Re: filmscanners: VueScan Histogram

[Bob Armstrong]

Larry wrote:

> I don't think it is necessary to adjust the white/black pts. by moving
> arrows on the histogram.  It would be nice to see the output of the
> preview as a histogram plot and then continue adjusting using the
> standard color tab items.


Yes, I agree.

Bob

Re: filmscanners: VueScan Histogram

[Colin Maddock]

I don't feel the need for a histogram in VueScan either. Setting the black and white 
clipping levels does all that is needed at the scanning stage, along with control of 
gamma.

A separate point. I liked having the picture, preview or scan, occupy the whole 
screen. It gave a more satisfying "first encounter" with the scan than the new setup.

Colin Maddock
  

Re: filmscanners: VueScan Histogram

[Hersch Nitikman]

Poor Ed. He can't win. Anything he does, no matter how many wanted it, some 
others will express disappointment. Personally, I like a histogram, and 
feel it helps me understand an image, and helps me improve it. And, on the 
current 7.3 versions, if I let VS have the whole screen, I get a very nice 
big image in the preview.
My only complaint is that I have found the Clean function less effective 
than it used to be with my PC-based LS-30. I have no idea why.
Hersch

At 10:07 AM 12/15/2001 +1300, you wrote:
>I don't feel the need for a histogram in VueScan either. Setting the black 
>and white clipping levels does all that is needed at the scanning stage, 
>along with control of gamma.
>
>A separate point. I liked having the picture, preview or scan, occupy the 
>whole screen. It gave a more satisfying "first encounter" with the scan 
>than the new setup.
>
>Colin Maddock
>

RE: filmscanners: VueScan Histogram

[Hemingway, David J]

It reminds me of the "Insight wars" Polaroid was beaten up mercilessly for
not having histograms and not just any histogram, it had to have the little
sliders on the bottom like Photoshop.
When I asked for new features Histograms was ALWAYS on the top of the list.
Shit, Ed's getting off easy!!  :)
David

 -Original Message-
From:   Hersch Nitikman [mailto:[EMAIL PROTECTED]] 
Sent:   Friday, December 14, 2001 10:12 PM
To: [EMAIL PROTECTED]
Subject:Re: filmscanners: VueScan Histogram

Poor Ed. He can't win. Anything he does, no matter how many wanted it, some 
others will express disappointment. Personally, I like a histogram, and 
feel it helps me understand an image, and helps me improve it. And, on the 
current 7.3 versions, if I let VS have the whole screen, I get a very nice 
big image in the preview.
My only complaint is that I have found the Clean function less effective 
than it used to be with my PC-based LS-30. I have no idea why.
Hersch

At 10:07 AM 12/15/2001 +1300, you wrote:
>I don't feel the need for a histogram in VueScan either. Setting the black 
>and white clipping levels does all that is needed at the scanning stage, 
>along with control of gamma.
>
>A separate point. I liked having the picture, preview or scan, occupy the 
>whole screen. It gave a more satisfying "first encounter" with the scan 
>than the new setup.
>
>Colin Maddock
>

Re: filmscanners: VueScan Histogram

[Julian Vrieslander]

I will add my vote to those VueScan users who do *not* want it to become 
an image editing application.  I would rather do my tweaks in Photoshop.

But I do encourage Ed to continue development of features which enable 
VueScan to save files with the best possible bits.  I think a histogram 
display is useful for this.

With the current version of VueScan I need to set the white point and 
black points by guesswork.  I pick a pair of percentage numbers and hope 
for the best, or if I have time, I will do a few Prev Mem or Scan Mem 
trials to see the effects of various settings.  But often, when I bring 
the scan into Photoshop and check the histograms, I discover that I 
chopped off the highlights or shadows.  Or I might have set VueScan's 
limits too conservatively, and there are a lot of empty bins in the 
histo, with the image data squished into a narrow range of tonal values.

Maybe with more experience I will get better at inspecting VueScan's 
displays and choosing the right values for WP, BP, and gamma.  But since 
these displays are not color managed, I also have to mentally compensate 
for how the image appearance is going to change when it goes into 
Photoshop.  That's a lot for my poor addled brain to handle.

A histogram would help, because the shape of the data distribution is 
often more informative than the clip percentages.  Ideally, VueScan would 
have the tried-and-true histo interface which provides draggable markers 
for WP, BP and gamma, with live image updates.  But I know Ed does not 
like to spend a lot of time implementing complex UI elements that require 
platform specific code.  If Ed retains the numeric entry UI, I suggest 
that the histogram has a calibrated axis for levels (e.g., 0-255 for 8 
bit data) and that the WP and BP can be entered using values from this 
axis (rather than percentages). 

--
Julian Vrieslander 

Re: filmscanners: VueScan Histogram

[Bob Armstrong]

Michael wrote:

>   Up until the latest versions I always put a hard number (%) in the Wp & Bp
> ... e.g., '0.1' or 0.01%... and because changing this number by minute
> amount changed the image more than you'd think, I had Vs write a small JPG
> direct to Photoshop (note ... such a small change in Wp% would never be
> gleened from visually inspecting a preview histogram).  When satisfied, I
> had Vs write a TIF.  Regarding newer versions, I understand Ed has changed
> what '0' means in the Bp field ... and I have to again evaluate what this
> means to the histogram.

