| Re: Relative strengths: IDL vs. PV-Wave ? [message #7608] |
Tue, 03 December 1996 00:00 |
Martin Szummer
Messages: 2
Registered: December 1996
|
Junior Member |
|
|
Dear All,
I did some searches using DejaNews and found previous
comments regarding the merits of IDL and PV-Wave.
I have taken the liberty to abbreviate the replies;
the context may be missing, my apologies.
Use DejaNews to retrieve the original posts. The query string was:
IDL & PV-Wave & (compar* | differen* | better | best | review*)
Also I do not know how up to date the information is with respect
to the latest versions, but I have included post dates.
If you think these facts and opinions are mistaken, please add yours!
--- Martin
------------------------------------------------------------ -------
grunes@news.nrl.navy.mil (Mitchell R Grunes)
sci.image.processing, comp.lang.apl posted 1996/09/24, 1996/08/15
IDL vs PV-Wave
IDL handles more image file formats
IDL handles 24 bit display, PV-Wave could not, at least in the past
PV-Wave has smooth animation on SGI's, but IDL flickers
IDL has better memory use
IDL has better world maps
IDL has possibly better documentation
------------------------------------------------------------ ------
Comparison IDL and Matlab
Source: http://www.amara.com/ftpstuff/IDL_Matlab.txt
IDL vs Matlab
Matlab
+ more compact syntax
- no explicit types
+ more beautiful graphics
- non-intuitive to customize graphics
+ more built-in functions than IDL
+ more toolkits
- difficult to batch-process a directory of files
+ better technical support than IDL
IDL
+ more of a true programming language: procedures, functions, explicit
types
+ more file formats supported (HDF, CDF, easier to write new ones)
+ file I/O easier
------------------------------------------------------------ ---------
William Clodius <wclodius@lanl.gov>
Date: 1996/09/23
Newsgroups: sci.image.processing
...
Both packages provide accesss to extensive numerial libraries, PV-Wave
to those of IMSL (this may be an optional add-on), IDL to an enxtensive
subset of Numerical Recipes Routines. No knowledgeable person would
argue that the Numerical Recipes routines are better than those of IMSL,
but the wide availability of the Numerical Recipes text may make
cutomization of those routines easier in IDL.
------------------------------------------------------------ -----------
Newsgroups: sci.image.processing
From: medimtintl@aol.com (MedImTIntl)
Date: 1996/09/21
(...) Any "additional library" from the VNI is
many times outweighed by the contributions in the User Library developed
by hundreds of users around the world that continuously contribute and
enrich the RSI offerings.
------------------------------------------------------------ ----------
comp.lang.idl-pvwave
Subject: Re: Can I do this without using loops?
From: David Ritscher <david.ritscher@zibmt.uni-ulm.de>
Date: 1996/06/14
Message-Id: <31C17DE4.50C@zibmt.uni-ulm.de>
IDL. Version 3.1.0 and PV-WAVE v6.01. Sometimes one was faster,
sometimes the other (often by a factor of two or more!).
Insert columns Insert rows
IDL PV-Wave IDL PV-Wave
1. 22.84 17.58 30.48 21.31
2. 22.77 30.81 4.45 1.48
3. 3.88 9.73 3.87 9.70
4. 12.05 11.93 5.67 7.17
5. 89.24 164.56 87.51 163.15
6. 141.93 385.72 140.45 389.53
7. 92.75 166.44 92.82 169.71
8. 1.30 4.96 9.78 10.29
9. 30.75 28.71 18.40 21.60
Tested using the following number of rows and columns: 1000, 30000
Insert columns Insert rows
IDL PV-Wave IDL PV-Wave
1. 84.95 71.09 Inf Inf
2. Inf Inf 28.52 22.69
3. 27.65 42.95 30.88 50.10
4. 129.53 138.84 124.39 126.26
5. 300.62 547.51 296.74 556.55
6. - 1284.83 - 1300.17
7. Inf Inf Inf Inf
8. 20.07 28.76 Inf Inf
9. - 3100.02 158.07 161.23
------------------------------------------------------------ --------
Newsgroups: comp.lang.idl-Pvwave
Date: 1996/06/15
Subject: Re: Can I do this without using loops?
From: David Ritscher <david.ritscher@zibmt.uni-ulm.de>
...
