Antoine De Pauw wrote: > Jeff, > > I totally agree this is due to missing values > > Again I've got difficulties to find good words so forgive me, what I tried > to say is that the ability to have that border transparent would be a good > feature in next releases, for people who need to interpolate and plot such > data and have an aesthetic result >
Antoine: First of all, let's not call it a 'border' - it happens to show up around the border of the image in your case, but fundamentally it's just missing pixels in the image. The missing pixels are already transparent - what you're seeing is the axes background showing through the missing values. > Imshow is the ideal candidate for satellite data as it has some nice > interpolation features and it is fast, so it can be batch-run on the server > every time we receive data, without too much computation time > > The alternative I'm using now is a double or quadruple size grid to reduce > the width of that border, with background color set to the lower colormap > color > > That way, the border is really hard to see and it makes (almost) quality > plots for publications > I think you're already doing everything that can be done - aside from making the plot region smaller so it's inside the convex hull of the data and there are no missing values. I don't see how changing the behavior of imshow would help you any further. -Jeff > -----Original Message----- > From: Jeff Whitaker [mailto:[EMAIL PROTECTED] > Sent: Thursday, 25 September, 2008 15:34 > To: De Pauw Antoine > Cc: 'Matplotlib Users' > Subject: Re: [Matplotlib-users] Information request > > De Pauw Antoine wrote: > >> Jeff, >> >> Thanks for the tip, it's now working perfectly >> >> However, there's still that border with the imshow plot, and I think it >> would be good to have it transparent >> >> There's a zoomed picture I made: >> http://img218.imageshack.us/img218/5833/imshowborderxz9.png >> >> You see the shadow around the data... >> >> It would be nice for next releases of Matplotlib to get rid of that, but >> > I'm > >> not able to patch it myself or so... I know there's still a lot of work >> > with > >> the lib but keep the good work, it is really fantastic >> >> Thanks for your help! >> >> Antoine De Pauw >> Collaborateur de recherches, Informatique - Research collaborator, IT >> Laboratoire de chimie quantique et photophysique - Quantum chemistry and >> photophysics laboratory >> Université Libre de Bruxelles - ULB >> >> > > Antoine: I thought we agreed that it's not an imshow bug - but rather > due to the griddata gridding procedure returning missing values outside > the convex hull of the input data. Do you disagree? I see no such border > around an imshow plot that contains no missing values. If you shrink the > size of the map plotting region so it's fully within the convex hull of > the data, the border disappears. > > -Jeff > >> -----Original Message----- >> From: Jeff Whitaker [mailto:[EMAIL PROTECTED] >> Sent: jeudi 25 septembre 2008 14:15 >> To: De Pauw Antoine >> Cc: 'Matplotlib Users' >> Subject: Re: [Matplotlib-users] Information request >> >> De Pauw Antoine wrote: >> >> >>> Hi Jeff, >>> >>> I finally found out how to fill my figure with a background color using >>> axes.set_axis_bgcolor(color), but I'm facing the following problem now: >>> >>> How could I get the lower color of a colormap? This is quite undocumented >>> and I don’t know the colormap properties I could use for that >>> >>> I know there must be an accessible value somewhere, like for the >>> ax.get_yticklabels() you gave me >>> >>> If someone had the clue, my problems would then be completely solved >>> >>> Antoine De Pauw >>> Collaborateur de recherches, Informatique - Research collaborator, IT >>> Laboratoire de chimie quantique et photophysique - Quantum chemistry and >>> photophysics laboratory >>> Université Libre de Bruxelles - ULB >>> >>> >>> >> Antoine: To get the RGBA value associated with a particular data value, >> just call the colormap as a function as pass it that value. For example >> >> >>> import matplotlib.pyplot as plt >> >>> plt.cm.jet(1) >> (0.0, 0.0, 0.517825311942959, 1.0) >> >> BTW: the 'fill_color' kwarg of drawmapboundary basemap method allows you >> to set the background color of the map. >> >> http://matplotlib.sourceforge.net/basemap/doc/html/api/basemap_api.html >> >> It fills only the map region (which for some projections, like the >> orthographic, is not the same as the axes region). >> >> >> -Jeff >> >> >>> -----Original Message----- >>> From: Jeff Whitaker [mailto:[EMAIL