I'm glad to see that we are using a similar process; I haven't read of any other VS 
user describe their scanning workflow including the use of a histogram.  (I get VS to 
send the full TIF image to VuePrint and look at the histogram there.  The trouble with 
this is the time taken to save the file, change B/W points and rescan until the image 
fills the range you want makes it impractical for regular use.)  

Your idea of using a JPG is a good one.  That will save time but it still won't be as 
efficient as a histogram in VS.

Agreed that very small changes in B/W points make very big changes in the image.  I 
don't find VS a very intuitive application to use but have spent some time getting to 
know it.  It was this that led me to think that using a histogram was a good way of 
determining B/W points in VS and so help to get the most from a scan.  I certainly 
don't want or expect VS to be used for image editing. 

VueScan gives me good results but I just don't have the time to scan the images I want 
to scan.

Bob

Re: filmscanners: VueScan Histogram

[DRP]

>> I don't think it is necessary to adjust the white/black pts. by moving
>> arrows on the histogram.  It would be nice to see the output of the
>> preview as a histogram plot and then continue adjusting using the
>> standard color tab items.

I agree, moving arrows are not necessary, but a Vs histogram would really be
time saving for those who use to preview, then scan, then  look at the
histogram inside Ps, then readjust color tabs items and rescan in Vs...

Would a Vs preview histogram difficult to build?

Best regards.

Didier

Re: filmscanners: VueScan Histogram

[Julian Vrieslander]

Looks good so far, Ed.  I like the layout, with the combined histo on 
top, and the RGB histos superimposed.

I'm sure there will be a raging torrent of suggestions on more gongs and 
whistles to add.  I would like to see some way of connecting the 
histograms to the numbers entered in the WP, BP, and gamma settings, but 
I am not sure about the best way to do this.  I like the way it is done 
in Nikon Scan, with sliders, numeric fields, and a button to toggle 
between before/after histos.  But that's a bit complex.

BTW, in my neck of the woods "slider" is the term we use to decribe 
greasy hamburgers.

Here's one request that I think makes sense, and which would be easy to 
implement.  If the user is viewing the Prev Hist tab and then executes a 
Preview or Prev Mem, leave the Prev Hist tab in front, rather than 
switching to the Preview tab.  Chances are, the user wants to see the 
effects of the new scan on the histogram.  Similarly, if Scan Hist is in 
front, leave it in front if the user does a Scan or Scan Mem.  My own 
preference would be for disabling automatic flipping of the display and 
histo tabs altogether, and let me choose when to flip them.

--
Julian Vrieslander 

Re: filmscanners: VueScan Histogram

[Mark T.]

I would second Julian's comments - expecially the part about leaving the 
histograms in view for the Preview Memory function - that's the first thing 
I tried to do and was a little frustrated by not being able to see what 
happened until I had hit the tab again.  Ideally leave the old histo there 
until the moment that it changes - then when we see the actual 'switch' 
take place I think we will quickly get the hang of the relationship between 
the numbers and the graph.

mark t


PS - I find it hard to believe we are all sitting here having our 
needs/wants met by a software developer, often in hours, let alone months 
or years..  Ed, I hope you never learn the 'normal' way to develop 
software...  :-)


Julian wrote:
>Looks good so far, Ed.  I like the layout, with the combined histo on
>top, and the RGB histos superimposed.
>
>I'm sure there will be a raging torrent of suggestions on more gongs and
>whistles to add.  I would like to see some way of connecting the
>histograms to the numbers entered in the WP, BP, and gamma settings, but
>I am not sure about the best way to do this.
..
>Here's one request that I think makes sense, and which would be easy to
>implement.  If the user is viewing the Prev Hist tab and then executes a
>Preview or Prev Mem, leave the Prev Hist tab in front, rather than
>switching to the Preview tab.  Chances are, the user wants to see the
>effects of the new scan on the histogram.  Similarly, if Scan Hist is in
>front, leave it in front if the user does a Scan or Scan Mem.  My own
>preference would be for disabling automatic flipping of the display and
>histo tabs altogether, and let me choose when to flip them.

Re: filmscanners: VueScan Histogram

[EdHamrick]

In a message dated 12/16/2001 3:34:28 AM EST, [EMAIL PROTECTED] 
writes:

> I would second Julian's comments - expecially the part about leaving the 
>  histograms in view for the Preview Memory function

It's fixed in 7.3.5, which I'll release in the next day or so.

Regards,
Ed Hamrick

Re: filmscanners: VueScan Histogram

[EdHamrick]

In a message dated 12/16/2001 4:15:12 AM EST, [EMAIL PROTECTED] writes:

> Just a question, is the y-axis in the histograms linear or log?

It's linear.  Would people prefer it as a log scale?

What scale do other programs use?

