Digital BW, The Print

I'm back to scanning for a 2nd round of learning and could use some additional perspectives. I'm scanning old 35mm Plus-X negs. and would like them to be as sharp as possible. The question I have is how do I best evaluate - on screen - scans of the same frame made at two different resolutions? While the answer seems to be to view them at "actual pixels" size, when doing a side by side comparison the larger (full optical rez) scan is more "zoomed in" in appearance than the smaller scan and is softer with more grain showing in the sky.

My scene for this test is a winter landscape with fully overcast sky and leafless trees about 150+ feet in the distance. I'm looking at the smallest branches, trunk detail, and sky grain. My largest scan is at full optical resolution and the smaller is one step down.

Does it make sense to look at these side by side at actual pixel size or should I go further and make prints of the same size? While making prints is a necessary final step, I'd like to first make the best onscreen, side by side evaluation that I can. Thoughts?

Michael wrote:
> Does it make sense to look at these side by side at actual pixel size or should I go further and make prints of the same size? While making prints is a necessary final step, I'd like to first make the best onscreen, side by side evaluation that I can. Thoughts?
>   

Make prints.

It's difficult to impossible to make any meaningful evaluation of 
"sharpness" on a monitor. The monitor dot pitch is too dissimilar from 
printing. This causes what you see on the monitor to be even farther 
enlarged (usually 3-5x more) than detail you'd see on a print. Which of 
course makes it "soft" on the monitor.

Once you've made prints and determined how your scanner settings effect 
the final print, you might be able to apply what you learned from making 
and evaluating the prints to what you see on the monitor. But even then, 
only in general terms.

This also applies to sharpening algorithms. The only way to tell what 
the real effect is, is make prints.
--
Bruce Watson

In the words of Dave Etchells, the Publisher of the Imaging Resource:
 
"Testing hundreds of digital cameras, we've found that you can only tell just so much about a camera's image quality by viewing its images on-screen. Ultimately, there's no substitute for printing a lot of images and examining them closely."

Experience tells me that the same applies to films scan.

Cheers,
Andre

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, "Michael" <michael3442@...> wrote:
>
> I'm back to scanning for a 2nd round of learning and could use some additional perspectives. I'm scanning old 35mm Plus-X negs. and would like them to be as sharp as possible. The question I have is how do I best evaluate - on screen - scans of the same frame made at two different resolutions? While the answer seems to be to view them at "actual pixels" size, when doing a side by side comparison the larger (full optical rez) scan is more "zoomed in" in appearance than the smaller scan and is softer with more grain showing in the sky.
> 


Scan at max resolution and keep that as a master for future prints.  Lower resolution will be fine for small prints, but when you want to go larger, you don't want to have to scan yet again.

Harry

Keep a log of the workflow and make prints and include the mundane; printer,
version of software, etc. Change a setting, make a test print. However, the
biggest issue that I have seen over the years is that today's workflow may
be quite a bit different that next months or next year. As the screen to
print  matching gets better, the sharpening tools - everything from pre
output, to RIPs and output devices, your interpretation of the data may
change. I believe that this is what has lead to the "non destructive" buzz
word. Early attempts to edit may have been destroying your (our) data. So as
Harry points out, scan big and save it. You might even scan the same image
big with several different key setting changed; USM, - different channels,
CI of your data.  I have gone back to some early conversions of 4x5 B&W negs
that I had RAW scans made of and made much better files to print from the
second time around.  

 

Eric   

 

Eric Neilsen

Eric Neilsen Photography

4101 Commerce Street, Suite 9

Dallas, TX 75226

 

www.ericneilsenphotography.com

skype me with ejprinter

www.ericneilsenphotography.com/forum1

Let's Talk Photography

 

  _____  

.

This also applies to sharpening algorithms. The only way to tell what 
the real effect is, is make prints.
--
Bruce Watson





[Non-text portions of this message have been removed]

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, "hflockwood" <hflockwood@...> wrote:
>
> 
Harry,

You are not reading what Michael has said.   He said that a higher resolution scan created a file that showed more grain and was softer. We all know the mantra to scan at high res for possible future use, but possibly higher res scans magnify grain.  Has this issue been covered?

