Not to derail the thread, but do you find that paper w/ compound on some type of board or something flat does not round the apex? I am wanting to try this. Denim on a hardbacked surface still rounds the teeth off from what I have tried. Also, when you use the paper do you elevate the angle as Fred was stating to make sure the burr is gone?
Some of our differences/preferences may also be due to our experiences w/ 2 different systems... I use a guided jig (WEPS) because I am not skilled enough to do it freehand like you

:thumbup:
Had to take a minute to write this up, hope it is comprehensible.
What I have found -
compound directly on hardwood tends not to round at all except for user error, but also makes it very difficult to cleanly work the apex. Often there will be a burr left, just like using a trailing pass on a hard fixed abrasive surface. Open cell wood like red oak works better than clear maple - this is actually a method used in many regions of N Europe and elsewhere to sharpen tools, on a knife board with loose grit. The effect is not as crisp as off a stone, but can still be pretty good.
compound on softer wood like balsa or poplar will allow the minerals to sink into the wood a bit more and still have some mobility - burr formation is reduced or reversed - burr is gone. Depending on abrasives used, edge will be considerably more polished and uniform at the apex. Feedback is not good, stock removal rates not very high compared to a stone, some rounding will become apparent with repeated or prolonger use. As long as you don't press so hard it deflects into the wood (which might not take that much force at all...) , or errantly elevate the spine too high, the rounding will be spread over the entire bevel surface and not localized at the apex.
Paper over a smooth stone allows the edge bevel to sink into the stock somewhat. The paper has formation to it, and will deflect and expand at pressure levels at or below what is needed to get a good polish. So yes, still some rounding at the apex. This tendency is also present in sandpapers and to lesser extent but still detectable, in lapping films. Is pressure dependent, so with a real light touch, few passes, working at a lower angle than the original grind, rounding of any sort will be slight and spread over the entire bevel. Thinner papers will display less rounding, all other things being equal, and can fall into the range of poplar or balsa.
If you wrap the paper around the coarse side of a vitreous stone, the high points on the stone press into the paper and the formation collapses (little or no re-expansion after pressure is removed) with less applied force - the safe pressure operating range increases as the paper density goes up. The open areas around the high points allow the paper to deflect away from the edge with less pressure, they do not push back as hard as they return to their original volume/location. There is less overall rounding of the apex and the bevel itself. And as with all methods that reduce the abrasive footprint, stock removal rates go up as well until the footprint gets so small it looses effectiveness. Taken to something of an extreme is my Washboard and its almost bladelike teeth pushing into the paper in a uniform, open pattern. It was directly influenced by the observation of paper over a coarse stone, among others. As with working off a board, most of the rounding is operator error, not an inherent issue with the materials. Again, thinner papers will display less rounding, all other things being equal.
Plain paper over a hard surface will not round at all that I can detect under magnification, even with a number of passes. The abrasives are not hard enough, or a large enough percentage of the surface area but along with the paper fibers themselves, are capable of lining up the edge, removing very small burrs, and if the spine is elevated, some slight burnishing of low carbide steels will occur. This is how I finish most sharpening, and all sharpening when finishing right off a stone, as much to apply a slight burnish as to make certain all burrs are gone. Plain paper burnishing is the safest way to finish an edge if no rounding is acceptable.
If compound is involved I do not elevate the spine at all - whatever slight deflection occurs is enough of an apex clean up for me. I can detect the original grind angle, I cannot accurately gauge the shift if I start elevating the spine in the absence of specific feedback, and the angle will open up to unacceptable levels almost immediately.
Much of the above has to do with repetition and pressure tolerance. A small number of very light passes on denim over a hard surface is unlikely to cause issues. Is when the spine is elevated, pressure levels high enough to compress the fabric are used that real rounding occurs. Any material with give is going to have this tendency - a range of safe pressure, a range where it begins to sag, a region where it re-expands to fill its previous volume as pressure is removed, rounding the bevel and possibly the edge. Putting abrasives on these materials allows the minerals to move around a bit, reducing/reversing burr formation and presenting a more uniform abrasive field. Managing the ill effects involves understanding the limitations of the specific materials.