Folks,
Hi,
I want to make some plots and make the plots invisible, that means just i want to save the plots like this:

p = plot(x(*,i),color='blue',thick=2)

p.save, 'plot1.jpg'
but without a window pops up,
Cheers,
Dave

Set the BUFFER keyword.

Cheers,

David

--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thue. ("Perhaps thou speakest truth.")

Cool, i just put BUFFER=1 and it works,
Thank you very much,
Cheers,
Dave

Totally amazing, considering it's a function graphics routine! ;-)

Cheers,

David

--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thue. ("Perhaps thou speakest truth.")

Hi David,

In fact, NG work very well for me, and for most of the people I talk to. We continue to improve them with every IDL release. The biggest issue I've had is that plots with large numbers of points (roughly 1e6) are slow to render, but we've greatly improved this for 8.2.2, due in January.

If you find problems with NG, please let us know. Constructive criticism will help us make them better.

mp

I agree they continue to improve with every release, but I am still too
gun-shy to adopt them for serious work. Sounds like we talk to different
people about function graphics. I run into few people who use them,
and those that do almost uniformly complain about how slow they are.

But, I hear fewer complaints about how they appear on the display. Most of
the problems now appear to be in creating file output from them. Lines
that suddenly grow to 10 times their normal size, and that kind of thing.

I should probably spend some time updating this page on contour
plots for IDL 8.2.1. I'd be curious to see how much things have improved
in both functionality and speed:

   http://www.idlcoyote.com/cg_tips/compcont.php

Cheers,

David

I wrote just a second ago:

This isn't what I meant to write. I run into few people who use them,
but I meant to say I run into a number of people who have tried them
and *they* uniformly complain about how slow they are. The people who
use them don't seem bothered by their speed. But, I would say most of
the people I run into who have tried them don't use them because of
their speed.

Sorry for the confusion.

Cheers,

David

Ok, I give up! I spent a couple of hours trying to
get this plot to work in IDL 8.2.1, and there is
no way I can get the damn colorbar to work! I have
NO idea what I am doing wrong. I am using code
straight out of the documentation and I don't
EVER see a taper on my color bar.

Can anyone help? Just replicate the contour
plot shown in that article in IDL 8.2.1.

Cheers,

David

--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thue. ("Perhaps thou speakest truth.")

Another hour, and still no joy. Here is what
I have so far, if anyone wants to have a crack
at this.

  Code: http://www.idlcoyote.com/misc/compare_contour_plots.pro
  Pic: http://www.idlcoyote.com/misc/idl_8.2.1_contour.png

Issues to resolve:

1. Contour plot doesn't honor POSITION keyword.
2. Contour plot doesn't use correct colors.
3. Only half of box axes in place.
4. Color bar has correct colors, but no taper.
5. Color bar not labeled correctly.

Cheers,

David

--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thue. ("Perhaps thou speakest truth.")

Hello,

I'd like to toss my $1.40 in ....

 From my perspective, the concept of the IDL new graphics functions are
exactly what I was looking for. I can add (or remove) things to them as
I want, and the usual qualitative investigative things one wants to do
with a data plot (zoom, translate, rotate) tools are finally built in
(without having to resort to an iTool). [Ed: Hey, I can jump on the OO
bandwagon just as fast as the next guy...]

And they're relatively easy to use -- although the keyword usage could
be applied more consistently across function (e.g. vertical_alignment
and horizontal_alignment is used for some, whereas for others the latter
is referred to as just "alignment"). I think the IDL folks have
discovered the right path for a better user interface to their object
graphics engine.

Finally, the output just looks much nicer than anything I ever produced
with DG.

Anyone who reads this newsgroup knows there are, uh, issues with certain
aspects of creating graphics in NG (e.g. altogether now: COLORBAR! :o)
but I'm encouraged by the IDL folk's attention to these issues over the
last year or two (e.g. MarkP and ChrisT regularly reply to this
newsgroup, updates are more frequent addressing user concerns, etc).

But the big hairy 800 pound primate in the room is the real time
execution speed of NG -- which must be addressed to make NG be taken
seriously (and by that I mean for day-to-day data visualisation, not the
"produce the final graphic for publication" usage). Until the speed of
the NG plots rivals that of DG, I reckon a lot of folks aren't going to
use them. And people new to IDL will likely ditch it for something
faster. And FWIW, not all of us have the latest greatest hardware out
there - my nearly 3-year old desktop linux box was a low-to-medium-end
machine (i.e. relativly slow) when I got it.

It's tangential, but:
   http://edition.cnn.com/2012/11/12/tech/web/video-loading-study

I really hope the NG speed up in v8.2.2 is at least O(1) to bring it
somewhat closer to DG speeds.

Here endeth my monologue.

cheers,

paulv

On 11/12/12 15:47, Mark Piper wrote:

If the color bar doesn't work, the graphics system
doesn't work, as far as I'm concerned. I mean, come
on, we are trying to do science here!

I'm still not sure it is possible, two and a half
years after this system first came out, to compare
two images side-by-side.