Lincoln

> 
> --- In DigitalBlackandWhiteThePrint@...m, "Michael" <michael3442@> wrote:
> >
> > I'm back to scanning for a 2nd round of learning and could use some additional perspectives. I'm scanning old 35mm Plus-X negs. and would like them to be as sharp as possible. The question I have is how do I best evaluate - on screen - scans of the same frame made at two different resolutions? While the answer seems to be to view them at "actual pixels" size, when doing a side by side comparison the larger (full optical rez) scan is more "zoomed in" in appearance than the smaller scan and is softer with more grain showing in the sky.
> > 
> 
> 
> Scan at max resolution and keep that as a master for future prints.  Lower resolution will be fine for small prints, but when you want to go larger, you don't want to have to scan yet again.
> 
> Harry
>

The interaction between grain, light source, lens, sensor... is
complicated.  Certain combinations can lead to surprising results.  Thus, it
is certainly possible that increasing the resolution of the scan can lead to
an increase in grain in the scan.


[Non-text portions of this message have been removed]

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, "outdoornm" <outdoornm@...> wrote:
>
> 
> 
> --- In DigitalBlackandWhiteThePrint@yahoogroups.com, "hflockwood" <hflockwood@> wrote:
> >
> > 
> Harry,
> 
> You are not reading what Michael has said.   He said that a higher resolution scan created a file that showed more grain and was softer. We all know the mantra to scan at high res for possible future use, but possibly higher res scans magnify grain.  Has this issue been covered?
> 
> Lincoln
> 

> > Scan at max resolution and keep that as a master for future prints.  Lower resolution will be fine for small prints, but when you want to go larger, you don't want to have to scan yet again.
> > 
> > Harry
> >

My workflow, from scanning (Nikon 4000ED) to printing (Cone K7 inks in an R2400), has been radically altered based on two articles I recently discovered.

The first, most relevant to this discussion, is by P.L. Andrews at www.photoscientia.co.uk/grain.htm. explains in detail how the interaction between the sensor's photo sites and film grain can lead to grain aliasing.  The effect, which I verified unequivocally in a troublesome scan, showed that apparent grain enhancement or scanner "noise" is neither but, instead, is grain aliasing.  Scanning at lower resolution is NOT the answer.  I urge anyone who is unfamiliar with this phenomenon to read Mr. Andrews's analysis.  I believe it may be relevant to the OP's observation.  [NikonScan's GEM feature minimizes the problem with little or no image degradation.]

The second article, by John Brownlow, supports Eric's comments, especially regarding one's sharpening workflow.  It can be found at www.luminous-landscape.com/tutorials/smart_sharp.shtml.  I'll never go back to the hit-or-miss technique that has served me (NOT) so well in the past.  I can now take my older high resolution scans and get much better prints with this more rational sharpening technique.

So, I'll stick to my original recommendation: scan at maximum resolution but understand the artifacts that can arise depending upon the interaction of film grain and your sensor's photo site geometry.

Harry

That link is not working.

Mark
http://www.stillrivereditions.com

On Jul 5, 2010, at 8:57 AM, hflockwood wrote:

> The first, most relevant to this discussion, is by P.L. Andrews at www.photoscientia.co.uk/grain.htm. explains in detail how the interaction between the sensor's photo sites and film grain can lead to grain aliasing.  



[Non-text portions of this message have been removed]

Just delete the /grain.htm and the article is there

Alan

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, Mark Savoia <mark@...> wrote:

>
> That link is not working.
> 
> Mark
> http://www.stillrivereditions.com
> 
> On Jul 5, 2010, at 8:57 AM, hflockwood wrote:
> 
> > The first, most relevant to this discussion, is by P.L. Andrews at www.photoscientia.co.uk/grain.htm. explains in detail how the interaction between the sensor's photo sites and film grain can lead to grain aliasing.  
> 
> 
> 
> [Non-text portions of this message have been removed]
>

Mark Savoia schreef:
> That link is not working.
> 
> Mark
> http://www.stillrivereditions.com

www.photoscientia.co.uk and then to the subject.

There are better sites in my opinion.

-- 
Met vriendelijke groeten,   Ernst

Try: http://tech.groups.yahoo.com/group/Wide_Inkjet_Printers/

|      Dinkla Grafische Techniek      |
|         www.pigment-print.com        |
|                 ( unvollendet )                 |

outdoornm schreef:
> 
> --- In DigitalBlackandWhiteThePrint@yahoogroups.com, "hflockwood"
> <hflockwood@...> wrote:
>> 
> Harry,
> 
> You are not reading what Michael has said.   He said that a higher
> resolution scan created a file that showed more grain and was softer.
> We all know the mantra to scan at high res for possible future use,
> but possibly higher res scans magnify grain.  Has this issue been
> covered?
> 
> Lincoln

Make it "aliased" grain.

