Seriously though, if you want to do any more playing around, I would strongly suggest that you don't use 14M or 56M sample lengths - since traditionally, one of the tools to battle against aliasing is to increase sample length (since that automatically increases sample rates and/or samples being decimated for the display). If anti-aliasing works at all on the Rigol, it should first and foremost be working when you have small sample lengths - so that switching it on might (in the background) automatically force the DSO to capture more samples for random decimation (or change sample speeds) in order to prevent the occurrence of the aliased waveform.
I didn't post pictures of the smaller memory depths because AA never did anything for them, not because I didn't try it. But going back at it, I am able to get a
very subtle change at 140k in some situations. I experimented with what happens to the sample dots themselves and could find no difference at any memory depth.
And that leads to the final nail in the coffin: You can toggle AA while STOPped, and it does the same thing as toggling it while running. So you're right, it's image improving, and not waveform improving. This is true for High Res, too! Perhaps that explains why you never get 10-bit values out of the scope with RUU: it's entirely a display-time trick.
High Res troubles me in general. For example, 1Vpp 66,666Hz sine wave at 10us/div, 200mV/div looks great in high res. Now go to 1ms/div. The signal shrinks to half amplitude! Now turn on Anti-Aliasing for some real fun: it...
aliases into a 8% amplitude, 5kHz wave. Changing the vertical scale to 50mv and it halves the amplitude again! Like I said, AA and High Res don't get along
If I try to zoom all the way into the sample dots with high res active, I can't. I always get a line.
AA does make some of my captures prettier, enhances certain glitches, and hasn't (yet) hidden any transients in my testing tonight. The biggest drawback seems to be the wfm/s hit (which varies considerably based on the time base and memory depth.)