[Curves->Marching Squares on Surface] gap fix.

[satabol]

Windows 11, Blender 3.3.1, Sverchok-master

In node's example of [Marching Squares on Surface] there is a gap:
http://nortikin.github.io/sverchok/docs/nodes/curve/marching_squares_on_surface.html

and:

Recreated scheme of nodes:

https://gist.github.com/a7846c7f7a3c1de01b182ae4981d79c2

In this fix I remove a gap:

[portnov]

I would suggest to make this tolerance value a node setting, instead of hardcoding. Maybe hide it in node's N panel.

[satabol]

@portnov This is not good idea. I found real problem and wrote new fix. The description of problem:

With some T value calculation do errors:

The problem is (4):

u_size = (u_max - u_min)/(samples_u-1)
u_size * (samples_u-1) > (u_max - u_min) (WTF??? - float precision)

and this go to next wrong result:

In this situation need decrease a bit u_size and I propose do this with next code:

This take good result:

If this is not enough then this can be do with (for range) u_size/v_size to calc a value where
u_size * (samples_u-1) <= (u_max - u_min)

What do you think?

[satabol]

[satabol]

Found some glitches. I create for-range with new formula based on:

sverchok/utils/marching_squares.py

Lines 31 to 32 in b7b06eb

	x = min_x + x_size * x0 + x_size/2.0
	y = min_y + y_size * y0 + y_size/2.0

https://gist.github.com/df4d5b173e4934aff740746f28450cf1

[satabol]

There is a connection in current version between these lines of code:

sverchok/nodes/curve/marching_squares_on_surface.py

Lines 116 to 119 in b7b06eb

	u_size = (u_max - u_min)/samples_u
	v_size = (v_max - v_min)/samples_v
	uv_contours, new_edges, _ = make_contours(samples_u, samples_v, u_min, u_size, v_min, v_size, 0, contours, make_faces=True, connect_bounds = self.connect_bounds)

sverchok/utils/marching_squares.py

Lines 31 to 32 in b7b06eb

	x = min_x + x_size * x0 + x_size/2.0
	y = min_y + y_size * y0 + y_size/2.0

x_size = u_size and on last iteration in line 31 result of x has to be u_max but with some T values result exceed u_max. This is a main problem. If refactor of make_contours function and send u_max and v_max as addition parameters and assign its in last [for-range] cycle like:

x = min_x + x_size * x0
y = min_y + y_size * y0

if x0==samples_u-1: x=u_max

if y0==samples_u-1: y=v_max

then problem will solved forever ))) But I have suppose that make_contours is used somewhere. That is why I create compensation formulas.

[portnov]

you can always add new parameter to make_contours if you make it optional.

[satabol]

@portnov Thank you! It work. Can you look at that?

[satabol]

https://gist.github.com/faa14b885442ea6a6388e6bf103faeb5

[portnov]

Looks good to me.