Obviously not those sample points. I didn't have access to them at first, and even now I have no reason to think they match exactly with the vector capture. The trigger setting is different, and dots/vectors trigger a little differently.
The trigger setting didn't really matter - here's another image I just captured of the same 100MHz uninterpolated sine wave (sorry - slightly less amplitude due to a loose 50 Ohm terminator) with the trigger set to the previous 180mV level. It's almost the same image as before - with the dots just shifted horizontally.
Compare the peak from your interpolated capture with one that has plenty of data and see how much rounder the correct curve is!
DSOs are imperfect devices - and besides doing the sin(x)/x interpolation, the scope is doing other transformations to the sample data to get it to the display (e.g. the Rigol is mapping 200 bits of vertical ADC resolution to 400 pixels of vertical screen resolution). Perfectly correct curves may or may not be precisely what you see on the display - although with a small number of sample points, the difference between linear and sin(x)/x interpolation is pretty noticeable.
Here are two images from a LeCroy Waverunner LT 224; the first one showing sin(x)/x using 10 samples per div. Can you see what looks like short line segments at the tops and bottoms of some of the sine waves?
This image shows both linear and sin(x)/x interpolation using 5 sample points per div as in my Rigol example (although with longer cycles).
If I then chop off the pointy bit, I get something very similar to your capture (effectively adding just one more sample per peak.)
Well sure, adding sample points in convenient locations can definitely help linear interpolation look more like sin(x)/x.
But in any case, to me, the difference between your linear interpolation and the 'bad' curve is still noticeable - one looks like straight vectors and the other looks like less-than-perfect curve fitting.
But either way, sorry to ask and I won't take any more of this long thread with what's way off topic.
It wasn't any problem to ask, and I don't think it's off topic since it's about the Rigol's interpolation (and I brought it up in the first place). I was just surprised at the question - and I thought I answered good-naturedly with a bit of ribbing - while trying to point out that, IMO, it would have been clear if linear interpolation had been used on that waveform with 5 points per div. - even though, as I mentioned in my later post, I understood your point and conceded that it would have been more clear if I had used a lower frequency sine wave in the example.