I remember getting the same piece of advice from the teacher. Problem is, even before getting it, I was already applying it, being a rare kid experienced in computer graphics. The teacher was just repeating a phrase she heard somewhere, without actual competence to direct me.

The way I see, and I think the way most people see, is that I have subpixels, not distributed in a square grid and small enough, too many to be able to count them - but I can see them when I close my eyes, it's somewhat similar to looking at a colored noise - something like this: https://i.imgur.com/1P3n80k.gif except you would have to display it on a ridiculously high resolution display (I don't know, 64k or maybe more) and it would represent just a small fragment of view.

Of course this unordered constellation of cones can be mapped into a grid of pixel or a space on a paper, so the only problem is I can't make a measurement in my head and I need to calibrate "eye-balling" measurement to figure out where on paper should I put what I see and I deal with it typically by imagining vertical and horizontal lines to subdivide my view, and then I likewise subdivide the paper.

So I don't really have a problem drawing what I see, the problem I have is the missing technique of how to use a pencil to draw what I actually want to draw.

I think most people work the same way but apparently you don't?