Memory allocation problem

[felix.schalck-2]

Hello,

I'm trying to create a huge topographic map of europe using srtm3 data.
Dowload and conversion seemed to work flawlessly; now i'm trying to
paste 41 rows of 1°lat and 98°long together in one single grd grid,
which will be used to create the map. Unfortunatly each row is 540 mb
big, and after having pasted the 5 first rows, grdpaste dies badly, with
the following error message:

GMT fatal error: grdpaste could not allocate memory [3.2 Gb]

I have an old amd athlon xp 1,6 ghz with 1gb RAM, and 2Gb SWAP. Do I
need the full 41x540Mb (which will be the size of the final grid) amount
of swap ? Is there no way to tell grdpaste to directly write the new row
a the end of the final grid file, on the hdd, without mapping the whole
thing into memory ?

Thank you for your help,

Felix

To unsubscribe, send the message "signoff gmt-help" to listserv@...

[Joaquim Luis]

felix.schalck wrote:

> Hello,
>
> I'm trying to create a huge topographic map of europe using srtm3
> data. Dowload and conversion seemed to work flawlessly; now i'm trying
> to paste 41 rows of 1°lat and 98°long together in one single grd grid,
> which will be used to create the map. Unfortunatly each row is 540 mb
> big, and after having pasted the 5 first rows, grdpaste dies badly,
> with the following error message:
>
> GMT fatal error: grdpaste could not allocate memory [3.2 Gb]
>
> I have an old amd athlon xp 1,6 ghz with 1gb RAM, and 2Gb SWAP. Do I
> need the full 41x540Mb (which will be the size of the final grid)
> amount of swap ? Is there no way to tell grdpaste to directly write
> the new row a the end of the final grid file, on the hdd, without
> mapping the whole thing into memory ?
>

Why are you trying to do that (which obviously you won't succeed)?

There is no way that on papers the size we are used to deal with you
would ever tell the difference between a 48x91º map produced with SRTM3
and , for example, SRTM+ plus or ETOPO2.
Even if you find a way to paste together 41x540 = 21.6 Gb your next
question would be "How can I make grdimage swallow a 21 Gb grid?"

J. Luis

To unsubscribe, send the message "signoff gmt-help" to listserv@...

[Walter H. F. Smith]

Dear Felix,

I think the "grd" tools are an inefficient way to deal with this much  
data.  A better way is to work in the native binary integer format of  
your machine.  Once the entire area is assembled you can make an  
entry for the native binary file in your "grdraster.info" and use  
GMT's "grdraster" to extract subsets of the data as "grd" files.

If you are trying to handle the entire area at once, think about  
whether you really need to do this.  There are (20 * 60)^2 data  
points in each 1x1 degree cell, and you have a 41 * 98 degree area.  
If I've done my math right, that is almost 6 x 10^9 data points.  If  
for example you wanted to use GMT's "grdimage" to image the entire  
area at once, you would need an absurd amount of memory.  The  
operating system can give you virtual memory (it shouldn't have to be  
RAM, unless the kernel has a limit on how much you can allocate to  
one process), but you would need 64-bit addressing and it would take  
forever to run.  Also, do you have a display device that can handle  
that many pixels?

If you want high-resolution images, work in smaller areas.  If you  
want a map of an area 41 by 98, consider reducing the data sampling.  
"grdraster" can do this if you simply want to sub-sample every N'th  
point.  If you are worried about aliasing, then build some averaging  
in when you assemble the tiles.

It sounds like you have used "grdpaste" to assemble the tiles in east-
west rows.  Good, because that isn't easy to do with any other tool.  
Once you have done that, you can use "grd2xyz -ZTLh ..." on each  
completed row's "grd" file, to dump it to native binary.  The native  
binaries can then be assembled using "cat" to append each row to the  
end of the row to the north, building from north to south.

Caution:  I know the SRTMv2 30 arc second tiles are "pixel  
registered", so this will work directly. But if I remember correctly  
maybe the 3 arc second tiles are "grid registered" which means that  
the edge values are repeated.  If you simply "cat" in the above  
recipe you will get a row of data repeated twice at each 1 degree  
edge.  You don't want that.  The solution is to skip writing the last  
row on every row except the last one, where you need to skip writing  
the first one.  To skip some bytes in a binary stream, use "dd".

So grd2xyz -ZTLh ... | dd ...

followed by "cat" should build something you can manage with  
"grdraster".

