Digital BW, The Print

It boils down  to what, if any advantage there is to working with 8-bit v. 16-bit images, and different genuine experts will give different answers - Dan Margulis says 16-bit is unnecessary and has issued a challenge to prove otherwise, while Bruce Fraser maintains that it is generally better to start with 16-bit files because (in color images anyway) of the additional color information available.

This is a debate which has been carried on to a great extent on Dan Margulis's Color Theory list, and there is an archive of the thread at
http://www.ledet.com/margulis/ACT_postings/ACT-8-bit-16-bit.html

Maris

----- Original Message ----- 
From: "Mark Tucker" <mark@marktucker.com>
To: <DigitalBlackandWhiteThePrint@yahoogroups.com>
Sent: Tuesday, December 04, 2001 2:35 PM
Subject: [Digital BW] 16-bit Scanning: Why?


| Can anybody tell me in twenty words or less what the advantage 
| is of scanning in 16-bit?
| 
| I heard one guy say that he can let his assistant scan images in 
| 16-bit, and have them just set the black and white point, and do 
| nothing else. Then he comes in after them and really tweaks the 
| file.
| 
| This scenario above makes sense I guess, but it sure doesn't 
| apply to me. Past that, if you neg is properly exposed, and your 
| scanner is good to begin with (Imacon), why would I ever want to 
| do this?
| 
| When I've tested it in the past, and resampled from 16 to 8-bit, I 
| see absolutely no change in the file. Am I missing something?
| 
| -Simpleton
| 
| 
| 
| Please visit the Group Homepage to check the Files, Bookmarks, Polls and other resources as they are often being updated. The page is at:
| 
| http://groups.yahoo.com/group/DigitalBlackandWhiteThePrint
| 
| Please follow these basic guidelines:
| - Include your full name with your message.
| - Include the address of your website, if you have one.
| - As threads develop, trim off excess portions of earlier messages to keep them short.
| - As the topic of a thread changes remember to change the subject header.
| - Good manners are required at all time. No personal attacks or "flames."
| - Complete your Yahoo profile.
| - Before posting a question, search the message archives and the various resources on the homepage. 
| 
| 
|  
| 
| Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/ 
| 
| 
|

> It boils down  to what, if any advantage there is to working with
> 8-bit v. 16-bit images, and different genuine experts will give
> different answers - Dan Margulis says 16-bit is unnecessary and
> has issued a challenge to prove otherwise, while Bruce Fraser
> maintains that it is generally better to start with 16-bit files
> because (in color images anyway) of the additional color
> information available.

Hi Maris,

I would like to see Margulis's challenge, because for tonal adjustments you
MUST MUST MUST do them with high-bit data, or you lose tonal values, period.
That isn't arguable at all.  And it DOES matter when doing B&W, especially
with Piezo, less so for color, but it still matters.

What I have found, is sometimes the purported expert isn't saying what you
think he's saying...problem with this "field" is not only the misuse of
terminology, but the reference one is talking from.

Regards,

Austin

Are you on Dan's Color Theory List?  If not, I will try to find the challenge (it was according to Dan mis-paraphrased at some point so I want to find it in accurate form) and will send it off-list.

As to the argument that you lose tonal values, his response is that theoretically you do but in the real world the result is identical.  I am not an expert so I will not take sides one way or the other.

Maris

----- Original Message ----- 
From: "Austin Franklin" <darkroom@...>
To: <DigitalBlackandWhiteThePrint@yahoogroups.com>
Sent: Tuesday, December 04, 2001 9:15 PM
Subject: RE: [Digital BW] 16-bit Scanning: Why?


