| Re: Philosophical Scaling Question [message #51593] |
Mon, 04 December 2006 10:24  |
Brian Larsen
Messages: 270
Registered: June 2006
|
Senior Member |
|
|
> LoadCT, 33
> image = LoadData(7)
> TVLCT, r, g, b, /GET
> data = Scale_Vector(image, 1, 1000) ; Data in log scale.
> pos = [0.1, 0.1, 0.9, 0.7]
> TVImage, LogScl(data), Position=pos, /Keep, /Erase
> index = Bindgen(256)
>
> TVLCT, r[LogScl(index)], g[LogScl(index)], b[LogScl(index)]
> Colorbar, Range=[1,1000], XLOG=1, Divisions=3, $
> Position=[pos[0], 0.87, pos[2], 0.93], Minor=5
>
Ah ha, this exactly what I wanted colorbar to do!! The reason I had
never been able to get it done is that was a lot of work (and work I
didn't know how to do), it seems like a flag to colorbar or tvlct or
something that was /logscl would do the trick. Really the issue was my
unfamiliarity with logscl, tvlct, and tvimage and how to put it all
together.
Cheers,
Brian
|
|
|
|
| Re: Philosophical Scaling Question [message #51595 is a reply to message #51593] |
Mon, 04 December 2006 10:00  |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Brian Larsen writes:
> I think I agree with this statement but for most all of (making an
> assumption there I guess) just having it plot in decades as opposed to
> linear is good enough. In that case I think this thread may have
> solved this.
>
> maybe give the user some options to pass onto logscl but have the
> defaults just as were used here...
Well, *exactly*. I've thought all along the request
for a "logarithmic color bar" didn't make much sense.
Because if this is all you want, COLORBAR already does
this.
LoadCT, 33
image = LoadData(7)
TVLCT, r, g, b, /GET
data = Scale_Vector(image, 1, 1000) ; Data in log scale.
pos = [0.1, 0.1, 0.9, 0.7]
TVImage, LogScl(data), Position=pos, /Keep, /Erase
index = Bindgen(256)
TVLCT, r[LogScl(index)], g[LogScl(index)], b[LogScl(index)]
Colorbar, Range=[1,1000], XLOG=1, Divisions=3, $
Position=[pos[0], 0.87, pos[2], 0.93], Minor=5
You can choose the logarithmic scale to use. Use LOGSCL
for a true log scaling. Choose GMASCL for a power-law
scaling. Or, even choose an inverse hyperbolic sine scaling
with with ASINHSCL:
http://www.dfanning.com/ip_tips/xstretch.html
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
|
|
|
|
|
|
| Re: Philosophical Scaling Question [message #51600 is a reply to message #51599] |
Mon, 04 December 2006 08:58 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Paolo Grigis writes:
> Well, as I see it, you are scaling the wrong quantity:
> it is not the *values* of r,g,b that should be scaled,
> but their *indices*... try:
>
> ind=bindgen(256)
> TVLCT, r[LogScl(ind)], g[LogScl(ind)], b[LogScl(ind)]
Yes, I just came to that conclusion myself. This works
as expected. Thank you!
Now the question remains: How would you make a
logarithmic color bar in a general way that would
accommodate various log scalings? I believe I could
do it if I were also displaying the image, but I
have grave doubts about it being possible to do it
generally.
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
| Re: Philosophical Scaling Question [message #51601 is a reply to message #51600] |
Mon, 04 December 2006 08:51 |
Brian Larsen
Messages: 270
Registered: June 2006
|
Senior Member |
|
|
This is a good question.
to ask another question... why when you add colobar after each of the
tvimage calls the colorbar is the same in windows 1 and 2?
Maybe this is my lack of understanding of what logscl does.
Brian
Loadct, 33, /Silent
TVLCT, r, g, b, /Get
pmm, r
pmm, g
pmm, b
image = Loaddata(7)
pmm, image
Window, XSize=400, YSize=350, 0
Loadct, 33, /Silent
TVImage, image
colorbar
Window, XSize=400, YSize=350, 1
Loadct, 33, /Silent
TVImage, LogScl(image)
colorbar
Window, XSize=400, YSize=350, 2
Loadct, 33, /Silent
TVLCT, r, g, b, /Get
TVLCT, logscl(r), logscl(g), logscl(b)
tvimage, image
colorbar
David Fanning wrote:
> Folks,
>
> With no answers to my weekend questions about logarithmic
> color bars I'm flying blind this morning. And I seem to be
> running into theoretical difficulties. Can anyone help?
>
> Suppose I had a color table (color table 33 comes to mind)
> where each color vector had a min of 0 and a max 0f 255.