Regards,
Ed Hamrick

Re: filmscanners: VueScan Histogram

[Julian Vrieslander]

On 12/16/01 4:33 AM, Mark T. <[EMAIL PROTECTED]>, wrote:

>PS - I find it hard to believe we are all sitting here having our 
>needs/wants met by a software developer, often in hours, let alone months 
>or years..  Ed, I hope you never learn the 'normal' way to develop 
>software...  :-)

Yeah, I think we should clone Ed and send a few copies to Microsoft, 
Apple, Sun, etc.

--
Julian Vrieslander 

Re: filmscanners: VueScan Histogram

[sirius]

i like it linear, it shows differences in comprehensible proportions.
i suppose the histo in pshop is also linear.

- Original Message -
From: <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Sunday, December 16, 2001 10:36 AM
Subject: Re: filmscanners: VueScan Histogram


> In a message dated 12/16/2001 4:15:12 AM EST, [EMAIL PROTECTED]
writes:
>
> > Just a question, is the y-axis in the histograms linear or log?
>
> It's linear.  Would people prefer it as a log scale?
>
> What scale do other programs use?
>
> Regards,
> Ed Hamrick
>

Re: filmscanners: VueScan Histogram

[Bob Armstrong]

Ed wrote:

> I think the best way to do this is for me to just add a histogram and
> then evolve it using people's suggestions.
> 
> It's quite possible that a simple histogram of the preview (within
> the crop window) and the scan will be good enough.  I can worry
> about fancy stuff like movable sliders later.


Ed, I've got 7.3.4 and have been trying out the histogram function and I am sure this 
is going to save me a lot of time.

One thing that has struck me is that one may be able to use the 'Prev Hist' and 'Scan 
Hist' as before and after histograms if the preview and scan are set to the same 
resolution.  This might be of help in guesstimating the numbers to enter as B/W 
points.  It may also be useful in understanding how changes in settings affect the 
image.  

I know the above is very much 'by the numbers' rather than being focused (oops, no pun 
intended) on the image but I do agree with Maris saying 'I want a good image, not a 
good histogram'.  My point is, Ed has given us a tool which helps our scanning and 
also helps in understanding the effect of adjusting all those options he tries to 
confuse me with ;)

Thanks very much for the histograms Ed.

Bob
looking for 7.3.5

Re: filmscanners: VueScan Histogram

[David Gordon]

Colin Maddock [[EMAIL PROTECTED]] wrote on Sat, 15 Dec 2001 10:07:11 +1300

>I don't feel the need for a histogram in VueScan either. Setting the
>black and white clipping levels does all that is needed at the scanning
>stage, along with control of gamma.

How do you know where to clip the black and white points?

-- 
David Gordon
[EMAIL PROTECTED]

Re: filmscanners: VueScan Histogram

[David Gordon]

Julian Vrieslander [[EMAIL PROTECTED]] wrote on Sat, 15 Dec
2001 00:33:41 -0500

>Maybe with more experience I will get better at inspecting VueScan's 
>displays and choosing the right values for WP, BP, and gamma.  But since 
>these displays are not color managed, I also have to mentally compensate 
>for how the image appearance is going to change when it goes into 
>Photoshop.

What I see in the VueScan window is what I get in Photoshop - am I doing
something right!

-- 
David Gordon
[EMAIL PROTECTED]

RE: filmscanners: VueScan Histogram

[michael shaffer]

Ed writes ...

> [EMAIL PROTECTED] writes:
>
> > Just a question, is the y-axis in the histograms linear or log?
>
> It's linear.  Would people prefer it as a log scale?
>
> What scale do other programs use?

  Others are linear ... but I have to admit, a log scale would be more
accurate as the data approached the Wp and Bp.  (... Hmmm, maybe not
accurate, but at least visible ...)  If we are worried about setting the
endpoints accurately and visually, maybe we ought to consider ... either a
log scale, or the ability to zoom in on the y-axis(?)

shAf  :o)

Re: filmscanners: VueScan Histogram

[EdHamrick]

In a message dated 12/16/2001 9:46:17 AM EST, [EMAIL PROTECTED] 
writes:

> > What scale do other programs use?
>  
>Others are linear ... but I have to admit, a log scale would be more
>  accurate as the data approached the Wp and Bp.  (... Hmmm, maybe not
>  accurate, but at least visible ...)  If we are worried about setting the
>  endpoints accurately and visually, maybe we ought to consider ... either a
>  log scale, or the ability to zoom in on the y-axis(?)

I've made the histogram type an option in the Prefs tab.  The options
are Linear, Square root and Logarithmic.  I'll release this in VueScan
7.3.5.

Regards,
Ed Hamrick

RE: filmscanners: VueScan Histogram

[michael shaffer]

Ed writes ...

> I've made the histogram type an option in the Prefs tab.
> The options are Linear, Square root and Logarithmic.  I'll
> release this in VueScan 7.3.5.


  What exactly are you plotting on the y-axis? ... e.g., any RGB value >0 &
<255, or some calculated luminosity value? Raw RGB, or color corrected RGB??
(... hopefully knowing I'm clipping (or including) the real whitepoint and
blackpoint(?) ...)