PROTECTED] >>> Sent: mardi 23 septembre 2008 20:38 >>> To: De Pauw Antoine >>> Cc: 'John Hunter'; 'Matplotlib Users' >>> Subject: Re: [Matplotlib-users] Information request >>> >>> De Pauw Antoine wrote: >>> >>> >>> >>>> Jeff, >>>> >>>> I still don't know how to either remove this artifact or fill my arrays >>>> >>>> >>>> >>> with >>> >>> >>> >>>> values to remove empty regions, and I'll make a last attempt to resolve >>>> >>>> >> it >> >> >>>> I uploaded a data file here: http://scqp.ulb.ac.be/20080821.b56 >>>> >>>> The actual code snippet is here: >>>> http://snipplr.com/view/8307/map-plotting-python-code-temporary/ >>>> >>>> I hope you'll be able to reproduce it, I set the cmap to winter for you >>>> >>>> >> to >> >> >>>> see the gap... setting it to hot will make the grayish border visible in >>>> high resolution by zooming it... I think the border (not the empty zone) >>>> could be an artifact with the hot colormap >>>> >>>> >>>> Antoine De Pauw >>>> Collaborateur de recherches, Informatique - Research collaborator, IT >>>> Laboratoire de chimie quantique et photophysique - Quantum chemistry and >>>> photophysics laboratory >>>> Université Libre de Bruxelles - ULB >>>> >>>> >>>> >>>> >>>> >>> Antoine: Here is a version that just plots the pixels directly, without >>> interpolating to a grid. I personally like this better, since you can >>> easily see where you actually have data. >>> >>> HTH, >>> >>> -Jeff >>> >>> from mpl_toolkits.basemap import Basemap >>> import matplotlib.pyplot as plt >>> import matplotlib.mlab as mlab >>> import numpy as np >>> import os >>> fileName = '20080821.b56' >>> titre='SO2' >>> legende='Delta Brightness Temperature (K)' >>> nbreligne=long(os.stat(fileName)[6])/(8*int(fileName[-2:])) >>> rawfile=np.fromfile(open(fileName,'rb'),'<d',-1) >>> Lat=rawfile[0:nbreligne] >>> Lon=rawfile[nbreligne:nbreligne*2] >>> Val=rawfile[nbreligne*21:nbreligne*22] >>> map=Basemap(projection='mill',llcrnrlat=-90,urcrnrlat=90,\ >>> urcrnrlon=180,llcrnrlon=-180,resolution='l') >>> x, y = map(Lon, Lat) >>> >>> >>> > plt.scatter(x,y,s=25,c=Val,marker='s',edgecolor="None",cmap=plt.cm.winter,vm > >> >> >>> in=-5,vmax=-1.2, >>> alpha=0.5) >>> cb=plt.colorbar(shrink=0.6) >>> cb.ax.set_ylabel(legende,fontsize=11) >>> for t in cb.ax.get_yticklabels(): >>> t.set_fontsize(7) >>> meridians = np.arange(-180,180,60) >>> parallels = np.arange(-90,90,30) >>> map.drawparallels(parallels,labels=[1,0,0,0],fontsize=7,linewidth=0.25) >>> map.drawmeridians(meridians,labels=[0,0,0,1],fontsize=7,linewidth=0.25) >>> map.drawcoastlines(0.25,antialiased=1) >>> plt.title(titre) >>> plt.show() >>> >>> >>> >>>> -----Original Message----- >>>> From: Jeff Whitaker [mailto:[EMAIL PROTECTED] >>>> Sent: lundi 22 septembre 2008 13:59 >>>> To: De Pauw Antoine >>>> Cc: 'John Hunter'; 'Matplotlib Users' >>>> Subject: Re: [Matplotlib-users] Information request >>>> >>>> De Pauw Antoine wrote: >>>> >>>> >>>> >>>> >>>>> Jeff, >>>>> >>>>> I included here a figure where you'll see the border problem for imshow >>>>> >>>>> >>>>> >>> in >>> >>> >>> >>>>> my case >>>>> >>>>> http://img217.imageshack.us/img217/5240/testfigzp3.png >>>>> >>>>> The border wraps at -180 and 180 to form the white line >>>>> >>>>> PS: it is atmospheric ice and not SO2, I just omitted to change the >>>>> >>>>> >> title >> >> >>>>> >>>>> >>>>> >>>>> >>>> ^^ >>>> >>>> >>>> >>>> >>>>> Antoine De Pauw >>>>> Collaborateur de recherches, Informatique - Research collaborator, IT >>>>> Laboratoire de chimie quantique et photophysique - Quantum chemistry >>>>> > and > >>>>> photophysics laboratory >>>>> Université Libre de Bruxelles - ULB >>>>> >>>>> >>>>> >>>>> >>>>> >>>> Antoine: I hate to keep repeating myself - but we can't do much if you >>>> don't provide a self-contained script, that I can run, which reproduces >>>> the problem. My guess is that the line along the dateline, and the >>>> point at the South Pole are missing values (which griddata set to >>>> missing because they are outside the extent of the data) - but that's >>>> just a guess until I can reproduce it. >>>> >>>> -Jeff >>>> >>>> >>>> >>>> >>>>> -----Original Message----- >>>>> From: Antoine De Pauw [mailto:[EMAIL PROTECTED] >>>>> Sent: jeudi 18 septembre 2008 17:23 >>>>> To: Jeff Whitaker; [EMAIL PROTECTED] >>>>> Cc: 'John Hunter'; 'Matplotlib Users' >>>>> Subject: re:Re: [Matplotlib-users] Information request >>>>> >>>>> Jeff, >>>>> >>>>> No the example doesn't show that line >>>>> >>>>> If I reduce the amount of data, the border will be on every side of the >>>>> >>>>> >>>>> >>>>> >>>> plot >>>> >>>> >>>> >>>> >>>>> I'll show you an orthographic plot with no maskinf tomorrow and you >>>>> > will > >>>>> >>>>> >>>>> >>>>> >>>> see >>>> >>>> >>>> >>>> >>>>> the problem easily, it wraps in a white line along the 0° meridian and >>>>> > a > >>>>> white circle in the pole >>>>> >>>>> I think it's the imshow layer that is not totally transparent on the >>>>> > map > >>>>> background.. I tried every trick I could for example to put some >>>>> >>>>> >>>>> >>>>> >>>> zero-valued >>>> >>>> >>>> >>>> >>>>> points on each corner to make imshow interpolate correctly the sides, >>>>> >>>>> >> but >> >> >>>>> that doesn't make any difference >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>>> De Pauw Antoine wrote: >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> >>>>>>> Jeff, >>>>>>> >>>>>>> Yes they disappear, and they fluctuate with the interpolation method >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>> used >>>> >>>> >>>> >>>> >>>>>>> For example, nearest interpolation don't show the line >>>>>>> >>>>>>> Also, if I reduce the grid resolution, the line is thicker, and if I >>>>>>> >>>>>>> >>>>>>> >>> use >>> >>> >>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>> a >>>>> >>>>> >>>>> >>>>> >>>>> >>>>>>> masked array to get rid of undesired values, the border shows really >>>>>>> strongly >>>>>>> >>>>>>> Here's an example everyone will see: >>>>>>> >>>>>>> http://img225.imageshack.us/img225/2671/testfigep2.png >>>>>>> >>>>>>> (everything except the clouds is noise) >>>>>>> >>>>>>> Antoine De Pauw >>>>>>> Collaborateur de recherches, Informatique - Research collaborator, IT >>>>>>> Laboratoire de chimie quantique et photophysique - Quantum chemistry >>>>>>> >>>>>>> >>>>>>> >>> and >>> >>> >>> >>>>>>> photophysics laboratory >>>>>>> Université Libre de Bruxelles - ULB >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>> Antoine: Sorry to seem dense, but I don't see anything wrong with >>>>>> > that > >>>>>> >>>>>> >> >> >>>>>> plot. I see a white border along the north and south pole, but I >>>>>> intrepret that to be missing values. However, my eyes are notoriously >>>>>> > > >>>>>> bad. I'd like to be to run a script that generates the artifacts >>>>>> myself, so I can zoom in and see the problem myself. Does the >>>>>> griddata_demo.py script show the same problem for you? >>>>>> >>>>>> -Jeff >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> >>>>>>> -----Original Message----- >>>>>>> From: Jeff Whitaker [mailto:[EMAIL PROTECTED] >>>>>>> Sent: mercredi 17 septembre 2008 19:05 >>>>>>> To: John Hunter >>>>>>> Cc: De Pauw Antoine; Matplotlib Users >>>>>>> Subject: Re: [Matplotlib-users] Information request >>>>>>> >>>>>>> John Hunter wrote: >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>>> On Wed, Sep 17, 2008 at 11:54 AM, John Hunter <[EMAIL PROTECTED]> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>> wrote: >>>> >>>> >>>> >>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>>> Attached is a screenshot (zoom.png) from the gimp, zoomed in near >>>>>>>>> >>>>>>>>> >> the >> >> >>>>>>>>> axes border. The black horizontal line is the top axes border, the >>>>>>>>> horizontal grey line is the artifact, the vertical dashed line is a >>>>>>>>> grid line. I don't know if this offers a clue, but if you look at >>>>>>>>> > a > >>>>>>>>> zoom in the upper right corner, the grey line seems to break up >>>>>>>>> > and > >>>>>>>>> curve down and to the right (corner.png) >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>> Sorry, screwed up corner.png (I attached the original and not the >>>>>>>> screenshot). The correct screenshot is attached >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>> John: OK, now I finally see it. Antoine: Do these artifacts >>>>>>> disappear if you comment out the imshow call? >>>>>>> >>>>>>> -Jeff >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>> -- >>>>>> Jeffrey S. Whitaker Phone : (303)497-6313 >>>>>> Meteorologist FAX : (303)497-6449 >>>>>> NOAA/OAR/PSD R/PSD1 Email : [EMAIL PROTECTED] >>>>>> 325 Broadway Office : Skaggs Research Cntr 1D-113 >>>>>> Boulder, CO, USA 80303-3328 Web : http://tinyurl.com/5telg >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>> >>>> >>>> >>>> >>> >>> >>> >> >> > > > -- Jeffrey S. Whitaker Phone : (303)497-6313 NOAA/OAR/CDC R/PSD1 FAX : (303)497-6449 325 Broadway Boulder, CO, USA 80305-3328 ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Matplotlib-users mailing list Matplotlib-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/matplotlib-users