| > It boils down  to what, if any advantage there is to working with
| > 8-bit v. 16-bit images, and different genuine experts will give
| > different answers - Dan Margulis says 16-bit is unnecessary and
| > has issued a challenge to prove otherwise, while Bruce Fraser
| > maintains that it is generally better to start with 16-bit files
| > because (in color images anyway) of the additional color
| > information available.
| 
| Hi Maris,
| 
| I would like to see Margulis's challenge, because for tonal adjustments you
| MUST MUST MUST do them with high-bit data, or you lose tonal values, period.
| That isn't arguable at all.  And it DOES matter when doing B&W, especially
| with Piezo, less so for color, but it still matters.
| 
| What I have found, is sometimes the purported expert isn't saying what you
| think he's saying...problem with this "field" is not only the misuse of
| terminology, but the reference one is talking from.
| 
| Regards,
| 
| Austin
| 
| 
| 
| Please visit the Group Homepage to check the Files, Bookmarks, Polls and other resources as they are often being updated. The page is at:
| 
| http://groups.yahoo.com/group/DigitalBlackandWhiteThePrint
| 
| Please follow these basic guidelines:
| - Include your full name with your message.
| - Include the address of your website, if you have one.
| - As threads develop, trim off excess portions of earlier messages to keep them short.
| - As the topic of a thread changes remember to change the subject header.
| - Good manners are required at all time. No personal attacks or "flames."
| - Complete your Yahoo profile.
| - Before posting a question, search the message archives and the various resources on the homepage. 
| 
| 
|  
| 
| Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/ 
| 
|

Maris,

> Are you on Dan's Color Theory List?  If not, I will try to find
> the challenge (it was according to Dan mis-paraphrased at some
> point so I want to find it in accurate form) and will send it off-list.

I am not on the list...please do send it to me.

> As to the argument that you lose tonal values, his response is
> that theoretically you do but in the real world the result is
> identical.  I am not an expert so I will not take sides one way
> or the other.

The fact that you lose tonal values isn't theoretical, it's just a plain old
fact...kind of like 1 apple + 1 apple = 2 apples.  It's a simple experiment
you can try your self ;-)

Regards,

Austin

> 
> The fact that you lose tonal values isn't theoretical, it's just a 
plain old
> fact...kind of like 1 apple + 1 apple = 2 apples.  It's a simple 
experiment
> you can try your self ;-)
> 
> Regards,
> 
> Austin

Not taking sides, I just think he was trying to say, that yes you do 
lose tonal value (numerically if you do the math), but with the 
quality of printers available, it won't show. Again, I don't have 
enough experience to say, one way or the other.

Austin,

The crux of the debate is do the tonal issues that you lose appear visibly in the finished product - the print or the web?  Using your analogy, can you tell the difference between a tray holding 5000 apples and a shallower tray holding 2500 apples just by looking at the tray?  Will the colors and tones appear different to the viewer, or can you only determine the difference scientifically - by counting them?

Anyway, the 8-bit v. 16-bit newsgroup thread I had referred to earlier at
http://www.ledet.com/margulis/ACT_postings/ACT-8-bit-16-bit.html
upon a re-reading only minimally touched on this issue and was a 1999 thread.

The actual challenge was apparently set forth in one of Dan's Newsletters which I don't have (probably issued before I joined the List and his Newsletter mailing list) but I do have it restated by Dan in a newsgroup message which I am forwarding directly to you.

There was also a thread or two in 2001 containing a very interesting debate on the issue which I have saved to .txt and am also forwarding directly.

If anyone else on the list wants either the challenge or both, please let me know offlist at
mlidaka@...

Maris

----- Original Message ----- 
From: "Austin Franklin" <darkroom@...>
To: <DigitalBlackandWhiteThePrint@yahoogroups.com>
Sent: Tuesday, December 04, 2001 9:42 PM
Subject: RE: [Digital BW] 16-bit Scanning: Why?