  (Personally, I appreciate asking these types of questions.)

cheerios ... shAf  :o)
Avalon Peninsula, Newfoundland

Re: filmscanners: VueScan Histogram

[Maris V. Lidaka, Sr.]

You cover it all!

Maris

- Original Message - 
From: <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Sunday, December 16, 2001 8:47 AM
Subject: Re: filmscanners: VueScan Histogram


| In a message dated 12/16/2001 9:46:17 AM EST, [EMAIL PROTECTED] 
| writes:
| 
| > > What scale do other programs use?
| >  
| >Others are linear ... but I have to admit, a log scale would be more
| >  accurate as the data approached the Wp and Bp.  (... Hmmm, maybe not
| >  accurate, but at least visible ...)  If we are worried about setting the
| >  endpoints accurately and visually, maybe we ought to consider ... either a
| >  log scale, or the ability to zoom in on the y-axis(?)
| 
| I've made the histogram type an option in the Prefs tab.  The options
| are Linear, Square root and Logarithmic.  I'll release this in VueScan
| 7.3.5.
| 
| Regards,
| Ed Hamrick
| 

RE: filmscanners: VueScan Histogram

[Austin Franklin]

> I've made the histogram type an option in the Prefs tab.  The options
> are Linear, Square root and Logarithmic.  I'll release this in VueScan
> 7.3.5.
> 

Hi Ed,

Why not make it selectable in the histogram window?

Regards,

Austin

Re: filmscanners: VueScan Histogram

[Julian Vrieslander]

On 12/16/01 8:03 AM, David Gordon <[EMAIL PROTECTED]>, wrote:

>Julian Vrieslander [[EMAIL PROTECTED]] wrote on Sat, 15 Dec
>2001 00:33:41 -0500
>
>>Maybe with more experience I will get better at inspecting VueScan's 
>>displays and choosing the right values for WP, BP, and gamma.  But since 
>>these displays are not color managed, I also have to mentally compensate 
>>for how the image appearance is going to change when it goes into 
>>Photoshop.
>
>What I see in the VueScan window is what I get in Photoshop - am I doing
>something right!

Maybe you are running on a PC and using sRGB as your color space.  If so, 
a color managed display is less important.  I run on a Mac with a gamma 
1.8 monitor, and I prefer to use Adobe RGB as my color space.  With 
VueScan set to Adobe RGB, images appear very different than how they 
appear in Photoshop: the VueScan version is very flat and desaturated.

I've figured out a workflow that gives me a somewhat more useful display 
in VueScan.  I set color space to Apple RGB and gamma 1.8 for my first 
look at the scan.  I set crop, exposure, white point, black point, 
brightness, and filter options, using Prev Mem and Scan Mem to check 
results in Apple RGB.  Then I change to Adobe RGB and gamma 2.2 (keeping 
other settings the same), and I do a Scan Mem to write the final output 
file.  This two-space two-step takes extra time, and it still does not 
give a really good match with what I see in Photoshop.  But it's the best 
I can do with the current version of VueScan.

--
Julian Vrieslander 

Re: filmscanners: VueScan Histogram

[Bernie Ess]

Could someone explain to me in a few words how to work with a histogram when
scanning? I can see it in VueScan but I don´t know what to do with it,
because there are no sliders anyway...

Thanks in advance,
Bernhard

Re: filmscanners: VueScan Histogram

[Colin Maddock]

David Gordon asked:  How do you know where to clip the black and white points?

I always leave black clipping at 0, and white clipping at a minimal amount, perhaps 
0.1, so that the VueScan scan contains all the tonal range that was on the neg/slide. 
Any further clipping of the end points for cosmetic reasons etc can be done as part of 
the operations in the image editing program, using, say, the Info facility to search 
the levels.

Colin Maddock
 

Re: filmscanners: VueScan Histogram

[David Gordon]

Julian Vrieslander [[EMAIL PROTECTED]] wrote on Sun, 16 Dec
2001 12:31:34 -0500

>Maybe you are running on a PC and using sRGB as your color space.  If so, 
>a color managed display is less important.

An outrageous suggestion, you shall be hearing from my lawyers in the
morning... ;)

>I run on a Mac with a gamma 
>1.8 monitor, and I prefer to use Adobe RGB as my color space.

Me to!

> With 
>VueScan set to Adobe RGB, images appear very different than how they 
>appear in Photoshop: the VueScan version is very flat and desaturated.

You seem to have a roundabout way of getting the file into Photoshop! How
are your ColorSync settings? I too am using Adobe RBB (1998), when I've
finnished fiddling in Vuescan the final preview looks the same as the
file when opened in PS. 

Isn't that the way it's meant to work, Ed?

