Digital sampling of course is that coarse wave drawn, but that kind of a blocky waveform is impossible to produce physically. Resulting output wave is always smoothened out by physical limitations DA conversion, and that circles back to the point earlier about suitable sampling frequency for creating that wave.
It's not even physical limitation, the curves are intentionally smoothed out by low pass filters. A step function (the "blocky waveform") in theory consists of areas of infinite frequency, the vertical part of the square wave. In practice this is impossible and undesirable, but if you put a lpf on the output of your DAC it will remove the areas of infinite frequency, leaving only the intentional wave left over.
The nyquist frequency has little to do with that. It tells us that the minimum sampling frequency should be at least twice the highest recorded frequency. Any sampling rate less than the nyquist frequency will not accurately capture the behavior of the signal. However the nyquist frequency isn't the lowest frequency that will yield a perfect recording, it is the lowest frequency that can be used to yield a usable model of the signal. With a sampling rate of 44kHz, we know that all signals recorded below 22kHz are accurate, but if we tried to record a steady 30kHz signal, it would not sample to a steady 30kHz signal. This is called aliasing, because one signal pretends to be another. For the audio world, sampling at the nyquist frequency for many cases is not actually all that usable.
Imagine you have a light bulb that slowly fades on and off at a frequency of one full cycle a minute. The
minimum frequency that it is possible to accurately record this behavior would be double that. You would take a picture at the minimum brightness of the bulb and at the maximum brightness. The lpf in this metaphor would be taking those pictures and using a blend effect to transition from one to the next when you are presenting them on your powerpoint. If you were to take pictures less than twice a minute, say once every 40 seconds, you wouldn't record the right behavior of the bulb. Your first picture would be accurate, your second picture would be accurate, but your third picture would be a duplicate of your second picture, and your fourth picture would be a duplicate of your first, making it look like the bulbs maximum brightness was 2/3rds of what it was, as well as different frequency. If you go to the UK where they have 50 Hz power and film a light bulb on your 60 fps camera, it is a great example of this. The light bulb will look like it's pulsing on the camera. A bit of a tangent, but we still don't quite understand how our eyes work with regards to sampling.
Sampling rate is horizontal resolution, bit depth is vertical resolution. There isn't really an equivalent to the nyquist frequency for bit depth.