Good luck.

walter

Walter H F Smith
Chairman, GEBCO SCDB
Geophysicist, Laboratory for Satellite Altimetry
NOAA NESDIS code E/RA-31
1335 East West Hwy, room 5408
Silver Spring MD 20910-3226
tel [phone], extension 126
fax [phone]
Walter.HF.Smith@...


On Sep 24, 2008, at 9:15 AM, felix.schalck wrote:

> Hello,
>
> I'm trying to create a huge topographic map of europe using srtm3  
> data. Dowload and conversion seemed to work flawlessly; now i'm  
> trying to paste 41 rows of 1°lat and 98°long together in one single  
> grd grid, which will be used to create the map. Unfortunatly each  
> row is 540 mb big, and after having pasted the 5 first rows,  
> grdpaste dies badly, with the following error message:
>
> GMT fatal error: grdpaste could not allocate memory [3.2 Gb]
>
> I have an old amd athlon xp 1,6 ghz with 1gb RAM, and 2Gb SWAP. Do  
> I need the full 41x540Mb (which will be the size of the final grid)  
> amount of swap ? Is there no way to tell grdpaste to directly write  
> the new row a the end of the final grid file, on the hdd, without  
> mapping the whole thing into memory ?
>
> Thank you for your help,
>
> Felix
>
> To unsubscribe, send the message "signoff gmt-help" to  
> listserv@...

To unsubscribe, send the message "signoff gmt-help" to listserv@...

[John Robbins-4]

Some parts of this message have been removed. Learn more about Nabble's security policy.

felix.schalck wrote:

Hello,

I'm trying to create a huge topographic map of europe using srtm3 data. Dowload and conversion seemed to work flawlessly; now i'm trying to paste 41 rows of 1°lat and 98°long together in one single grd grid, which will be used to create the map. Unfortunatly each row is 540 mb big, and after having pasted the 5 first rows, grdpaste dies badly, with the following error message:

GMT fatal error: grdpaste could not allocate memory [3.2 Gb]

I have an old amd athlon xp 1,6 ghz with 1gb RAM, and 2Gb SWAP. Do I need the full 41x540Mb (which will be the size of the final grid) amount of swap ? Is there no way to tell grdpaste to directly write the new row a the end of the final grid file, on the hdd, without mapping the whole thing into memory ?

Thank you for your help,

Felix

Interesting task, Felix. It sounds Herculean in nature!

I couldn't gather what the specifics were for your topo map, but if you are merely making a colorized map with grdimage, you probably could "layer" a series of grdimage calls (to each srtm3 row) to a single, output postscript file (it too might get large, but it only resides on the hdd). In other words, instead of trying to build a single huge .grd file, take each row's .grd file as a "slice," apply a single color table across all slices, and "build" the map directly.

If you were hoping to plot contour lines, that might be more problematic as the contours may not line-up across slices. I've never tried it myself, so I'm not sure what the edge effects might look like.

Also, if you were thinking of using grdimage and hoping to apply a grdgradient, you'll have to work out the specifics for that and apply them uniformly for each row.

Its an ugly problem, but there are alternatives. Furthermore, others may have tricks I'm not aware of. But, as you've discovered, trying to create a single .grd file using grdpaste just isn't going to work.

As Joaquim wrote, what is the final size of the graphic? If you're putting it on a A4-sized sheet, then you can reduce the resolution a huge amount with grdsample, then do your grdpastes. If you're wall-papering the hallway of your office, then split the rows into regions and work that way. Again, a little creative thinking will work, either way.

I just saw Walter's reply and that sounds like yet another possible solution. Good luck.

John Robbins

To unsubscribe, send the message "signoff gmt-help" to listserv@...

[felix.schalck-2]

Many thanks for all three replies; i did not expect so fast and so
precise replies. Basically, the main problem i read out of your emails
is the lack of details, so here are some more:

The project is mainly designed for personal use (even though people
around me show some interest); it could be described as sort of a
therapy - and is thus quite important for me. The main objectives of the
map is precision; time is no problem, since the whole work can take
weeks if necessary. The final map is not intended to be printed, but
rather to be directly read on my 22" screen, by scrolling and zooming to
the desired area. I was planning a resolution of about 5000x4200 px,
which gives a scala of 1:4000000, but this resolution is not fixed yet,
and depends on the SRTM3 based grdimage results. I made a first map with
GLOBE data, but the details level is just not enough, and at my
resolution, it gives soapy results; and that's why I turned to SRTM3 data.