-- 
David Gordon
[EMAIL PROTECTED]

Re: filmscanners: VueScan Histogram

[Julian Vrieslander]

On 12/16/01 4:19 PM, David Gordon <[EMAIL PROTECTED]>, wrote:

>Julian Vrieslander [[EMAIL PROTECTED]] wrote on Sun, 16 Dec
>2001 12:31:34 -0500
>
>>Maybe you are running on a PC and using sRGB as your color space.  If so, 
>>a color managed display is less important.
>
>An outrageous suggestion, you shall be hearing from my lawyers in the
>morning... ;)

No problem.  The white box on the porch is for subpoenas, the red box is 
for injunctions, the blue box is for arrest and search warrants.

>>I run on a Mac with a gamma 
>>1.8 monitor, and I prefer to use Adobe RGB as my color space.
>
>Me to!
>
>> With 
>>VueScan set to Adobe RGB, images appear very different than how they 
>>appear in Photoshop: the VueScan version is very flat and desaturated.
>
>You seem to have a roundabout way of getting the file into Photoshop! How
>are your ColorSync settings? I too am using Adobe RBB (1998), when I've
>finnished fiddling in Vuescan the final preview looks the same as the
>file when opened in PS. 

Do you mean the settings in the ColorSync control panel?  I don't see 
that they have much to do with this.  My monitor is profiled (formerly 
with Adobe Gamma, now with a Colorvision Spyder), and the correct profile 
is selected in ColorSync CP.  Once my scans are in Photoshop 6, my 
workflow is color managed, including screen displays, file writes, and 
prints on an Epson 1280.

Why do I set VueScan to use Apple RGB color space (with 1.8 gamma)?  
Because this provides the most accurate rendition of image brightness and 
saturation on a non-color-managed Mac application driving a 1.8 gamma, 
6500K monitor.  VueScan's default color space (sRGB) does not match my 
monitor setup and renders the image too dark.  Using Adobe RGB in VueScan 
makes the image look washed out and desaturated.

I switch VueScan to Adobe RGB (and 2.2 gamma) only when I have finished 
my adjustments in the color and filter tabs, and before writing the TIF 
file.  When I open the TIF file in Photoshop, its calibrated displays 
show the image with approximately correct saturation and contrast.  It's 
not a perfect match to the appearance i saw in VueScan, using Apple RGB, 
because VueScan was not using my monitor profile.

Are you claiming that your scans look the same in VueScan and Photoshop 
with both configured for Adobe RGB?  When I compare those displays, the 
difference is huge and unmistakable.  If I adjust the image to look 
correct in VueScan under Adobe RGB, it looks totally wrong in Photoshop.  
I can only think of one possible explanation.  Are you using Photoshop 5 
or 5.5, with "Display using monitor calibration" disabled in the prefs?  
If so, you have monitor color management disabled, and I would expect the 
Photoshop and VueScan displays to look similar.  

--
Julian Vrieslander 

RE: filmscanners: VueScan Histogram

[Jawed Ashraf]

Adobe RGB has (ironically enough, this is after all a "Mac" company!) a
built-in gamma of 2.2.  Therefore on a Mac operating at gamma 1.8, with an
application (Vuescan?) that isn't performing any translation of the image
from source colour space into monitor colour space, Adobe RGB will look
"wrong".  Not to mention that Adobe RGB also has a larger gamut than
monitors can show.  So, not only does the image look wrong tonally, but
saturation is also incorrect.

Obviously, when an image in Adobe RGB is viewed in a profile-aware
application, such as Photoshop, a translation to the monitor gamut *and*
gamma is performed.  So in these circumstances the image will look "right".
Except there's a chance that some extremely saturated colours (I think
yellows are problematic on monitors) will be noticeably "wrong".   It is
simply because the monitor can't cope.

Jawed

> -Original Message-
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED]]On Behalf Of Julian
> Vrieslander
> Sent: 16 December 2001 17:32
> To: [EMAIL PROTECTED]
> Subject: Re: filmscanners: VueScan Histogram
>
>
> On 12/16/01 8:03 AM, David Gordon <[EMAIL PROTECTED]>, wrote:
>
> >Julian Vrieslander [[EMAIL PROTECTED]] wrote on Sat, 15 Dec
> >2001 00:33:41 -0500
> >
> >>Maybe with more experience I will get better at inspecting VueScan's
> >>displays and choosing the right values for WP, BP, and gamma.
> But since
> >>these displays are not color managed, I also have to mentally
> compensate
> >>for how the image appearance is going to change when it goes into
> >>Photoshop.
> >
> >What I see in the VueScan window is what I get in Photoshop - am I doing
> >something right!
>
> Maybe you are running on a PC and using sRGB as your color space.  If so,
> a color managed display is less important.  I run on a Mac with a gamma
> 1.8 monitor, and I prefer to use Adobe RGB as my color space.  With
> VueScan set to Adobe RGB, images appear very different than how they
> appear in Photoshop: the VueScan version is very flat and desaturated.
>
> I've figured out a workflow that gives me a somewhat more useful display
> in VueScan.  I set color space to Apple RGB and gamma 1.8 for my first
> look at the scan.  I set crop, exposure, white point, black point,
> brightness, and filter options, using Prev Mem and Scan Mem to check
> results in Apple RGB.  Then I change to Adobe RGB and gamma 2.2 (keeping
> other settings the same), and I do a Scan Mem to write the final output
> file.  This two-space two-step takes extra time, and it still does not
> give a really good match with what I see in Photoshop.  But it's the best
> I can do with the current version of VueScan.
>
> --
> Julian Vrieslander <mailto:[EMAIL PROTECTED]>
>
>

Re: filmscanners: VueScan Histogram

[P Elkin]

- Original Message -
From: "Colin Maddock" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Sunday, December 16, 2001 7:32 PM
Subject: Re: filmscanners: VueScan Histogram


> David Gordon asked:  How do you know where to clip the black and white
points?
>
> I always leave black clipping at 0, and white clipping at a minimal
amount, perhaps 0.1, so that the VueScan scan contains all the tonal range
that was on the neg/slide.