Compare what a scanner with a true higher optical  resolution resolves
and one with a lower resolution. Both will make aliased grain but not
the same.

Reduce the aliased grain in both a lower resolution and higher
resolution scan from the same scanner and compare the scans on detail.
If there is no extra detail the scanner just doesn't resolve more than
what the lower resolution already did.

The Epson 4870 has resolution settings that go way above the actual
optical resolution it has. So higher resolutions will not create more
detail yet the grain particle will be made with more pixels than a lower
resolution scan that gives the same detail, the same result you would
get with upsampling. At pixel view the grain will be bigger, the image
softer.

The Epson hardware already oversamples by nature. Not a bad thing as it
can create a bit more dynamic range that way due to lower noise as a
result of oversampling. The scan can use a lot of sharpening so some
softness isn't an issue. Either use multi sampling in the driver or the
highest resolution, it will bring similar results.

True optical resolution of that scanner is  1500 -2000 PPI depending on
the scan method. But you will need a higher resolution (actually
sampling rate) set in the driver to get that true optical resolution or
use an odd number say 2351 PPI to force the driver to the highest
sampling rate and let the driver do the downsampling. As the last will
not have the best anti-aliasing on downsampling I would prefer to use
the highest sampling rate and the downsampling done by one of the better
downsampling programs.

http://www.xs4all.nl/~bvdwolf/main/foto/down_sample/example1.htm

Your image doesn't have to have repeating detail to make a good
anti-aliasing filter necessary, grain or more correct aliased grain in
the scan behaves the same in downsampling.

Or use a noise reducer like Neat Image and do the downsampling after
that to a resolution that holds the detail but removes the void pixels.

So in one case we think we see larger (aliased) grain but it is just
more pixels per grain particle, in the other case we get larger aliased
grain because aliased grain is even more aliased in downsampling.

Check Time Vitale's PDF on aliased grain:
http://cool.conservation-us.org/coolaic/sg/emg/library/pdf/vitale/2007-04-vitale-filmgrain_resolution.pdf

http://www.boeringa.demon.nl/menu_technic_ektar100_grainstructure.htm

What might be an interesting route to smaller aliased grain sizes is the 
use of ultraviolet light source in a B&W film scanner. I think the blue 
filtered sensor will still be sensitive enough down to 350Nm. The lens 
will cope as well and other glass in the optical path has to be tested. 
Whether an Epson 4870 will be the right Guinea pig is another matter.

-- 
Met vriendelijke groeten,   Ernst

Try: http://tech.groups.yahoo.com/group/Wide_Inkjet_Printers/

|      Dinkla Grafische Techniek      |
|         www.pigment-print.com        |
|                 ( unvollendet )                 |

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, Ernst Dinkla <edinkla@...> wrote:

SNIP

> 
> Make it "aliased" grain.
> 
> Compare what a scanner with a true higher optical  resolution resolves
> and one with a lower resolution. Both will make aliased grain but not
> the same.
> 
> Reduce the aliased grain in both a lower resolution and higher
> resolution scan from the same scanner and compare the scans on detail.
> If there is no extra detail the scanner just doesn't resolve more than
> what the lower resolution already did.
> 

ANOTHER BIG SNIP.


> Met vriendelijke groeten,   Ernst
> 
> Try: http://tech.groups.yahoo.com/group/Wide_Inkjet_Printers/
> 
> |      Dinkla Grafische Techniek      |
> |         www.pigment-print.com        |
> |                 ( unvollendet )                 |
>

Thanks for all that, Ernst; clearly a lot of work has been done on the subject.  

Harry

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, "Alan" <alanj.martin@...> wrote:
>
> Just delete the /grain.htm and the article is there
> 
> Alan
> 
> --- In DigitalBlackandWhiteThePrint@yahoogroups.com, Mark Savoia <mark@> wrote:
> >
> > That link is not working.
> > 
> > Mark
> > http://www.stillrivereditions.com
> > 
> > On Jul 5, 2010, at 8:57 AM, hflockwood wrote:
> > 
> > > The first, most relevant to this discussion, is by P.L. Andrews at www.photoscientia.co.uk/grain.htm. explains in detail how the interaction between the sensor's photo sites and film grain can lead to grain aliasing.  
> > 
> > 
> > 
> > [Non-text portions of this message have been removed]
> >
>
The link actually does work - but without the period.  Sorry about that.

Harry

On 05/07/2010 outdoornm wrote:
> You are not reading what Michael has said. He said that a higher 
> resolution scan created a file that showed more grain and was softer. 
> We all know the mantra to scan at high res for possible future use, 
> but possibly higher res scans magnify grain. Has this issue been 
> covered?