In terms of gmt,  the first map was an Albers projection of aera  
R-8/32/75/61r, with settings Jb15/30/40/60/1:4000000. The new map is
expected to have the same settings, and thus be drawn in three steps:

1 - barymetry, with grdimage
2 - topography, with grdimage
3 - Coastlines, Rivers and polit. borders, with ps coast

I think I'm going to make 2 or three separate files, one with
topography/barymetry, and one with coastlines, the latter in vectorized
format, and then merge the files via GIMP.

Note: Since it's a conical projection, the top portion of the aera needs
to be much larger than the average one; that's why I got data from -25
to 67. Unfortunately, i did not find out how to build a trapezoid grid
with grd tools, and thus end with some 3-5 Gigs of empty space (ocean)
below Iceland, and west of France.

So, up to your reactions: As Mr Smith pointed it, perhaps processing the
whole data in one command might not be a good idea, even though I have
access every night to a 64bit cpu, and the native binary solution seems
to be very promising. That's why I would like more details about Mr
Robbin's grdimage "slices" solution, which could not only relieve the
memory problem, but also further graphical processing via GIMP, and on
top of that address the "ocean-waste-problem". How is that doable ? Do I
only need to keep the same projection settings for every slice, change
the aera, and run the grdimage command 40 times ? And than paste
everything together with gimp. Would they fit to each other ? What about
the gradient ?
Map title, coastlines and other details could be created in another
file, and become a "background" layer to paste the topographic slices.

I'm rather new to the GMT tools, and any comments, advices of more
experienced users are really welcome, to help me create this map.

Thanks in advance,

Felix


John Robbins wrote:

> felix.schalck wrote:
>> Hello,
>>
>> I'm trying to create a huge topographic map of europe using srtm3
>> data. Dowload and conversion seemed to work flawlessly; now i'm
>> trying to paste 41 rows of 1°lat and 98°long together in one single
>> grd grid, which will be used to create the map. Unfortunatly each row
>> is 540 mb big, and after having pasted the 5 first rows, grdpaste
>> dies badly, with the following error message:
>>
>> GMT fatal error: grdpaste could not allocate memory [3.2 Gb]
>>
>> I have an old amd athlon xp 1,6 ghz with 1gb RAM, and 2Gb SWAP. Do I
>> need the full 41x540Mb (which will be the size of the final grid)
>> amount of swap ? Is there no way to tell grdpaste to directly write
>> the new row a the end of the final grid file, on the hdd, without
>> mapping the whole thing into memory ?
>>
>> Thank you for your help,
>>
>> Felix
>>
>>
> Interesting task, Felix. It sounds Herculean in nature!
>
> I couldn't gather what the specifics were for your topo map, but if
> you are merely making a colorized map with grdimage, you probably
> could "layer" a series of grdimage calls (to each srtm3 row) to a
> single, output postscript file (it too might get large, but it only
> resides on the hdd). In other words, instead of trying to build a
> single huge .grd file, take each row's .grd file as a "slice," apply a
> single color table across all slices, and "build" the map directly.
>
> If you were hoping to plot contour lines, that might be more
> problematic as the contours may not line-up across slices. I've never
> tried it myself, so I'm not sure what the edge effects might look like.
>
> Also, if you were thinking of using grdimage and hoping to apply a
> grdgradient, you'll have to work out the specifics for that and apply
> them uniformly for each row.
>
> Its an ugly problem, but there are alternatives. Furthermore, others
> may have tricks I'm not aware of. But, as you've discovered, trying to
> create a single .grd file using grdpaste just isn't going to work.
>
> As Joaquim wrote, what is the final size of the graphic? If you're
> putting it on a A4-sized sheet, then you can reduce the resolution a
> huge amount with grdsample, then do your grdpastes. If you're
> wall-papering the hallway of your office, then split the rows into
> regions and work that way. Again, a little creative thinking will
> work, either way.
>
> I just saw Walter's reply and that sounds like yet another possible
> solution. Good luck.
>
> John Robbins
>
> --
>
> Cryospheric Branch, Code 614.1
>
> NASA Goddard Space Flight Center
>
> Greenbelt, Maryland 20771
>
> Web: ICESat Home Page <http://icesat.gsfc.nasa.gov/>
>
> Email: John.Robbins@... <mailto:John.Robbins@...>
>
> To unsubscribe, send the message "signoff gmt-help" to
> listserv@...

