is a constant in that strategy. Either way, this would be a huge improvement to one or the other of those strategies. I posted a similar toolpath produced in Bobcad using the spiral with constant stepover as an example of the difference, and I'm sure that no one would argue it produces by far and away the best result for his kind of condition due to the spiraling contiguous toolpath combined with a completely constant (and therefor consistent) stepover all the way around the roundover. This would be a very powerful option in my experience. Please have a look here and vote:
Spiral Option in Scallop Strateg... - Autodesk Community
Attachment 296962
Attachment 296964
Attachment 296966
Attachment 296968
Attachment 296970