No, higher res does not "magnify grain", but you can get grain aliasing 
artifacts when scanning at any resolution below the Nyquist limit for the 
emulsion.

Grain aliasing is similar to moire as a random interference between 
spatial frequency of sampling and grain distribution, and produces false 
grain as an artifact.

Because grain size and clumping is highly variable, depending on emulsion, 
exposure and development, the Nyquist limit for the grain structures in 
any emulsion is is a range of frequencies. Aliasing effects can therefore 
be localised and limited to certain areas of the image, or across the 
frame, or absent.

I first described this c.1999 on the filmscanners list when Polaroid were 
puzzled by exaggerated grain in customer scans. You'll find more thorough 
subsequent discussion about the effect by Pete Andrews at 
http://www.photoscientia.co.uk/Grain.htm

Some 2700ppi scanners like the earlier Nikon Coolscans, with their 
semi-collimated LED lightsource, could produce horrendous false grain in 
areas such as blue sky on ISO100 slide materials. Many users thought they 
were seeing real ISO100 grain, and that meant that 2700ppi was more than 
adequate. In fact they were seeing aliasing products that looked rather 
like grain, from material that appears essentially grainless to the eye.

Higher res scans mitigate the effect.  IME upwards of 4000ppi largely 
avoids the effect on most materials. However I've seen it with overexposed 
Fuji200 colour neg and also with some of the ISO3200 B&W neg films. 
TMax3200 is essentially unscannable for me, using a Polaroid SS4000, the 
aliasing turns out an image made of gravel.

-- 
Regards

Tony Sleep
http://tonysleep.co.uk

Thank you Tony, Lincoln and others. What is getting lost is my original question: "The question I have is how do I best evaluate - on screen -
scans of the same frame made at two different resolutions?" Before you say to view at actual pixels, I know that part of the answer. But I think it's more complicated than that. Viewing a 6 foot length of tree branch at actual pixels from 2 differently sized scans makes that same length of branch different lengths, in inches, on the screen. This makes it hard to look at the same section and make a direct comparison. It the lower rez scan is magnified beyond actual pixels then interpolation(?) from software becomes an issue. But without a same sized, side by side comparison of the two scans, how do I best evaluate them?

>
> Thank you Tony, Lincoln and others. What is getting lost is my original
> question: "The question I have is how do I best evaluate - on screen -
> scans of the same frame made at two different resolutions?"
>  _
>

Well, you have a couple of choices, none of which is perfect.  First, you
could up-sample the smaller photo.  The problem is that this will make that
up-sampled file look worse than it it.  Second, you could down sample the
bigger file, which will make it look better than it is.  Third, you could
view the files at different magnifications, such that the image elements are
the same size on your screen.  All of these methods interpolate the data.
As such, how this is done will effect the results.  If it were me, I'd
choose the second option along with looking at prints.  After all, the main
question is whether the higher res scan gives any benefit.  If after
down-sampling, which should increase the quality of the higher-res file,
there still isn't any difference with the lower res scan, then there's no
point in doing the higher res scan.


[Non-text portions of this message have been removed]

On 06/07/2010 Michael wrote:
> "The question I have is how do I best evaluate - on screen -
> scans of the same frame made at two different resolutions?" 

I think you are into territory where the answer is simply "experience" 
through doing it. There are no objective methods because the concept of 
quality is fuzzy, and as you say, interpolation will confuse matters. Like 
comparing 6x6 negs with 35mm, it's a matter of "good enough" by whatever 
your personal standards are. I found comparing both images at 100% was 
quite good enough for comparing scanners of different resolutions, usually 
investigating differences and artifacts at higher and lower mag.
-- 
Regards

