milton_zoom.txt

Not sure if I understood what you meant by "tracking positions", but
I'm guessing that you need clarification on how change of coordinates
works between screen coordinates and what I call "canvas space".

This is how things work as of 1.4.0 (previous versions had a 32 bit
canvas instead of 64, but fundamentally it works the same)

A milton canvas is a -2^64 X 2^64 image. In that sense, it's not
really "infinite", but 64 bits is enough so that no one will ever run
out of canvas.

The screen is a window onto the canvas. When you zoom-out, the window
becomes larger. When you zoom-in, the window becomes smaller.

The smallest a window can get is a 1-to-1 correspondence between a
screen pixel and a canvas pixel. Milton sets the limit to be a little
bit larger than 1-to-1 because of some very good reasons which are not
relevant at the moment :P

We change coordinates from a screen coordinate S to a canvas
coordinate C with this formula:

C = S*scale + pan_center

So `scale` is the zoom-level, i.e. how big the window is, and
pan_center is how much we have panned away from the origin.

The inverse function is then:

S = (C - pan_center) / scale

If you take a look at canvas.cc, you can see these formulas with an
extra parameter called zoom_center, which we use to set the zoom
center to the location of the cursor. (See how when you zoom with the
mouse-wheel the zoom is always relative to your cursor position). For
the purpose of this explanation, we can think that zoom_center is
zero.

Let me know if this answers your question, or if I should clarify
anything else!

If you prefer email, you can contact me at bigmonachus@gmail.com