*From my observations if you just scan with the WP at 0.1 as default, you
can compress the tonal values rather than making sure the image ocupies the
full range available. If your image allows a WP of say 0.4 just before
clipping then the available range isn't wasted in the 0-0.4 area*

Any further clipping of the end points for cosmetic reasons etc can be done
as part of the operations in the image editing program, using, say, the Info
facility to search the levels.

*Making sure the actual points at just before clipping are fully filling the
histogram maximises the image quality output to file etc. I also found
before the histograms were implimented that the best visual tonal range on
screen ie unclipped whites but not vailed or flat as is often the case with
WP <0.3 actually do fill the histogram perfectly. The image editor then can
do the 'cosmetics' if required on an image mapped to make use of the full
available dynamic range.*

Philip Elkin

RE: filmscanners: VueScan Histogram

[michael shaffer]

Philip writes ...

> *From my observations if you just scan with the WP at 0.1
> as default, you can compress the tonal values rather than
> making sure the image ocupies the full range available.
> If your image allows a WP of say 0.4 just before clipping
> then the available range isn't wasted in the 0-0.4 area*
> ...

  That's my impression as well.  For example, any given photo depending
exposure and subject, the scan may be asked to include "white without
detail" into the histogram ... which may sacrifice levels for "white with
detail".  I believe as you force the histogram to include the former and
play with the Wp=0-.4 region, you'll notice significant effect on brightness
(or apparent gamma).  Once you set Wp so as to exclude most "white w/o
detail" you'll see less effect ... and it is here where you should probably
set the Wp.  A good histogram presentation, which might display a "small
bump" due to specular highlights would be beneficial, and reduce time wasted
at guessing.

shAf  :o)

RE: filmscanners: VueScan Histogram

[michael shaffer]

Ed writes ...

> ...
> I've made the histogram type an option in the Prefs tab.  The options
> are Linear, Square root and Logarithmic.  ...

  The upper "luminosity(?)" histogram shows an uncomfortable likeness to the
"green" histogram(?)  How is the upper histogram calculated? (... surely not
the "green weighted" grayscale conversion ...)

shAf  :o)

Re: filmscanners: VueScan Histogram

[EdHamrick]

In a message dated 12/17/2001 10:46:16 AM EST, [EMAIL PROTECTED] 
writes:

> How is the upper histogram calculated?

m_hst[3][(306L*pix[0]+601L*pix[1]+117L*pix[2])>>18]++;


Regards,
Ed Hamrick

Re: filmscanners: VueScan Histogram

[John Rylatt]

Now, even I can understand that!!

John.


[EMAIL PROTECTED] wrote:
> 
> In a message dated 12/17/2001 10:46:16 AM EST, [EMAIL PROTECTED]
> writes:
> 
> > How is the upper histogram calculated?
> 
> m_hst[3][(306L*pix[0]+601L*pix[1]+117L*pix[2])>>18]++;
> 
> Regards,
> Ed Hamrick

RE: filmscanners: VueScan Histogram

[michael shaffer]

Ed writes ...

> In a message dated 12/17/2001 10:46:16 AM EST, [EMAIL PROTECTED]
> writes:
>
> > How is the upper histogram calculated?
>
> m_hst[3][(306L*pix[0]+601L*pix[1]+117L*pix[2])>>18]++;

  at 1st guess ... it is "weighted green" ... a normalized
(30.6%RED + 60.1%GREEN + 11.7%BLUE) ...

  ... and I agree with weighted green being one of the best methods for
evaluating the "lightness" of color ... however, because for "white" and
"black", when R=G=B, I have to wonder how useful "weighted green" is for
evaluating "how many white pixels" & "how many black pixels"(???)

  (... by no means a criticism, and I thusfar like Ed's implimentation ...
but I believe how lightness is calculated is important to properly
evaluating Wp & Bp ...)

shAf  :o)

Re: filmscanners: VueScan Histogram

[Brad Smith]

Some lists generate a lot of sudden, surpise laughs.  This list
doesn't...till I saw this response to the 
"How is it calculated?" question.

Ed..great response!!!
Brad

[EMAIL PROTECTED] wrote:
> 
> In a message dated 12/17/2001 10:46:16 AM EST, [EMAIL PROTECTED]
> writes:
> 
> > How is the upper histogram calculated?
> 
> m_hst[3][(306L*pix[0]+601L*pix[1]+117L*pix[2])>>18]++;
> 
> Regards,
> Ed Hamrick

Re: filmscanners: VueScan Histogram (explained)

[Erik Kaffehr]

Hi!