To unsubscribe, send the message "signoff gmt-help" to listserv@...

[Walter H. F. Smith]

Dear Felix,

I do not disagree with the other suggestions sent in reply to your  
first posting.

Since you also want to do the Albers Conic projection there is  
additional overhead in projecting the coordinates, and this makes it  
even harder to work in the full resolution of SRTM3.

But I do not think you have to.  Here is why:

You say you want a final resolution of about 5000 by 4200 pixels.  
Your previous email said you were trying to assemble an area 98 by 41  
degrees, with, as I pointed out, 1200x1200 pixels per degree, or a  
total of 5.8 x 10^9 input data points.  But your total number of  
output pixels is 2.1 x 10^7.  In other words, the input data quantity  
is more than 100 times the needed output quantity.

(The above calculation is rough, as it ignores what happens when the  
data are projected.  One could use mapproject to compute the  
dimensions of your map and do a smarter calculation.  I'm just trying  
to get an order of magnitude estimate here.)

So it seems that you could decimate (literally, reduce by a factor of  
10) the input data quantity in each dimension and things should be  
satisfactory.  If this is true, then it should be possible to build  
your map from an input of 30-arc-second, not 3-arc-second, data.  
Thus I am surprised that you were unsatisfied with the results from  
the GLOBE data.  Try the SRTM (version 2) 30 arc second data, and see  
what you think.  Or, to include the ocean area, use SRTM30PLUS  
(there's a new version in the last few days) at http://topex.ucsd.edu

If working with 30-arc-second data as input proves to be  
unsatisfactory, then the following approach may work:

Use GMT tools as you did to assemble the 1x1 degree tiles of SRTM 3  
data.  Use "grd2xyz" to dump each one as lon,lat,height and pipe this  
through "mapproject" with your desired projection, using the -R and -
J for the entire area, not just the tile.  Concatenate the results  
for all tiles, or at least for many tiles in a "chunk" (a subset  
area, to avoid having total size too big) and pipe the result through  
"blockmean" with the -R and -I [and -F] set not in lon,lat  
coordinates but now in x,y map projected coordinates, so you average  
the data into the final pixels you intend to use.  If this has to be  
done in several chunks then later cat all chunks together and run  
through blockmean with the same parameters again; this will clean up  
the overlap zones of the chunks.  Finally, the output of blockmean  
can be input to xyz2grd (again, -R and -I are now in projected xy) to  
make your grid file from which you can image, contour, etc.

this will work.  it just takes patience.

w



Walter H F Smith
Chairman, GEBCO SCDB
Geophysicist, Laboratory for Satellite Altimetry
NOAA NESDIS code E/RA-31
1335 East West Hwy, room 5408
Silver Spring MD 20910-3226
tel [phone], extension 126
fax [phone]
Walter.HF.Smith@...


On Sep 24, 2008, at 11:41 AM, felix.schalck wrote:

> Many thanks for all three replies; i did not expect so fast and so  
> precise replies. Basically, the main problem i read out of your  
> emails is the lack of details, so here are some more:
>
> The project is mainly designed for personal use (even though people  
> around me show some interest); it could be described as sort of a  
> therapy - and is thus quite important for me. The main objectives  
> of the map is precision; time is no problem, since the whole work  
> can take weeks if necessary. The final map is not intended to be  
> printed, but rather to be directly read on my 22" screen, by  
> scrolling and zooming to the desired area. I was planning a  
> resolution of about 5000x4200 px, which gives a scala of 1:4000000,  
> but this resolution is not fixed yet, and depends on the SRTM3  
> based grdimage results. I made a first map with GLOBE data, but the  
> details level is just not enough, and at my resolution, it gives  
> soapy results; and that's why I turned to SRTM3 data.
>
> In terms of gmt,  the first map was an Albers projection of aera  
> R-8/32/75/61r, with settings Jb15/30/40/60/1:4000000. The new map  
> is expected to have the same settings, and thus be drawn in three  
> steps:
>
> 1 - barymetry, with grdimage
> 2 - topography, with grdimage
> 3 - Coastlines, Rivers and polit. borders, with ps coast
>
> I think I'm going to make 2 or three separate files, one with  
> topography/barymetry, and one with coastlines, the latter in  
> vectorized format, and then merge the files via GIMP.
>
> Note: Since it's a conical projection, the top portion of the aera  
> needs to be much larger than the average one; that's why I got data  
> from -25 to 67. Unfortunately, i did not find out how to build a  
> trapezoid grid with grd tools, and thus end with some 3-5 Gigs of  
> empty space (ocean) below Iceland, and west of France.
>
> So, up to your reactions: As Mr Smith pointed it, perhaps  
> processing the whole data in one command might not be a good idea,  
> even though I have access every night to a 64bit cpu, and the  
> native binary solution seems to be very promising. That's why I  
> would like more details about Mr Robbin's grdimage "slices"  
> solution, which could not only relieve the memory problem, but also  
> further graphical processing via GIMP, and on top of that address  
> the "ocean-waste-problem". How is that doable ? Do I only need to  
> keep the same projection settings for every slice, change the aera,  
> and run the grdimage command 40 times ? And than paste everything  
> together with gimp. Would they fit to each other ? What about the  
> gradient ?
> Map title, coastlines and other details could be created in another  
> file, and become a "background" layer to paste the topographic slices.
>
> I'm rather new to the GMT tools, and any comments, advices of more  
> experienced users are really welcome, to help me create this map.
>
> Thanks in advance,
>
> Felix
>
>
> John Robbins wrote:
>> felix.schalck wrote:
>>> Hello,
>>>
>>> I'm trying to create a huge topographic map of europe using srtm3  
>>> data. Dowload and conversion seemed to work flawlessly; now i'm  
>>> trying to paste 41 rows of 1°lat and 98°long together in one  
>>> single grd grid, which will be used to create the map.  
>>> Unfortunatly each row is 540 mb big, and after having pasted the  
>>> 5 first rows, grdpaste dies badly, with the following error message:
>>>
>>> GMT fatal error: grdpaste could not allocate memory [3.2 Gb]
>>>
>>> I have an old amd athlon xp 1,6 ghz with 1gb RAM, and 2Gb SWAP.  
>>> Do I need the full 41x540Mb (which will be the size of the final  
>>> grid) amount of swap ? Is there no way to tell grdpaste to  
>>> directly write the new row a the end of the final grid file, on  
>>> the hdd, without mapping the whole thing into memory ?
>>>
>>> Thank you for your help,
>>>
>>> Felix
>>>
>>>
>> Interesting task, Felix. It sounds Herculean in nature!
>>
>> I couldn't gather what the specifics were for your topo map, but  
>> if you are merely making a colorized map with grdimage, you  
>> probably could "layer" a series of grdimage calls (to each srtm3  
>> row) to a single, output postscript file (it too might get large,  
>> but it only resides on the hdd). In other words, instead of trying  
>> to build a single huge .grd file, take each row's .grd file as a  
>> "slice," apply a single color table across all slices, and "build"  
>> the map directly.
>>
>> If you were hoping to plot contour lines, that might be more  
>> problematic as the contours may not line-up across slices. I've  
>> never tried it myself, so I'm not sure what the edge effects might  
>> look like.
>>
>> Also, if you were thinking of using grdimage and hoping to apply a  
>> grdgradient, you'll have to work out the specifics for that and  
>> apply them uniformly for each row.
>>
>> Its an ugly problem, but there are alternatives. Furthermore,  
>> others may have tricks I'm not aware of. But, as you've  
>> discovered, trying to create a single .grd file using grdpaste  
>> just isn't going to work.
>>
>> As Joaquim wrote, what is the final size of the graphic? If you're  
>> putting it on a A4-sized sheet, then you can reduce the resolution  
>> a huge amount with grdsample, then do your grdpastes. If you're  
>> wall-papering the hallway of your office, then split the rows into  
>> regions and work that way. Again, a little creative thinking will  
>> work, either way.
>>
>> I just saw Walter's reply and that sounds like yet another  
>> possible solution. Good luck.
>>
>> John Robbins
>>
>> --
>>
>> Cryospheric Branch, Code 614.1
>>
>> NASA Goddard Space Flight Center
>>
>> Greenbelt, Maryland 20771
>>
>> Web: ICESat Home Page <http://icesat.gsfc.nasa.gov/>
>>
>> Email: John.Robbins@... <mailto:John.Robbins@...>
>>
>> To unsubscribe, send the message "signoff gmt-help" to  
>> listserv@...
>
> To unsubscribe, send the message "signoff gmt-help" to  
> listserv@...

