search

GenericMappingTools / gmt

The Generic Mapping Tools
https://www.generic-mapping-tools.org
Other
858 stars 359 forks source link

Double frame on grdview perspective plot #4181

Closed weiji14 closed 4 years ago

weiji14 commented 4 years ago

Description of the problem

There appears to be a double-frame on the x and y-axis when making a perspective plot using grdview with the -Jz flag. Originally detected at https://github.com/GenericMappingTools/pygmt/pull/589#discussion_r484320626.

Actual (incorrect) plot Expected (correct) plot
test_grdview_with_perspective_and_zaxis_frame-expected test_grdview_with_perspective_and_zaxis_frame

Full script that generated the error

gmt begin doubleframe png
    gmt grdcut @earth_relief_01d_g -R-116/-109/-47/-44 -Gtempgrid.nc
    gmt grdview tempgrid.nc -p225/30 -Jz0.005 -Bzaf -Vd
gmt end

Full error message

``` gmt [DEBUG]: Obtained the ppid from parent: 176 gmt [DEBUG]: Enter: gmtinit_new_GMT_ctrl gmt [DEBUG]: GMT->session.SHAREDIR = /srv/conda/envs/notebook/share/gmt gmt [DEBUG]: GMT->session.HOMEDIR = /home/jovyan gmt [DEBUG]: GMT->session.USERDIR = /home/jovyan/.gmt [created] gmt [DEBUG]: GMT->session.CACHEDIR = /home/jovyan/.gmt/cache [created] gmt [DEBUG]: GMT: 0. Will try to find subdir=postscriptlight stem = PSL_custom_fonts suffix=.txt gmt [DEBUG]: GMT: 1. gmt_getsharepath trying current dir gmt [DEBUG]: GMT: 2. gmt_getsharepath trying USERDIR /home/jovyan/.gmt gmt [DEBUG]: GMT: 3. gmt_getsharepath trying USERDIR subdir /home/jovyan/.gmt/postscriptlight gmt [DEBUG]: GMT: 4. gmt_getsharepath trying SHAREDIR subdir /srv/conda/envs/notebook/share/gmt/postscriptlight gmt [DEBUG]: GMT: 5. gmt_getsharepath trying SHAREDIR /srv/conda/envs/notebook/share/gmt gmt [DEBUG]: GMT: 6. gmt_getsharepath failed gmt [DEBUG]: Map distance calculation will be Cartesian gmt [DEBUG]: Exit: gmtinit_new_GMT_ctrl gmt [DEBUG]: Enter: New_PSL_Ctrl gmt [DEBUG]: Exit: New_PSL_Ctrl gmt [DEBUG]: Enter: gmt_manage_workflow gmt [DEBUG]: GMT now running in modern mode [Session ID = 176] gmt [DEBUG]: Exit : gmt_manage_workflow gmt [DEBUG]: Enter: PSL_beginsession gmt [DEBUG]: Exit : PSL_beginsession gmt [DEBUG]: Enter: PSL_setdefaults gmt [DEBUG]: Exit : PSL_setdefaults gmt [DEBUG]: Enter: gmtlib_io_init gmt [DEBUG]: Exit : gmtlib_io_init gmt [DEBUG]: Enter: gmt_hash_init gmt [DEBUG]: Exit: gmt_hash_init gmt [DEBUG]: Enter: gmt_hash_init gmt [DEBUG]: Exit: gmt_hash_init gmt [DEBUG]: Enter: gmt_reload_settings gmt [DEBUG]: The PROJ_GEODESIC set to Vincenty gmt [DEBUG]: gmtlib_get_graphics_item: Fig: 0 Subplot: 2 Panel: () Inset: 0 gmt [DEBUG]: Reading GMT Default parameters from file: /home/jovyan/.gmt/sessions/gmt_session.176/gmt.conf gmt [DEBUG]: Exit: gmt_reload_settings gmt [DEBUG]: Enter: gmtlib_plot_C_format gmt [DEBUG]: Exit: gmtlib_plot_C_format gmt [DEBUG]: Enter: gmtinit_get_history gmt [DEBUG]: gmtlib_get_graphics_item: Fig: 0 Subplot: 2 Panel: () Inset: 0 gmt [DEBUG]: Enter: gmt_hash_init gmt [DEBUG]: Exit: gmt_hash_init gmt [DEBUG]: Exit: gmtinit_get_history gmt [DEBUG]: Initialize FFTW with 6 threads. gmt [DEBUG]: GMT_Create_Session initialized GMT structure gmt [DEBUG]: Loading core GMT shared library: libgmt.so gmt [DEBUG]: Shared Library # 0 (core). Path = libgmt.so gmt [DEBUG]: Loading GMT plugins from: /srv/conda/envs/notebook/lib/gmt/plugins gmt [DEBUG]: Shared Library # 1 (supplements). Path = /srv/conda/envs/notebook/lib/gmt/plugins/supplements.so gmt [DEBUG]: GMT now running in modern mode [Session ID = 176] gmt [DEBUG]: Use PS filename /home/jovyan/.gmt/sessions/gmt_session.176/gmt_0.ps- gmt [DEBUG]: gmtinit_get_current_panel: No current panel selected so not in subplot mode gmt [DEBUG]: Modern mode: Added -R to options since history is available. gmt [DEBUG]: Modern: Adding -JX15c to options since there is no history available. gmt [DEBUG]: Revised options: tempgrid.nc -p225/30 -Jz0.005 -Bzaf -Vd -R -JX15c grdview [DEBUG]: History: Process -p225/30 grdview [DEBUG]: History: Process -Jz0.005 grdview [DEBUG]: History: Process -R grdview [DEBUG]: History: Process -JX15c grdview [DEBUG]: gmt_get_filename: In: -116/-109/-47/-44 Out: -116/-109/-47/-44 grdview [DEBUG]: Look for file -116/-109/-47/-44 in /home/jovyan/.gmt grdview [DEBUG]: Look for file -116/-109/-47/-44 in /home/jovyan/.gmt/cache grdview [DEBUG]: Look for file -116/-109/-47/-44 in /home/jovyan/.gmt/server grdview [DEBUG]: Got regular w/e/s/n for region (-116/-109/-47/-44) grdview [DEBUG]: gmt_get_filename: In: tempgrid.nc Out: tempgrid.nc grdview [DEBUG]: Found readable file tempgrid.nc grdview [DEBUG]: Replace file tempgrid.nc with path tempgrid.nc grdview [DEBUG]: Replace file tempgrid.nc with tempgrid.nc grdview [DEBUG]: Local file /home/jovyan/.gmt/server/gmt_data_server.txt found grdview [DEBUG]: File /home/jovyan/.gmt/server/gmt_data_server.txt less than 24 hours old, refresh is premature. grdview [DEBUG]: Load contents from /home/jovyan/.gmt/server/gmt_data_server.txt grdview [DEBUG]: Local file /home/jovyan/.gmt/server/gmt_hash_server.txt found grdview [DEBUG]: File /home/jovyan/.gmt/server/gmt_hash_server.txt less than 24 hours old, refresh is premature. grdview [DEBUG]: gmt_get_filename: In: tempgrid.nc Out: tempgrid.nc grdview [DEBUG]: Found readable file tempgrid.nc grdview [DEBUG]: Replace file tempgrid.nc with path tempgrid.nc grdview [DEBUG]: gmt_get_filename: In: tempgrid.nc Out: tempgrid.nc grdview [DEBUG]: Found readable file tempgrid.nc grdview [DEBUG]: Replace file tempgrid.nc with path tempgrid.nc grdview [DEBUG]: gmt_get_filename: In: tempgrid.nc Out: tempgrid.nc grdview [DEBUG]: gmt_get_filename: In: tempgrid.nc Out: tempgrid.nc grdview [DEBUG]: Found readable file tempgrid.nc grdview [DEBUG]: Object ID 0 : Registered Grid File tempgrid.nc as an Input resource with geometry Surface [n_objects = 1] grdview [DEBUG]: gmtapi_begin_io: Input resource access is now enabled [container] grdview [DEBUG]: gmtapi_import_grid: Passed ID = 0 and mode = 1 grdview [DEBUG]: Found readable file tempgrid.nc grdview [DEBUG]: gmt_get_filename: In: tempgrid.nc Out: tempgrid.nc grdview [DEBUG]: Call gmtgrdio_doctor_geo_increments on a geographic grid grdview [DEBUG]: Geographic input grid, longitudes span less than 360 grdview [DEBUG]: GMT_End_IO: Input resource access is now disabled grdview [DEBUG]: Projected values in meters: -3.5 3.5 -47 -44 grdview [DEBUG]: Auto-frame interval for axis 2 item 0: d = 200 f = 100 grdview [INFORMATION]: Auto-frame interval for z-axis (item 0): a200f100 grdview [INFORMATION]: Map scale is 0.000466667 km per cm or 1:46.6667. grdview [INFORMATION]: Processing shape grid grdview [DEBUG]: gmtapi_begin_io: Input resource access is now enabled [container] grdview [DEBUG]: gmtapi_import_grid: Passed ID = 0 and mode = 2 grdview [INFORMATION]: Reading grid from file tempgrid.nc grdview [DEBUG]: packed z-range: [-3606.5,-2614.5] grdview [DEBUG]: Geographic input grid, longitudes span less than 360 grdview [DEBUG]: Chosen boundary condition for all edges: geographic grdview [INFORMATION]: Set boundary condition for all edges: natural grdview [INFORMATION]: Set boundary condition for left edge: natural grdview [INFORMATION]: Set boundary condition for right edge: natural grdview [INFORMATION]: Set boundary condition for bottom edge: natural grdview [INFORMATION]: Set boundary condition for top edge: natural grdview [DEBUG]: GMT_End_IO: Input resource access is now disabled grdview [DEBUG]: Octant 4 (az = 153.435) one = 1 grdview [DEBUG]: Outer loop over x doing 6:-1:-1 grdview [DEBUG]: Inner loop over y doing 1:1:4 grdview [INFORMATION]: Start creating PostScript plot grdview [DEBUG]: Running in PS mode modern grdview [DEBUG]: Use PS filename /home/jovyan/.gmt/sessions/gmt_session.176/gmt_0.ps- grdview [DEBUG]: Create hidden PS file /home/jovyan/.gmt/sessions/gmt_session.176/gmt_0.ps- grdview [DEBUG]: No figure file /home/jovyan/.gmt/sessions/gmt_session.176/gmt.figures - nothing to do grdview [DEBUG]: Got session name as doubleframe and default graphics formats as png grdview [INFORMATION]: Do mesh plot with mesh color white grdview [DEBUG]: Current size of half-baked PS file /home/jovyan/.gmt/sessions/gmt_session.176/gmt_0.ps- = 22543. grdview [DEBUG]: gmtlib_garbage_collection: Destroying object: C=0 A=0 ID=0 W=Input F=Grid M=File S=Used P=56499f4a5350 N=tempgrid.nc grdview [DEBUG]: GMTAPI_Garbage_Collection freed 1 memory objects grdview [DEBUG]: gmtlib_unregister_io: Unregistering object no 0 [n_objects = 0] grdview (gmtlib_free_tmp_arrays): tried to free unallocated memory gmt [DEBUG]: Entering GMT_Destroy_Session gmt [DEBUG]: gmtlib_get_graphics_item: Fig: 0 Subplot: 2 Panel: () Inset: 0 ```

System information

I've reproduced this both on my personal computer and on the try-gmt.

joa-quim commented 4 years ago

You can avoid it with

gmt grdview @earth_relief_01d_g -R-116/-109/-47/-44 -p225/30 -JZ5 -Bzaf --MAP_FRAME_TYPE=plain -pdf map
seisman commented 4 years ago

The "double frame" is actually the frame for geographic projections (i.e., the fancy frame). If you add -Bxaf -Byaf to the command, it would be more clear:

image

The debug message from @weiji14 says:

gmt [DEBUG]: Revised options: tempgrid.nc -p225/30 -Jz0.005 -Bzaf -Vd -R -JX15c

it's unclear to me why GMT chooses -JX15c for a geographic grid.

joa-quim commented 4 years ago

it's unclear to me why GMT chooses -JX15c for a geographic grid.

It's the poor man's choice. Alternative would be to give the right projection for the given region. Doable but never done.

weiji14 commented 4 years ago

Ok, put it down as user error for now (setting a projection like -JT-112.5/45.5 works). Still think the zebra stripes should show up properly though instead of having a double lined frame, or that we just plot a single lined frame.

joa-quim commented 4 years ago

Agree. It's a mild bug

PaulWessel commented 4 years ago

Not sure what you expect here. There is only -Bzaf given, right? So nothing for x-y. It cannot draw checker-board without knowing or setting increments, but no -B was given. So what it did was basically -B0. When that happens I totally agree that the double lines are not helpful, so it could switch to a plain frame in that case when no increments are set.

seisman commented 4 years ago

I actually expect no X and Y frames here, because not -Bxaf and -Byaf are given.

BTW, adding -B0 to the grdview command gives me a plain frame.

PaulWessel commented 4 years ago

Yes, I can go with that, I was just assuming we have to do some backwards compatibility here by drawing that -B0. But I would be fine and happier with just the z-axis as requested.

seisman commented 4 years ago

Ping @weiji14.

weiji14 commented 4 years ago

I think it would look weird with just the z-axis plotted and no x and y frames (maybe someone would plot it that way?), but happy to go with it (just z-axis plotted). Anything is better than the strange double frame really.

joa-quim commented 4 years ago

Remember that we have MAP_FRAME_AXES defaulting to WSEN so this case has to plot those axes but in plain. In the unlikely case the user won't them, he can always do --MAP_FRAME_AXES=Z

PaulWessel commented 4 years ago

Or -B+n I think.

seisman commented 4 years ago

Or -B+n I think.

No, -Bzaf -B+n doesn't plot the Z axis.

Remember that we have MAP_FRAME_AXES defaulting to WSEN so this case has to plot those axes but in plain. In the unlikely case the user won't them, he can always do --MAP_FRAME_AXES=Z

For 2D maps, the frames are not plotted unless -B, -Bx or -By is given. However, for the 3D case, giving -Bz only also plots the X and Y axes seems weird to me.

PaulWessel commented 4 years ago

Yes, it breaks the logic. -Bz should not magically also turn on horizontal frames I think. That is the error here.