You  probably want to choose White point in such a way that the only a small 
percentage falls outside the histogram. Setting WP to 0.0 mean that you get 
everything, but that's probably not what you. Specular highlights should 
probably not be included in the density range. So you would use histogram to 
see what you get rid off . Ed essentially said that he puts this feature in 
Vuescan and let the users tell him how they want it, so I presume we are 
going to see some significant evolution in this area.

Best regards,

Erik




söndagen den 16 december 2001 18.47 skrev du:
> Could someone explain to me in a few words how to work with a histogram
> when scanning? I can see it in VueScan but I don´t know what to do with it,
> because there are no sliders anyway...
>
> Thanks in advance,
> Bernhard

-- 
Erik Kaffehr[EMAIL PROTECTED] alt. [EMAIL PROTECTED]
Mariebergsvägen 53  +46 155 219338 (home)
S-611 66 Nyköping   +46 155 263515 (office)
Sweden  -- Message sent using 100% recycled electrons --

Re: filmscanners: VueScan Histogram was VueScan 7.3.2

[Bob Armstrong]

michael shafferwrote:
>   I wonder if a problem might be, that such a histogram ... to be more
> accurate than setting the endpoints at exactly (e.g.,) 0.03% ... would first
> necessitate an actual scan.  That is, a histogram based on a preview will
> not be as accurate as placing a decimal, as in the example I used.

I guess if the preview is done at a different resolution to the scan then their 
histograms are likely to differ.  However, it doesn't take long to rescan from memory 
if the file isn't saved until the user is satified with the B/W point settings.  Of 
course, this would mean (the option) of separating the scan from the file save 
function.  

And, how can you know where to set the B/W points if you don't use a histogram ;)

Bob 

RE: filmscanners: VueScan Histogram was VueScan 7.3.2

[michael shaffer]

Bob writes ...

> ...
> And, how can you know where to set the B/W points if
> you don't use a histogram ;)

  I have 2 problems with that question: (1)  I suppose I don't know exactly
what you are trying to do(?) ... (2) a histogram is a visual tool, but it
would be extremely difficult to present a truely useful histogram.  A case
on point is Photoshop's histogram ... for accurately using it, do you pay
attention to the visual or do you pay attention to the numbers?

  (1)  It has always been the philosohy for Vuescan ... "if there is a
better tool in the subsequent RGB editing software, leave Vs to the basics."
I believe Vs's present Wp and Bp tools can put ^all^ the RGB into the scan,
and make it stretch from one endpoint of the histogram to the other, as well
as exclude a certain percentage (which might be considered noise, or
specular highlights without detail).  Once this is accomplished, Photoshop
will do a better job of adjusting your endpoints for subsequent editing
artifacts or accommodating the characteristics of your printer.

  (2)  What would you have the histogram show?  The preview's pixels or
absolutely ^all^ of the pixels? Numeric feedback for 8bit RGB or 16bit? ...
each individual RGB channel or some calculated "luminosity" value?  Which
luminosity value? ... 'L' of Lab? ... 'B' of HSB?  Or, would we ultimately
ask Ed for ^all^ options?

  I don't have a problem with Ed adding a histogram, but I cannot imagine it
otherwise being only a visual presentation, and not practically that useful.
But let's all decide first on which histogram we want before we ask Ed to
change it again ... and again.

my $0.02 ... shAf  :o)

Re: filmscanners: VueScan Histogram was VueScan 7.3.2

[Bob Armstrong]

michael wrote:

> Bob writes ...

> > And, how can you know where to set the B/W points if
> > you don't use a histogram ;)
> 
>I have 2 problems with that question: 

>(1) I suppose I don't know exactly what you are trying to do?

Trying to find out how VS users get the most out of a scan when using VS.  (Which begs 
the question of what 'most' means.  You say VS can put *all* the RGB into a scan and 
make it stretch from one end of the histogram to the other, as well
as exclude a certain percentage (which might be considered noise, or specular 
highlights without detail).  I would agree that generally this would be a suitable 
definition for getting the most out of a scan.)   

>(2) a histogram is a visual tool, but it would be extremely difficult to 
>present a truely useful histogram.  A case on point is Photoshop's 
>histogram ... for accurately using it, do you pay attention to the 
>visual or do you pay attention to the numbers?

Both.
 
>(1)  It has always been the philosohy for Vuescan ... "if there is a 
>better tool in the subsequent RGB editing software, leave Vs to the 
>basics."

I've heard that said and don't disagree with it.   

>(2)  What would you have the histogram show?  The preview's pixels
>or absolutely ^all^ of the pixels? Numeric feedback for 8bit RGB or 
>16bit? ... each individual RGB channel or some calculated "luminosity" 
>value?  Which luminosity value? ... 'L' of Lab? ... 'B' of HSB?

