plot -Sf gap - unwanted size change #2995

KristofKoch · 2020-03-30T19:54:03Z

Description of the problem

When using -Sf the size of the gap between the symbols changes depending on supplied path.

Two identical fronts are used for illustration, boxes are 80p in size with a 160p spacing.
The right (blue) front is shifted 80p to create an alternating pattern.
The paths both run from bottom right to upper left.
In both examples the plot call is absolutely identical, only the supplied path changes.

Full script that generated the error

cat > path.txt << END
 1.3 -1.5
-1.8  1.2
END

cat > path2.txt << END
 1.3 -1.5
 0.2  0.4
-0.8  0.1
-1.8  1.2
END

gmt begin small_gap
  gmt basemap -R-2/2/-2/2 -JM6i -Bafg1
  gmt plot path.txt -W0p -Sf160p/80p+b+l+pred
  gmt plot path.txt -W0p -Sf160p/80p+b+r+pblue+o80p
  gmt plot path.txt -Skcrosshair/0.5c -Wthinnest,black
gmt end show

gmt begin large_gap
  gmt basemap -R-2/2/-2/2 -JM6i -Bafg1
  gmt plot path2.txt -W0p -Sf160p/80p+b+l+pred
  gmt plot path2.txt -W0p -Sf160p/80p+b+r+pblue+o80p
  gmt plot path2.txt -Skcrosshair/0.5c -Wthinnest,black
gmt end show

Actual outcome

The two different paths lead to two different gap sizes all other things being the same.

Expected outcome

I expected the gaps to be as wide as the boxes and not changing size depending on supplied path in an unpredictable manner.

System information

Operating system: macOS
GMT version: a6f3e31

The text was updated successfully, but these errors were encountered:

KristofKoch · 2020-03-30T21:05:47Z

Some further experimentation lead me to believe that this might actually be an intentional feature. By sightly nudging the gap size the line terminates on a symbol making for a nice line end. This is hardly noticeable at small symbol sizes but gets very obvious at larger sizes as shown above.

cat > path.txt << END
 1.3 -1.5
-1.8  1.2
END

gmt begin small_gap
  gmt basemap -R-2/2/-2/2 -JM6i -Bafg1
  gmt plot path.txt -W0p -Sf160p/80p+b+l+pred
  gmt plot path.txt -W0p -Sf160p/80p+b+r+pblue+o80p
  gmt plot path.txt -W0p -Sf80p/40p+b+l+pred
  gmt plot path.txt -W0p -Sf80p/40p+b+r+pblue+o40p
  gmt plot path.txt -W0p -Sf16p/8p+b+l+pred
  gmt plot path.txt -W0p -Sf16p/8p+b+r+pblue+o8p
  gmt plot path.txt -W0p -Sf8p/4p+b+l+pred
  gmt plot path.txt -W0p -Sf8p/4p+b+r+pblue+o4p
  gmt plot path.txt -Skcrosshair/0.5c -Wthinnest,black
gmt end show

See this heavily zoomed in crop of above test:

The uneven gap is always there, it just get less and less noticeable the smaller the symbol gets.

I would have never noticed this if not for my attempt for a checkered line. Can someone confirm my suspicions?

PaulWessel · 2020-03-30T22:33:59Z

Remember that for geographic data, the default behavior is to resample crude paths with great-circle resampling. This can be modified or turned off with -A. So here, the paths you are prescribing are not the ones being used inside plot. Please try with -A and see if there are any issues then,

KristofKoch · 2020-03-31T05:56:33Z

Hi @PaulWessel, I tried again with -A as suggested but same result. In my understanding there should be not much difference between a great circle and a crude path with this projection and region?

cat > path.txt << END
 1.3 -1.5
-1.8  1.2
END

gmt begin small_gap_A
  gmt basemap -R-2/2/-2/2 -JM6i -Bafg1
  gmt plot path.txt -W0p -Sf160p/80p+b+l+pred -A
  gmt plot path.txt -W0p -Sf160p/80p+b+r+pblue+o80p -A
  gmt plot path.txt -Skcrosshair/0.5c -Wthinnest,black
gmt end show

Same result when using -A:

PaulWessel · 2020-03-31T05:58:00Z

Thanks, @KristofKoch. If you want to be sure you can plot your points as well with and without -A and see if they differ.

KristofKoch · 2020-03-31T06:05:26Z

I have the feeling I'm not understanding your argument but I do not see how the points could be influenced by -A?

cat > path.txt << END
 1.3 -1.5
-1.8  1.2
END

gmt begin small_gap_A
	gmt basemap -R-2/2/-2/2 -JM6i -Bafg1
  gmt plot path.txt -W0p -Sf160p/80p+b+l+pred -A
  gmt plot path.txt -W0p -Sf160p/80p+b+r+pblue+o80p -A
  gmt plot path.txt -Skcrosshair/0.5c -Wthinnest,black
  gmt plot path.txt -Skcrosshair/0.5c -W0p,pink -A
gmt end show

Crop from resulting plot - they match perfectly

PaulWessel · 2020-03-31T08:03:49Z

I mean where the points plot with vs without -A.