PaulWessel commented 4 years ago

Based on experiments yesterday, I think we need to add a bool that keeps track if -Bx and|or -By (typically via plain -B) is set during parsing. Only if that is true shall we plot the x-y basemap. Right now there is various confusions. For instance, test/psxyz/filler.sh (which passes) uses -Bwesn and it somehow draws the frame (just like -B0) would do. But not sure if specifying axes and format shall automatically imply -B0? We have a system default of WESN which kicks in if you specify intervals with -B. So in that sense -Bwesn just overrides that default but should not result in any frame. Agreed?

joa-quim commented 4 years ago

Yes, -Bwesn should not result in any frame.

PaulWessel commented 4 years ago

What should -B0 result in? When I replaced -Bwesn in fillter.sh with -B0 I get an empty 3-D stick for the z-axis. I think that is why I did -Bwesn so I could do -B0 -Bwesn, but we should define what -B0 means, i.e., should it imply anything about the z-axis.

joa-quim commented 4 years ago

-B0 is a xy frame with no annots, no ticks and no grid, right?

PaulWessel commented 4 years ago

I am totally confident in that is what we meant, yes. But

gmt basemap -R0/5/0/5/0/1 -JX3i -JZ1i -B0 -p170/35 -pdf map

gives that z-stick.

PaulWessel commented 4 years ago

I am learning that if I turn the z-axis off then the map title is plotted as 2-D instead of in-plane 3-D. So while the above command no longer has a stick. Compare these:

In master, running gmt basemap -R0/5/0/5/0/1 -JX3i -JZ1i -B0 -B+tTITLE -p170/35 -png master gives

master

while in my branch that lets -B0 only set x/y frame the same command looks like this:

branch

In that branch, I can get the first plot by adding -Bz0. As you can see, a true 3-D plot places a 2-D title while a perspective 2-D plot gets a title in the same perspective. Might also want to document taht, but I think the branch is better, no? What does @seisman think? I should add that perhaps the real problem is that -JZ is being used but nothing is plotted in z. So perhaps master is correct in that sense. Yet, if I take out -JZ and run master I get a 2-D title:

master

joa-quim commented 4 years ago

Having the z stick with -JZ1i -B0 seems correct because there is a -JZ. The title is more confuse. Middle plot seems more correct.

PaulWessel commented 4 years ago

I agree. So we have these rules:

  1. A 3-D plot is one where -p and -Jz|Z are given and they imply a 3-D basemap and a conventional flat 2-D title
  2. A perspective 2-D plot has -p but no -Jz|Z and implies a 2-D basemap and a perspective 2-D title
  3. The -B0 always apply to the x-y frame but also applies to the z-axis if a true 3-D plot.
joa-quim commented 4 years ago

Yes, seems to cover all cases.

seisman commented 4 years ago

4293 still doesn't fix the issue, right?

seisman commented 4 years ago

Trying the latest master branch.

The following result doesn't make sense. Here I give -Bx -By, but the Y annotations and ticks are not shown.

gmt basemap -R0/5/0/5/0/1 -JX3i -JZ1i -B+tTITLE -p170/35 -png map -Bx -By -Bz

image

PaulWessel commented 4 years ago

So in the highest level of parsing I know we look (in modern mode only) for options like plain -B and we append to make it -Baf. I probably did not check for them all individually so will do that later today).

PaulWessel commented 4 years ago

No, I have checks for those cases and they should all end up as -Bxaf -Byaf and -Bzaf. So then the question is what happens next in the auto-interval section.

PaulWessel commented 4 years ago

After -Bx is parsed, GMT->current.map.frame.set is true and -By is not processed, only -Bz. I think set needs to be set[GMT_X|Y|Z] so we have more control here.

PaulWessel commented 4 years ago

Note: This has been fixed in PR #4274 and will go away once that branch is merged.

PaulWessel commented 4 years ago

Branched merged so closing this issue.