Ratios of execution times of rebin(/sample) / rebin():
problem with 3000, 3000 rows and columns:
IDL PV-Wave
test8 1.22997 0.676364
test8_rows 0.999363 0.956496
problem with 1000, 30000 rows and columns:
IDL PV-Wave
test8 1.06719 0.936516
|
|
|
|
| Re: Relative strengths: IDL vs. PV-Wave ? [message #7613 is a reply to message #7608] |
Mon, 02 December 1996 00:00  |
Robert Jansen
Messages: 1
Registered: December 1996
|
Junior Member |
|
|
Martin Szummer wrote:
>
> Dear All,
>
> I am planning to acquire IDL or PV-Wave, and I was looking for
> information about their respective strenghts. The FAQ avoids
> this issue, so I was hoping you could enlighten me, while
> being fair to both parties.
>
> The following things are very important in my work:
> . image and video processing
> . machine learning and neural nets
> . rich programming language (batch processing)
> . linking with C/C++ code (under Windows NT, but also perhaps Linux)
> . speed (including speed of the IDL language)
> . many active users
> How do the products compare in the above regards?
>
> What product has the largest active user community?
>
> From IDL and PV-Wave's Web pages, I get the impression that
> PV-Wave sells more different commercial components/toolboxes
> (signal processing, beta image processing)
>
> What are your thoughts?
>
> Thank you,
> Martin Szummer
IDL is better. It has better color handling, more complete algorithms,
better memory management, faster algorithms. This is comming from
someone who uses PV-Wave (Since it's what is on most of the machines
I use). For example the image rotation routines in IDL allow for
bicubic-spline interpolation which is not in PV-Wave. They have
also provided a general interpolation routine "interpolate" which
should be a part of any image analysis package but is not in
PV-Wave. They also have 24bit color graphics.
What PV-Wave has done is to write a ton of silly little utilities
which could easily be written by a beginner user. For example
they have a function called avg.pro which takes the average of an
array. It is nothing more than total(a)/n_elements(a). There
are a million other examples. These tend to reduce the usefullness
of the language as far as I'm concerned since they use up names
and clutter up the language.
However, you may have to consider costs. I think IDL may be more
expensive.
You of course should look at MATLAB also which may beat out
both of these due to the ease with wich it connects with
C/C++
jansen@tsunami.nrl.navy.mil
--
Robert Jansen o /o
NRL-SEALAB o /o\ \o__
Tel: 202-767-8244 / | /
FAX: 202-767-5599 /_\o / \ /\
/ \ / / /
Pass Set Spike
|
|
|
|
| Re: Relative strengths: IDL vs. PV-Wave ? [message #7616 is a reply to message #7613] |
Mon, 02 December 1996 00:00 |
mcook
Messages: 12
Registered: March 1994
|
Junior Member |
|
|
Martin,
I can't answer all of your questions, but I can relate my own
experiences for you to consider. First of all we have site
licenses for PV-Wave on many of our computers here, so it
is more widespread than IDL. I use IDL on a couple of projects
primarily because I want to use their ENVI package for spectral
remote sensing which is based on IDL. I have used both IDL and
PV-Wave for my own uses however, and here are some of my
observations:
* There isn't much difference as far as coding goes for the most
part. For some reason, IDL seems to run faster for me, but
that may be my particular applications. Don't use me as
a benchmark.
* The main differences come from when the two products split
from the common source and additions were made: GUI and
numerical/stat routines. Review the two offerings yourself
to compare their relative merits.
* I'm not crazy about the GUI for either. If I have to make
a GUI now, I will use a TCL/TK driver with PV-Wave as an
engine (I haven't tried it with IDL, simply because I've
never had to). This opinion is primarily because I don't
like messing with the intricacies of low-level widgets and
the like. Also they seem much less flexible than I like.
* Some of my work involves using polygon graphics. PV-Wave
has quite a few useful routines for polygon manipulation
(both will render polygons) already included that really
help here. It is easy to write your own routines to mimic
the PV-Wave routines in IDL however.
* The only bug I've noticed between the two was a consistently
incorrect rendering from PV-Wave of a 3-D surface with shading.
IDL produced the correct rendering with essentially the same
code and was much faster about it.
* I've only done a little interfacing to other languages, with
IDL, but didn't have much problem. Again, it was only minor
routines.
* I haven't noticed any difference for the image processing
routines.
* I would surmise that you want your own code for the machine
learning or NN applications. I suspect your code would run
on either about equally well.
* I was told there may be OpenGL support in a future version
of IDL, which is something particularly useful to me.
* Even though PV-Wave is widespread here, I've noticed most
engineers use MATLAB. I firmly believe however that is
because they have little or no experience with PV-Wave.
They don't realize the rich offerings Wave and IDL have
for their work.
Hope this helps. I don't think you'll go wrong with either.
If however you have need for a product like ENVI, I have
found it very useful for my work and it isn't available for
PV-Wave.
Regards,
Mark Cook
mcook@ti.com
Dr. Mark K. Cook, MGTS
Systems Group
Texas Instruments, Inc.
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