>
> IDL> Loadct, 33, /Silent
> IDL> TVLCT, r, g, b, /Get
> IDL> MinMax, r
> 0 255
> IDL> MinMax, g
> 0 255
> IDL> MinMax, b
> 0 255
>
> And suppose I also have an image that is scaled in
> the same way:
>
> IDL> image = Loaddata(7)
> IDL> MinMax, image
> 0 255
>
> And finally, suppose I have a way to scale such data
> sets in a logametric way, say a function LOGSCL.
>
> IDL> .compile LOGSCL
> Compiled module: LOGSCL.
>
> My hypothesis is that there are two ways to display this
> data "logarithmically". I can leave the color table vectors
> alone, and scale the image data. Or, I can leave the image
> alone and scale the color vectors. Either way should result
> in exactly the same display.
>
> The problem is, it doesn't. :-(
>
> Window, XSize=400, YSize=350, 0
> Loadct, 33, /Silent
> TVImage, image
>
> Window, XSize=400, YSize=350, 1
> Loadct, 33, /Silent
> TVImage, LogScl(image)
>
> Window, XSize=400, YSize=350, 2
> Loadct, 33, /Silent
> TVLCT, r, g, b, /Get
> TVLCT, LogScl(r), LogScl(g), LogScl(b)
> TVImage, image
>
> Does anyone have a good idea for why not?
>
> Cheers,
>
> David
>
> P.S. And please don't tell me there is something wrong
> with LOGSCL, as this is *not* the answer I want to
> hear. :-(
>
> --
> David Fanning, Ph.D.
> Fanning Software Consulting, Inc.
> Coyote's Guide to IDL Programming: http://www.dfanning.com/
> Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
| Re: Philosophical Scaling Question [message #51602 is a reply to message #51601] |
Mon, 04 December 2006 08:48 |
Paolo Grigis
Messages: 171
Registered: December 2003
|
Senior Member |
|
|
Well, as I see it, you are scaling the wrong quantity:
it is not the *values* of r,g,b that should be scaled,
but their *indices*... try:
ind=bindgen(256)
TVLCT, r[LogScl(ind)], g[LogScl(ind)], b[LogScl(ind)]
instead.
Ciao,
Paolo
David Fanning wrote:
> Folks,
>
> With no answers to my weekend questions about logarithmic
> color bars I'm flying blind this morning. And I seem to be
> running into theoretical difficulties. Can anyone help?
>
> Suppose I had a color table (color table 33 comes to mind)
> where each color vector had a min of 0 and a max 0f 255.
>
> IDL> Loadct, 33, /Silent
> IDL> TVLCT, r, g, b, /Get
> IDL> MinMax, r
> 0 255
> IDL> MinMax, g
> 0 255
> IDL> MinMax, b
> 0 255
>
> And suppose I also have an image that is scaled in
> the same way:
>
> IDL> image = Loaddata(7)
> IDL> MinMax, image
> 0 255
>
> And finally, suppose I have a way to scale such data
> sets in a logametric way, say a function LOGSCL.
>
> IDL> .compile LOGSCL
> Compiled module: LOGSCL.
>
> My hypothesis is that there are two ways to display this
> data "logarithmically". I can leave the color table vectors
> alone, and scale the image data. Or, I can leave the image
> alone and scale the color vectors. Either way should result
> in exactly the same display.
>
> The problem is, it doesn't. :-(
>
> Window, XSize=400, YSize=350, 0
> Loadct, 33, /Silent
> TVImage, image
>
> Window, XSize=400, YSize=350, 1
> Loadct, 33, /Silent
> TVImage, LogScl(image)
>
> Window, XSize=400, YSize=350, 2
> Loadct, 33, /Silent
> TVLCT, r, g, b, /Get
> TVLCT, LogScl(r), LogScl(g), LogScl(b)
> TVImage, image
>
> Does anyone have a good idea for why not?
>
> Cheers,
>
> David
>
> P.S. And please don't tell me there is something wrong
> with LOGSCL, as this is *not* the answer I want to
> hear. :-(
>
|
|
|
|
|
|
| Re: Philosophical Scaling Question [message #51604 is a reply to message #51603] |
Mon, 04 December 2006 08:25 |
Foldy Lajos
Messages: 268
Registered: October 2001
|
Senior Member |
|
|
On Mon, 4 Dec 2006, David Fanning wrote:
> Folks,
>
> With no answers to my weekend questions about logarithmic
> color bars I'm flying blind this morning. And I seem to be
> running into theoretical difficulties. Can anyone help?
>
> Suppose I had a color table (color table 33 comes to mind)
> where each color vector had a min of 0 and a max 0f 255.