KristofKoch · 2020-03-31T08:18:41Z

My apologies @PaulWessel, but I think I still don't understand it. You mean comparing the placement of the two points from the textfile? This is what I tried with the exaggerated bullseye above: black without -A and the faint pink superimposed one with -A.

PaulWessel · 2020-03-31T09:01:17Z

Looking at the source code for gmt_draw_front:

if (f->f_gap > 0.0) {	/* Gave positive interval; adjust so we start and end with a tick on each line */
	ngap = irint (s[n-1] / f->f_gap);
	gap = s[n-1] / ngap;
	dist = f->f_off;	/* Start off at the offset distance [0] */
	ngap++;
}

So looks like we are marching along the line in steps of gap, which can differ from the given gap (f_gap) and that difference is likely to be larger fro fewer gaps.

KristofKoch · 2020-03-31T10:56:12Z

So my suspicions were right – the user supplied gap is more of a "recommendation". As far as I understand gmt_plot.c (no real C experience) there is no way to easily force an exact gap. Thank you for pointing that out, @PaulWessel

Is there a way to suppress the plotting of the line? Something like -W- then I would be able to get a wider spacing and wash over the "flexible" gap and have more leeway to toy around with --PS_LINE_CAP to generate different line designs.

KristofKoch · 2020-03-31T12:17:49Z

To highlight my point why a not plotted line would be beneficial (for me at least) see the following example. All styles without a distinct centerline have a problem with the (very faint) line.

PaulWessel · 2020-03-31T19:13:59Z

Thanks @KristofKoch, I will look at the front symbol option for improvements. It should be possible to not draw the line. You are probably aware of the decorated symbol option as well (-S~); is that of interest to you as well?

KristofKoch · 2020-03-31T19:52:36Z

For some inexplicable reason I wasn't aware of -S~.

head -> desk (repeatedly)

This promises to solve quite a few headaches (not the head banging – the -S~ option). Even my dashed line with arrow heads problem.

Thank you, @PaulWessel - now I have a lot to test and experiment. And thank you for looking into -Sfas well.

If the front gap is specified with an explicity plus-symbol then we take taht to mean the gap should be used as is. The default remains to distribute ticks exactly along the liine and thus adjust the gap accordingly. See #2995 for context.

* Let positive front gaps be used as given If the front gap is specified with an explicity plus-symbol then we take taht to mean the gap should be used as is. The default remains to distribute ticks exactly along the liine and thus adjust the gap accordingly. See #2995 for context. * Update gmt_init.c

PaulWessel · 2020-04-02T23:07:41Z

Hi @KristofKoch, I will close this one for now. Feel free to open an issue about allowing the lines in quoted, decorated, and front to be invisible.

PaulWessel mentioned this issue Apr 1, 2020

Let positively defined front gaps be used exactly as given #3006

Merged

seisman added this to the 6.1.0 milestone Apr 1, 2020

PaulWessel closed this as completed Apr 2, 2020

KristofKoch mentioned this issue Apr 3, 2020

Allow the lines in quoted, decorated, and front to be invisible. #3017

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

plot -Sf gap - unwanted size change #2995

plot -Sf gap - unwanted size change #2995

KristofKoch commented Mar 30, 2020

KristofKoch commented Mar 30, 2020

PaulWessel commented Mar 30, 2020

KristofKoch commented Mar 31, 2020

PaulWessel commented Mar 31, 2020

KristofKoch commented Mar 31, 2020

PaulWessel commented Mar 31, 2020

KristofKoch commented Mar 31, 2020

PaulWessel commented Mar 31, 2020

KristofKoch commented Mar 31, 2020 •

edited

Loading

KristofKoch commented Mar 31, 2020

PaulWessel commented Mar 31, 2020

KristofKoch commented Mar 31, 2020

PaulWessel commented Apr 2, 2020

plot -Sf gap - unwanted size change #2995

plot -Sf gap - unwanted size change #2995

Comments

KristofKoch commented Mar 30, 2020

KristofKoch commented Mar 30, 2020

PaulWessel commented Mar 30, 2020

KristofKoch commented Mar 31, 2020

PaulWessel commented Mar 31, 2020

KristofKoch commented Mar 31, 2020

PaulWessel commented Mar 31, 2020

KristofKoch commented Mar 31, 2020

PaulWessel commented Mar 31, 2020

KristofKoch commented Mar 31, 2020 • edited Loading

KristofKoch commented Mar 31, 2020

PaulWessel commented Mar 31, 2020

KristofKoch commented Mar 31, 2020

PaulWessel commented Apr 2, 2020

KristofKoch commented Mar 31, 2020 •

edited

Loading