I literally spent three hours yesterday, trying
to create a simple contour plot (which no one
has yet helped me with, probably because it is
impossible to create). So, clearly, these
things are not "intuitive". And, to call the
documentation "abysmal" is to be WAY too
generous.

I really don't know what to make of people telling
me these things "work well". What, exactly, are
they doing with the damn things!? It makes me
feel like an idiot when I can't get one simple
thing to work without literally spending hours and
hours trying to figure it out. You'd think it was
the remote control to the TV the way I struggle
with it! :-(

Cheers,

David

--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thue. ("Perhaps thou speakest truth.")

Seriously? No help whatsoever!?

OK, then I think three conclusions are possible.

1. Despite claims to the contrary, no one outside
   of Exelis uses function graphics.

2. If people do use function graphics, they can't
   figure out how to do this simple contour plot
   example with them.

3. The example is so trivial that no one wants to
   embarrass me further by pointing out my obvious
   inability to understand the very simplest things.

If conclusion 3 is the true conclusion, then I
urge you to come forward anyway. I'm going to
blame everything on Coyote, of course, so no permanent
damage can be done. ;-)

Here is what I have so far:

  Code: http://www.idlcoyote.com/misc/compare_contour_plots.pro
  Pic: http://www.idlcoyote.com/misc/idl_8.2.1_contour.png

Issues to resolve:

1. Contour plot doesn't honor POSITION keyword.
2. Contour plot doesn't use correct colors.
3. Only half of box axes in place.
4. Color bar has correct colors, but no taper.
5. Color bar not labeled correctly.

Cheers,

David

--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thue. ("Perhaps thou speakest truth.")

By the way, the article only seeks to record the
time it takes to render the graphics result, not
the time it takes to write the code! This is a
learning experience, after all. :-)

   http://www.idlcoyote.com/cg_tips/compcont.php

Cheers,

David

--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thue. ("Perhaps thou speakest truth.")

Le mercredi 14 novembre 2012 18:30:25 UTC+1, David Fanning a écrit :

Hi David,
Ten minutes with your example gave me the following:

;-----------------------------------
restore, 'air.nc.data.sav'
t = systime(1)

Loadct, 53, RGB_TABLE=rgb
rgb = reverse(rgb,1)

nlevels = 12
levels = round((max(data, MIN=mindata) - floor(mindata))*findgen(nlevels)/nlevels) + floor(mindata)
levels = [levels[0], levels]

win = Window(Dimension=[640, 512])
win.Refresh, /DISABLE

mp = map('Equirectangular', CENTER_LONGITUDE=180, $
  POSITION=[0.1,0.05,0.95,0.80], $
  LABEL_POSITION = 0, BOX_AXES=1, $
  GRID_LATITUDE = 30, GRID_LONGITUDE = 45, $
  /CURRENT)
mp['Longitudes'].LABEL_ANGLE = 90

cn = contour(data, lon, lat, /OVERPLOT, $
  GRID_UNITS=2, MAP_PROJECTION='Equirectangular', $
  RGB_TABLE=rgb, /CURRENT, $
  C_VALUE=levels, /FILL)

cb = COLORBAR(POSITION=[0.1, 0.91, 0.95, 0.95], $
    BORDER=1, TARGET=cn, TICKVALUES=levels)

m = MapContinents(COLOR=!Color.red)

win.Refresh

print,'elaps:', systime(1) - t

end
;------------------------------------------

looking to solve most of the issues you pointed out.
I do not understand the '0°E' appearing twice along X-axis. A NG bug ? Who knows ?

alain.

Ah. I would have thought the word "duplicate" came
from the French, but apparently not. ;-)

OK, I have completely given up on getting an exact
duplication of the plot I want, but it has to be
considerably closer than this example!

Robert Vincent and I spent another couple of hours
working on this together last night. Although Robert
had decided to give up on function graphics by the end
of the night, I'm dumb enough to keep plodding onward.

Anyway, together we came up with something that was
showing some promise. Here is a comparison between
the plot we are trying to duplicate on the left,
and the plot we have come up with on the right:

   http://www.idlcoyote.com/misc/fg_map_side_by_side.png

It took nearly forever to figure out the color bar,
but we finally got it.

There are obvious problems. The FG contour plot is
still VERY small in the window. We haven't been able
to figure out how to make it bigger. It seems to
ignore the POSITION keyword we are passing to it.
The box axes are a total joke. You can only label
them on two sides, and even then, you can't label
them as normal box axes.

But, still, it was looking pretty good. At least
it was until I took a close look at the lines that
we overplot onto the filled contours. You have to
blow up the plot to see this really well, so here
is a blow up of the FG contour plot:

   http://www.idlcoyote.com/misc/fg_map_blowup.png

You will notice that the overplotted lines don't
line up AT ALL with the filled contours!!!!

This, despite the fact that the two contour commands
we use to create the filled and outline contours are
IDENTICAL as far as we can tell!

Here is the code we are currently using:

   http://www.idlcoyote.com/misc/fg_map_contour.pro

We welcome any and all to tell us what we are doing
wrong.

Cheers,

David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thue. ("Perhaps thou speakest truth.")