A histogram of the scan at the bit level the scanner can produce.  Preferably 
luminosity, if not RGB would do if they were displayed together.

>I don't have a problem with Ed adding a histogram, but I cannot 
>imagine it otherwise being only a visual presentation, and not 
>practically that useful. But let's all decide first on which histogram we 
>want before we ask Ed to change it again ... and again.

VuePrint has a histogram with some numbers but not the numeric detail of Photoshop's 
histogram.  Why would a histogram in VS be 'not practically that useful'?  Last I 
heard indicated Ed didn't wants to add a histogram to VS so we may not get a chance to 
ask him to change it again and again...

Hope this helps you understand what I am trying to do.  Do you change B/W points in 
VS?  If so why and how do you decide what they should be?  

Bob
still very confused about scanning 

RE: filmscanners: VueScan Histogram was VueScan 7.3.2

[michael shaffer]

Bob writes ...

> ...
> Hope this helps you understand what I am trying to do.  Do you
> change B/W points in VS?  If so why and how do you decide what
> they should be?

  Up until the latest versions I always put a hard number (%) in the Wp & Bp
... e.g., '0.1' or 0.01%... and because changing this number by minute
amount changed the image more than you'd think, I had Vs write a small JPG
direct to Photoshop (note ... such a small change in Wp% would never be
gleened from visually inspecting a preview histogram).  When satisfied, I
had Vs write a TIF.  Regarding newer versions, I understand Ed has changed
what '0' means in the Bp field ... and I have to again evaluate what this
means to the histogram.

shAf  :o)

Re: filmscanners: VueScan Histogram was VueScan 7.3.2

[Hill James]

I'm all for having a histogram but I would be just as happy to see the
clipped pixels shown with a warning color.  I depend more on this feature in
Insight than I do the actual histogram.  (partly because Insight doesn't
allow gamma changes when saving in AdobeRGB)  At least with the clipped
pixels highlighted I can set WP & BP well enough to get a reasonable
histogram and finish in Photoshop.

James Hill
Freelance Photographer
Mebane, NC
[EMAIL PROTECTED]

- Original Message -
From: <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Saturday, December 15, 2001 7:54 AM
Subject: Re: filmscanners: VueScan Histogram was VueScan 7.3.2


> In a message dated 12/14/2001 12:46:00 PM EST, [EMAIL PROTECTED]
writes:
>
> > Last I heard indicated Ed didn't wants to add a histogram to VS so we
may
> not
> > get a chance to ask him to change it again and again...
>
> I think the best way to do this is for me to just add a histogram and
> then evolve it using people's suggestions.
>
> It's quite possible that a simple histogram of the preview (within
> the crop window) and the scan will be good enough.  I can worry
> about fancy stuff like movable sliders later.
>
> Regards,
> Ed Hamrick

Re: filmscanners: VueScan Histogram was VueScan 7.3.2

[sirius]

I agree with Hill James:
using the histogram is functional when you have feedback on what is going to
be clipped.
This could be achieved by
1. indicator color on the going to be clipped part of the image (seems
complex to me to realize)
or by
2. a measurement tool indicating rgb levels in 16 bit and in 8 bit numbers
on mouse position. This can be used for the numeric input of clipping point
of B and W. (should be easier to implement)

A histogram is a big aid, but should be paired with some visual feed back on
what pixels in teh image are BP and WP.

jan albrecht


- Original Message -
From: <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Saturday, December 15, 2001 1:54 PM
Subject: Re: filmscanners: VueScan Histogram was VueScan 7.3.2


> In a message dated 12/14/2001 12:46:00 PM EST, [EMAIL PROTECTED]
writes:
>
> > Last I heard indicated Ed didn't wants to add a histogram to VS so we
may
> not
> > get a chance to ask him to change it again and again...
>
> I think the best way to do this is for me to just add a histogram and
> then evolve it using people's suggestions.
>
> It's quite possible that a simple histogram of the preview (within
> the crop window) and the scan will be good enough.  I can worry
> about fancy stuff like movable sliders later.
>
> Regards,
> Ed Hamrick
>

Re: filmscanners: VueScan Histogram (should be log?)

[Erik Kaffehr]

Hi!

I don't know. My reasoning is that it would be useful to have a log scale, in 
case where there is only a small fraction of very bright pixels. I think
Photoshop uses linear scale and GIMP log scale, but I don't know. I of course 
have the source code for GIMP and could look if I really wanted.

Regards

Erik

söndagen den 16 december 2001 10.36 skrev du:
> In a message dated 12/16/2001 4:15:12 AM EST, [EMAIL PROTECTED] 
writes:
> > Just a question, is the y-axis in the histograms linear or log?
>
> It's linear.  Would people prefer it as a log scale?
>
> What scale do other programs use?
>
> Regards,
> Ed Hamrick

-- 
Erik Kaffehr[EMAIL PROTECTED] alt. [EMAIL PROTECTED]
Mariebergsvägen 53  +46 155 219338 (home)
S-611 66 Nyköping   +46 155 263515 (office)
Sweden  -- Message sent using 100% recycled electrons --