| Maris,
| 
| > Are you on Dan's Color Theory List?  If not, I will try to find
| > the challenge (it was according to Dan mis-paraphrased at some
| > point so I want to find it in accurate form) and will send it off-list.
| 
| I am not on the list...please do send it to me.
| 
| > As to the argument that you lose tonal values, his response is
| > that theoretically you do but in the real world the result is
| > identical.  I am not an expert so I will not take sides one way
| > or the other.
| 
| The fact that you lose tonal values isn't theoretical, it's just a plain old
| fact...kind of like 1 apple + 1 apple = 2 apples.  It's a simple experiment
| you can try your self ;-)
| 
| Regards,
| 
| Austin
| 
| 
| Please visit the Group Homepage to check the Files, Bookmarks, Polls and other resources as they are often being updated. The page is at:
| 
| http://groups.yahoo.com/group/DigitalBlackandWhiteThePrint
| 
| Please follow these basic guidelines:
| - Include your full name with your message.
| - Include the address of your website, if you have one.
| - As threads develop, trim off excess portions of earlier messages to keep them short.
| - As the topic of a thread changes remember to change the subject header.
| - Good manners are required at all time. No personal attacks or "flames."
| - Complete your Yahoo profile.
| - Before posting a question, search the message archives and the various resources on the homepage. 
| 
| 
|  
| 
| Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/ 
| 
| 
|  
|

>> As to the argument that you lose tonal values, his response is
>> that theoretically you do but in the real world the result is
>> identical.  I am not an expert so I will not take sides one way
>> or the other.
> 
> The fact that you lose tonal values isn't theoretical, it's just a plain old
> fact...kind of like 1 apple + 1 apple = 2 apples.  It's a simple experiment
> you can try your self ;-)

Dan's was speaking about photographs, which contain a lot of noise from
grain and the like, as opposed to "perfect" computer generated gradients.
(Remember, adding noise is the conventional cure for banding). His point
isn't that tones don't get dropped, it's that in all his years experience he
hasn't seen an instance where that shows up as detrimental in print, and if
someone would show him an instance where it did he'd eat his words and
publish it in one of his books or magazine articles.

He knows histograms, and he knows the math, he knows tones get dropped. He
feels between output screening methods, the limited ability of humans to
differentiate anywhere near 256 tones, the amount of colors that can be
combined from as few as say 50 tones in each channel, etc, add up to
deteriorated 8-bit images that might look *slightly* different than their
16-bit counterparts, but not necessarily worse, and sometimes better.

No one has stepped up to the plate as yet.

Todd

--- In DigitalBlackandWhiteThePrint@y..., "Maris V. Lidaka, Sr." 
<mlidaka@a...> wrote:
Using your analogy, can you tell the difference between a tray 
holding 5000 apples and a shallower tray holding 2500 apples 
just by looking at the tray?


NOW we're getting it down to MY LEVEL! Yes, Maris -- you tell that 
damn Austin! You GO, girl! I love that analogy; I feel the same 
way -- I convert from 16 to 8bit, and I'm still seeing the same tray 
of rotten apples on the topmost level. But they're MY apples, so I 
love them anyway.

Here, here, for keeping this in simple English...

-Dumber

Uh - I'm not a girl.

Maris

----- Original Message ----- 
From: "mtucker508" <mark@...>
To: <DigitalBlackandWhiteThePrint@yahoogroups.com>
Sent: Tuesday, December 04, 2001 11:16 PM
Subject: Re: [Digital BW] 16-bit Scanning: Why?


| --- In DigitalBlackandWhiteThePrint@y..., "Maris V. Lidaka, Sr." 
| <mlidaka@a...> wrote:
| Using your analogy, can you tell the difference between a tray 
| holding 5000 apples and a shallower tray holding 2500 apples 
| just by looking at the tray?
| 
| 
| NOW we're getting it down to MY LEVEL! Yes, Maris -- you tell that 
| damn Austin! You GO, girl! I love that analogy; I feel the same 
| way -- I convert from 16 to 8bit, and I'm still seeing the same tray 
| of rotten apples on the topmost level. But they're MY apples, so I 
| love them anyway.
| 
| Here, here, for keeping this in simple English...
| 
| -Dumber
| 
| 
| 
| Please visit the Group Homepage to check the Files, Bookmarks, Polls and other resources as they are often being updated. The page is at:
| 
| http://groups.yahoo.com/group/DigitalBlackandWhiteThePrint
| 
| Please follow these basic guidelines:
| - Include your full name with your message.
| - Include the address of your website, if you have one.
| - As threads develop, trim off excess portions of earlier messages to keep them short.
| - As the topic of a thread changes remember to change the subject header.
| - Good manners are required at all time. No personal attacks or "flames."
| - Complete your Yahoo profile.
| - Before posting a question, search the message archives and the various resources on the homepage. 
| 
| 
|  
| 
| Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/ 
| 
| 
|  
|