To unsubscribe, send the message "signoff gmt-help" to listserv@...

[felix.schalck-2]

Hi,

How can you convert a huge GMT-made ps file to any other format, and
especially PNG or PDF, while keeping the entire content ? I tried every
conversion app I have, including ps2raster, and always end with a
a4-size cut picture. Only GIMP import-option keeps original size; but
the antialiasing sucks.
Perhaps this is a problem of media-size ? Can't you make custom sized pdfs ?

Thanks for your help,

Felix

To unsubscribe, send the message "signoff gmt-help" to listserv@...

[hadi salimi]

Hi

I think you can try with Gsview or Acrobat Distiller.

Hadi

On Tue, Sep 30, 2008 at 1:50 PM, felix.schalck <felix.schalck@...> wrote:

Hi,

How can you convert a huge GMT-made ps file to any other format, and especially PNG or PDF, while keeping the entire content ? I tried every conversion app I have, including ps2raster, and always end with a a4-size cut picture. Only GIMP import-option keeps original size; but the antialiasing sucks.
Perhaps this is a problem of media-size ? Can't you make custom sized pdfs ?

Thanks for your help,

Felix

To unsubscribe, send the message "signoff gmt-help" to listserv@...

To unsubscribe, send the message "signoff gmt-help" to listserv@...

[felix.schalck-2]

Thanks for your reply. Unfortunately Gsview cuts to A4 too, and adobe
distiller seems to be a commercial software...


Hadi Salimi wrote:

> Hi
> I think you can try with Gsview or Acrobat Distiller.
> Hadi
>
> On Tue, Sep 30, 2008 at 1:50 PM, felix.schalck
> <felix.schalck@... <mailto:felix.schalck@...>> wrote:
>
>     Hi,
>
>     How can you convert a huge GMT-made ps file to any other format,
>     and especially PNG or PDF, while keeping the entire content ? I
>     tried every conversion app I have, including ps2raster, and always
>     end with a a4-size cut picture. Only GIMP import-option keeps
>     original size; but the antialiasing sucks.
>     Perhaps this is a problem of media-size ? Can't you make custom
>     sized pdfs ?
>
>     Thanks for your help,
>
>     Felix
>
>     To unsubscribe, send the message "signoff gmt-help" to
>     listserv@... <mailto:listserv@...>
>
>
> To unsubscribe, send the message "signoff gmt-help" to
> listserv@...

To unsubscribe, send the message "signoff gmt-help" to listserv@...

[Milton P. Plasencia Linares]

Perhaps the command convert of ImageMagick package can you help.


**********************************
Milton P. Plasencia Linares

Istituto Nazionale di Oceanografia e di Geofisica Sperimentale - OGS
Borgo Grotta Gigante 42/C
(34010) Sgonico - TRIESTE - ITALIA

Tel: [phone]
Fax: [phone]

E-mail: mplasencia@...

ASAIN (Antarctic Seismographic Argentinean Italian Network)
*********************************


Quoting "felix.schalck" <felix.schalck@...>:

 Thanks for your reply. Unfortunately Gsview cuts to A4 too, and adobe
 distiller seems to be a commercial software...
 
 
 Hadi Salimi wrote:
 > Hi
 > I think you can try with Gsview or Acrobat Distiller.
 > Hadi
 >
 > On Tue, Sep 30, 2008 at 1:50 PM, felix.schalck
 > <felix.schalck@... <mailto:felix.schalck@...>> wrote:
 >
 >     Hi,
 >
 >     How can you convert a huge GMT-made ps file to any other format,
 >     and especially PNG or PDF, while keeping the entire content ? I
 >     tried every conversion app I have, including ps2raster, and always
 >     end with a a4-size cut picture. Only GIMP import-option keeps
 >     original size; but the antialiasing sucks.
 >     Perhaps this is a problem of media-size ? Can't you make custom
 >     sized pdfs ?
 >
 >     Thanks for your help,
 >
 >     Felix
 >
 >     To unsubscribe, send the message "signoff gmt-help" to
 >     listserv@... <mailto:listserv@...>
 >
 >
 > To unsubscribe, send the message "signoff gmt-help" to
 > listserv@...
 
 To unsubscribe, send the message "signoff gmt-help" to listserv@...
 