Le jeudi 15 novembre 2012 14:44:52 UTC+1, David Fanning a écrit :

You should have read more carefully and executed the code I sent previously. It is pretty close of what you want. The trick is to draw the map first, then to overplot the filled and line contours. A slightly modified version follows:

;-----------------------------------------
restore, 'air.nc.data.sav'
t = systime(1)

Loadct, 53, RGB_TABLE=rgb
rgb = reverse(rgb,1)

nlevels = 12
levels = round((max(data, MIN=mindata) - floor(mindata))*findgen(nlevels)/nlevels) + floor(mindata)
levels = [levels[0], levels]

win = Window(Dimension=[640, 512])
win.Refresh, /DISABLE

mp = map('Equirectangular', CENTER_LONGITUDE=180, $
  POSITION=[0.1,0.05,0.95,0.80], $
  LABEL_POSITION = 0, BOX_AXES=1, $
  GRID_LATITUDE = 30, GRID_LONGITUDE = 45, $
  /CURRENT, ASPECT_RATIO=0)
mp['Longitudes'].LABEL_ANGLE = 90

cn = contour(data, lon, lat, /OVERPLOT, $
  GRID_UNITS=2, MAP_PROJECTION='Equirectangular', $
  RGB_TABLE=rgb, /CURRENT, $
  C_VALUE=levels, /FILL)

cn1 = contour(data, lon, lat, /OVERPLOT, $
  GRID_UNITS=2, MAP_PROJECTION='Equirectangular', $
  RGB_TABLE=rgb, /CURRENT, $
  C_VALUE=levels, C_COLOR=!Color.White)

cb = COLORBAR(POSITION=[0.1, 0.91, 0.95, 0.95], $
    BORDER=1, TARGET=cn, TICKVALUES=levels, TITLE='Temperature °K')

m = MapContinents(COLOR=!Color.red)

win.Refresh

print,'elaps:', systime(1) - t

end
;----------------------------------------------

There is no particular difficulty nor surprise, except the '0°E' extra label.

alain.

OK, close. I think this is nearly what we want, except
for those extraneous labels. Here is a picture of
the final output:

   http://www.idlcoyote.com/misc/fg_map_final.png

Yes, I would have probably discovered this trick for
myself, had I spent another 8-10 hours on the problem!
I certainly wouldn't have discovered it in the
documentation.

Thanks for your help!

David

;****************************************************
PRO FG_Map_Contour

  restore, 'air.nc.data.sav'
  nlevels = 12
  xrange = [Min(lon), Max(lon)]
  yrange = [Min(lat), Max(lat)]
  center_lon = (xrange[1]-xrange[0])/2.0 + xrange[0]
  nlevels = 12
  levels = cgConLevels(Float(data), NLevels=nlevels+1, $
       MinValue=Floor(Min(data)), STEP=step, Factor=1)
  mylat = 40.6  ; Latitude of Fort Collins, Colorado.
  mylon = 254.9 ; Longitude of Fort Collins, Colorado.

  cgLoadCT, 2, /Reverse, /Brewer, NColors=nlevels, RGB_Table=rgb
  rgb[11,*] = [255, 255, 255]

  win = WINDOW(WINDOW_TITLE = 'Function Graphics', $
    dimensions = [700, 600])

  win.Refresh, /Disable

mp = map('Equirectangular', CENTER_LONGITUDE=180, $
  POSITION=[0.1,0.1,0.90,0.80], $
  LABEL_POSITION = 0, BOX_AXES=1, $
  GRID_LATITUDE = 30, GRID_LONGITUDE = 45, $
  /CURRENT, ASPECT_RATIO=0)
mp['Longitudes'].LABEL_ANGLE = 90

cn = contour(data, lon, lat, /OVERPLOT, $
  GRID_UNITS=2, MAP_PROJECTION='Equirectangular', $
  RGB_TABLE=rgb, /CURRENT, RGB_INDICES=Indgen(nlevels), $
  C_VALUE=levels, /FILL)

cn1 = contour(data, lon, lat, /OVERPLOT, $
  GRID_UNITS=2, MAP_PROJECTION='Equirectangular', $
  RGB_TABLE=rgb, /CURRENT, $
  C_VALUE=levels, C_COLOR=!Color.White)

  c = MapContinents(Color=cgColor('tomato', /Triple, /Row))

  cb = Colorbar(Tickname=levels, RGB_TABLE=rgb, Range=[Min(levels), $
    Max(levels)], $
    Major=11, /Border_On, Title='Temperature $\deg$K', $
    Position=[0.1, 0.88, 0.9, 0.93], Minor=0, TAPER=3)

  s = Symbol(mylon, mylat, /Data, /Current, 'Star', $
     /Sym_Filled, Sym_Color='red')

  win.Refresh
END
;****************************************************

--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thue. ("Perhaps thou speakest truth.")

Alright. After all is said and done, the function graphics
output, if not the programming time, *is* getting faster.

I've updated the original article:

  http://www.idlcoyote.com/cg_tips/compcont.php

Cheers,

David

--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thue. ("Perhaps thou speakest truth.")