>
> IDL> Loadct, 33, /Silent
> IDL> TVLCT, r, g, b, /Get
> IDL> MinMax, r
> 0 255
> IDL> MinMax, g
> 0 255
> IDL> MinMax, b
> 0 255
>
> And suppose I also have an image that is scaled in
> the same way:
>
> IDL> image = Loaddata(7)
> IDL> MinMax, image
> 0 255
>
> And finally, suppose I have a way to scale such data
> sets in a logametric way, say a function LOGSCL.
>
> IDL> .compile LOGSCL
> Compiled module: LOGSCL.
>
> My hypothesis is that there are two ways to display this
> data "logarithmically". I can leave the color table vectors
> alone, and scale the image data. Or, I can leave the image
> alone and scale the color vectors. Either way should result
> in exactly the same display.
>
> The problem is, it doesn't. :-(
>
> Window, XSize=400, YSize=350, 0
> Loadct, 33, /Silent
> TVImage, image
>
> Window, XSize=400, YSize=350, 1
> Loadct, 33, /Silent
> TVImage, LogScl(image)
>
> Window, XSize=400, YSize=350, 2
> Loadct, 33, /Silent
> TVLCT, r, g, b, /Get
> TVLCT, LogScl(r), LogScl(g), LogScl(b)
> TVImage, image
>
> Does anyone have a good idea for why not?
>
> Cheers,
>
> David
>
> P.S. And please don't tell me there is something wrong
> with LOGSCL, as this is *not* the answer I want to
> hear. :-(
>
> --
> David Fanning, Ph.D.
> Fanning Software Consulting, Inc.
> Coyote's Guide to IDL Programming: http://www.dfanning.com/
> Sepore ma de ni thui. ("Perhaps thou speakest truth.")
>
Probably log(0)=-inf plays a role in your experiment.
Does LOGSCL handle it?
regards,
lajos
|
|
|
|
| Re: Philosophical Scaling Question [message #51732 is a reply to message #51593] |
Mon, 04 December 2006 12:43 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Brian Larsen writes:
>> LoadCT, 33
>> image = LoadData(7)
>> TVLCT, r, g, b, /GET
>> data = Scale_Vector(image, 1, 1000) ; Data in log scale.
>> pos = [0.1, 0.1, 0.9, 0.7]
>> TVImage, LogScl(data), Position=pos, /Keep, /Erase
>> index = Bindgen(256)
>>
>> TVLCT, r[LogScl(index)], g[LogScl(index)], b[LogScl(index)]
>> Colorbar, Range=[1,1000], XLOG=1, Divisions=3, $
>> Position=[pos[0], 0.87, pos[2], 0.93], Minor=5
>>
>
> Ah ha, this exactly what I wanted colorbar to do!! The reason I had
> never been able to get it done is that was a lot of work (and work I
> didn't know how to do), it seems like a flag to colorbar or tvlct or
> something that was /logscl would do the trick. Really the issue was my
> unfamiliarity with logscl, tvlct, and tvimage and how to put it all
> together.
Don't get too excited about this yet. It looks beautiful.
I'm just not sure it is correct. :-)
I'm trying to write an article about this, and I keep
running into complications. It is a lot more difficult
than I expected it to be.
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
| Re: Philosophical Scaling Question [message #51739 is a reply to message #51593] |
Mon, 04 December 2006 10:39 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Brian Larsen writes:
> Ah ha, this exactly what I wanted colorbar to do!! The reason I had
> never been able to get it done is that was a lot of work (and work I
> didn't know how to do), it seems like a flag to colorbar or tvlct or
> something that was /logscl would do the trick. Really the issue was my
> unfamiliarity with logscl, tvlct, and tvimage and how to put it all
> together.
Alas, the problem is for every simple case, there are an infinite
number of more complex cases. And as soon as someone like me
distributes software, I hear from EVERYONE who wants to
do something complex. :-(
I can't keep up. So I try to provide tools that do simple
things (which I can support), and which can be combined in
complex ways (which I can't support, generally).
Note the difference between this philosophy and the
thinking behind iTools, for example. Using my tools
an "ah ha" moment can be experienced in less than 10
lines of code (as you just discovered). Using iTools,
your "ah ha" moment can be a long way off (as I discover
just about every time I try to use one of those darn things).
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
| Re: Philosophical Scaling Question [message #51741 is a reply to message #51593] |
Mon, 04 December 2006 10:27 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Brian Larsen writes:
> Ah ha, this exactly what I wanted colorbar to do!! The reason I had
> never been able to get it done is that was a lot of work (and work I
> didn't know how to do), it seems like a flag to colorbar or tvlct or
> something that was /logscl would do the trick. Really the issue was my
> unfamiliarity with logscl, tvlct, and tvimage and how to put it all
> together.
I'll put you on the list of people who want to buy the new book. :-)
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