 Email secured by **CeSIT** Check Point gateway
 


-------------------------------------------------
This mail sent through IMP: http://horde.org/imp/

To unsubscribe, send the message "signoff gmt-help" to listserv@...

[hadi salimi]

if you send your file to me, I am ready convert it with Distiller

Hadi

On Tue, Sep 30, 2008 at 2:48 PM, Milton P. Plasencia Linares <mplasencia@...> wrote:

Perhaps the command convert of ImageMagick package can you help.


**********************************
Milton P. Plasencia Linares

Istituto Nazionale di Oceanografia e di Geofisica Sperimentale - OGS
Borgo Grotta Gigante 42/C
(34010) Sgonico - TRIESTE - ITALIA

Tel: [phone]
Fax: [phone]

E-mail: mplasencia@...

ASAIN (Antarctic Seismographic Argentinean Italian Network)
*********************************

Quoting "felix.schalck" <felix.schalck@...>:

 Thanks for your reply. Unfortunately Gsview cuts to A4 too, and adobe
 distiller seems to be a commercial software...


 Hadi Salimi wrote:
 > Hi
 > I think you can try with Gsview or Acrobat Distiller.
 > Hadi
 >
 > On Tue, Sep 30, 2008 at 1:50 PM, felix.schalck
 > <felix.schalck@... <mailto:felix.schalck@...>> wrote:
 >
 >     Hi,
 >
 >     How can you convert a huge GMT-made ps file to any other format,
 >     and especially PNG or PDF, while keeping the entire content ? I
 >     tried every conversion app I have, including ps2raster, and always
 >     end with a a4-size cut picture. Only GIMP import-option keeps
 >     original size; but the antialiasing sucks.
 >     Perhaps this is a problem of media-size ? Can't you make custom
 >     sized pdfs ?
 >
 >     Thanks for your help,
 >
 >     Felix
 >
 >     To unsubscribe, send the message "signoff gmt-help" to
 >     listserv@... <mailto:listserv@...>
 >
 >
 > To unsubscribe, send the message "signoff gmt-help" to
 > listserv@...

 To unsubscribe, send the message "signoff gmt-help" to listserv@...

 Email secured by **CeSIT** Check Point gateway



-------------------------------------------------
This mail sent through IMP: http://horde.org/imp/

To unsubscribe, send the message "signoff gmt-help" to listserv@...

To unsubscribe, send the message "signoff gmt-help" to listserv@...

[H.Fagard]

At 12:20 +0200 30/09/08, felix.schalck wrote:
>How can you convert a huge GMT-made ps file

can you put an example on a FTP server to try it?

Hervé

To unsubscribe, send the message "signoff gmt-help" to listserv@...

[J.J.Green-2]

Most of the suggestions so far have been for utilities
which call gs (ps2raster, convert, gv) but it seems that
the underlying problem is gs.

Perhaps this could by sidesptepped by producing a
png with n times the density at 1/n-th of the size.
So if you want an a3 png at 300dpi, instead use
ps2raster to make an a4 png at 600dpi or an a5 png
at 1200dpi.

Just a thought

Jim
--
J.J. Green, Dept. Applied Mathematics, Hicks Bld.,
University of Sheffield, UK.   [phone]
http://sview01.wiredworkplace.net/pub/jjg

To unsubscribe, send the message "signoff gmt-help" to listserv@...

[Walter H. F. Smith]

Is this happening with the -A option on ps2raster?

I have had no trouble converting poster-sized files with "gs", but  
you have to specify the size of the output image (or have it set  
correctly in the BoundingBox statement in the PostScript; ps2raster -
A is supposed to fix this).  If you simply accept the defaults on  
"gs" then you will get something cut down to your standard page size  
format.

walter

On Sep 30, 2008, at 06:20, felix.schalck wrote:

> Hi,
>
> How can you convert a huge GMT-made ps file to any other format,  
> and especially PNG or PDF, while keeping the entire content ? I  
> tried every conversion app I have, including ps2raster, and always  
> end with a a4-size cut picture. Only GIMP import-option keeps  
> original size; but the antialiasing sucks.
> Perhaps this is a problem of media-size ? Can't you make custom  
> sized pdfs ?
>
> Thanks for your help,
>
> Felix
>
> To unsubscribe, send the message "signoff gmt-help" to  
> listserv@...

To unsubscribe, send the message "signoff gmt-help" to listserv@...