Tony Sleep
http://tonysleep.co.uk

Michael,
I'm sliding in a tad late but it seems, to me anyway, that it's getting hard to see the forest thru the trees in this post series... lots of good thoughts.. 
Ok first what is sharpness (sorry) ..For me it is about contrast between various tonalities in a minimum with of pixels such that the perception of sharp is achieved..  So a transition over 5 pixels will be perceived very differently then a transition over 15 pixels.. When we look at image on the screen we actually don't get to see it on size we see a version that is show at say 72 dpi.. then interpolated by the video card to show us it's interpretation of what that image might look like as we make it smaller on our screen..  So in your effort to compare the sharpness of two scans done at different resolutions I think looking at the entire image on your screen is causing one to get lost..  I would check a few areas of each image at 100% ..in other words at a pixel level and literally note or count the pixels for a given area of both do the transition... I realize this may be subjective at times but it works..  ..In the end one might say it's meaning less but that will help you understand the variance.. Lets say one is a 6 pixel transition and the other is an 11 pixel transition.. To me this in information but not really that valuable in the end unless you interpolated the larger image down and got a bump of a smaller pixel transition then the smaller original scan. Typically larger prints reflect a greater viewing distance..so it's really not appropriate to say, in my opinion, both items will be viewed from the same distance ..but I guess they could be.. Lastly, the proof in the pudding is not the screen it's the print.. The reason I say that is that again your video card scales images on your screen ..but we've all seen jaggies on screen and none in a print. The video card is a rendering of the image at various sizes.. Only when viewing it at 100% can you see sharpening artifacts properly.. As technicians we must learn thru experience to be able to understand what we see.. B&W is really incredibly different then color imagery.. Color can use itself to show different object transitions by just using a different hue to do it.. B&W has only shades of gray to do the same job so we must learn to use that to it's maximum to show all that is in an image. That's going away from your inquiry I guess but to me it a part of it as when we push to make these transitions happen we also can get into messing with perceived sharpness at times. Anyway spending the time on screen to evaluate two tree branches at different resolutions in their entirety may not be the best use of time. Simply check a few areas one to the other for transitional length followed by maybe small proofs showing those areas..

jimbo

----- Original Message ----- 
  From: Michael 
  To: DigitalBlackandWhiteThePrint@yahoogroups.com 
  Sent: Tuesday, July 06, 2010 8:31 AM
  Subject: [Digital BW] Re: Scanning b&w negs, revisited


    
  Thank you Tony, Lincoln and others. What is getting lost is my original question: "The question I have is how do I best evaluate - on screen -
  scans of the same frame made at two different resolutions?" Before you say to view at actual pixels, I know that part of the answer. But I think it's more complicated than that. Viewing a 6 foot length of tree branch at actual pixels from 2 differently sized scans makes that same length of branch different lengths, in inches, on the screen. This makes it hard to look at the same section and make a direct comparison. It the lower rez scan is magnified beyond actual pixels then interpolation(?) from software becomes an issue. But without a same sized, side by side comparison of the two scans, how do I best evaluate them? 



  

[Non-text portions of this message have been removed]

Michael schreef:
> Thank you Tony, Lincoln and others. What is getting lost is my
> original question: "The question I have is how do I best evaluate -
> on screen - scans of the same frame made at two different
> resolutions?" Before you say to view at actual pixels, I know that
> part of the answer. But I think it's more complicated than that.
> Viewing a 6 foot length of tree branch at actual pixels from 2
> differently sized scans makes that same length of branch different
> lengths, in inches, on the screen. This makes it hard to look at the
> same section and make a direct comparison. It the lower rez scan is
> magnified beyond actual pixels then interpolation(?) from software
> becomes an issue. But without a same sized, side by side comparison
> of the two scans, how do I best evaluate them?
> 
> 

Set the largest file at pixel view. Scale the smallest one so it matches
the detail size. Roll back your chair backward and forward. Set "Zoom
all windows" and zoom both windows in and out simultaneously. If there
is no difference to be seen then there is no difference.

-- 
Met vriendelijke groeten,   Ernst

Try: http://tech.groups.yahoo.com/group/Wide_Inkjet_Printers/

|      Dinkla Grafische Techniek      |
|         www.pigment-print.com        |
|                 ( unvollendet )                 |

Hi,
I'm considering buying a V700 scanner for MF negs and wondered if anyone had tried wet mounting with these scanners as I've been told it's possible - and how the results were.

thanks,
Richard

____________
http://www.richardsmallfield.com

Op 18-10-2010 3:30, Richard Smallfield schreef:
> Hi,
> I'm considering buying a V700 scanner for MF negs and wondered if anyone had tried wet mounting with these scanners as I've been told it's possible - and how the results were.
>
> thanks,
> Richard
>
> ____________
> http://www.richardsmallfield.com


http://www.betterscanning.com/scanning/fluidresources.html

Doug Fisher makes the better wet mount kit for Epson scanners including 
the V700.


-- 
Met vriendelijke groeten,   Ernst

Try: http://tech.groups.yahoo.com/group/Wide_Inkjet_Printers/

|      Dinkla Grafische Techniek      |
|         www.pigment-print.com        |
|                 ( unvollendet )                 |