Todd,

> >> As to the argument that you lose tonal values, his response is
> >> that theoretically you do but in the real world the result is
> >> identical.  I am not an expert so I will not take sides one way
> >> or the other.
> >
> > The fact that you lose tonal values isn't theoretical, it's
> just a plain old
> > fact...kind of like 1 apple + 1 apple = 2 apples.  It's a
> simple experiment
> > you can try your self ;-)
>
> Dan's was speaking about photographs, which contain a lot of noise from
> grain and the like, as opposed to "perfect" computer generated gradients.
> (Remember, adding noise is the conventional cure for banding). His point
> isn't that tones don't get dropped, it's that in all his years
> experience he
> hasn't seen an instance where that shows up as detrimental in
> print, and if
> someone would show him an instance where it did he'd eat his words and
> publish it in one of his books or magazine articles.

But I DO see it all the time...most any time I try to do moves to 8 bit
files, and then print them using Piezo, I end up with posterization.  If I
need to do moves, I re-scan.

> He knows histograms, and he knows the math, he knows tones get dropped. He
> feels between output screening methods, the limited ability of humans to
> differentiate anywhere near 256 tones, the amount of colors that can be
> combined from as few as say 50 tones in each channel, etc, add up to
> deteriorated 8-bit images that might look *slightly* different than their
> 16-bit counterparts, but not necessarily worse, and sometimes better.
>
> No one has stepped up to the plate as yet.

Well, I know that I see exactly what he claims doesn't with my B&W images
printed with Piezo...

Regards,

Austin

Maris,

> Austin,
>
> The crux of the debate is do the tonal issues that you lose
> appear visibly in the finished product - the print or the web?

Yes, absolutely.

> Using your analogy, can you tell the difference between a tray
> holding 5000 apples and a shallower tray holding 2500 apples just
> by looking at the tray?

I don't see how that follows my analogy...nor do I understand the analogy.
Perhaps you could explain.

> Will the colors and tones appear
> different to the viewer, or can you only determine the difference
> scientifically - by counting them?

I don't know, but I'm talking B&W only.  I have no idea whether it matters
with color or not, since 8 bit color is really 24 bits, not 8 bits...and it
makes sense that it would not matter as much with color as it does with B&W
because color has 3x as many bits to play with.  Even if you end up moving 8
bits to 5 bits, you still get 15 bits total, or 32k colors, so I doubt it
matters much.

But...as I said, this IS a B&W list, and we are strictly talking B&W images.
With B&W, it is FAR more significant (16 bit vs 8 bit tonal moves) by at
least two orders of magnitude!

Regards,

Austin

>
>He knows histograms, and he knows the math, he knows tones get dropped. He
>feels between output screening methods, the limited ability of humans to
>differentiate anywhere near 256 tones, the amount of colors that can be
>combined from as few as say 50 tones in each channel, etc, add up to
>deteriorated 8-bit images that might look *slightly* different than their
>16-bit counterparts, but not necessarily worse, and sometimes better.

I can't help wondering if there's a corollary here to digital audio, 
if I can drift off-topic. The standard CD format, unchanged since 
1980 or so, is 16 bit, 44 Khz sampling rate. One can easily 
demonstrate mathematically (Nyquist, etc.) that this standard 
provides complete reproduction of the audible spectrum and dynamic 
range. It was actually introduced under the banner of "Perfect Sound 
Forever".

Problem was (and is) that the 16/44 CD never sounded as "good" as 
analog tape (or the venerable LP) in a high resolution audio system. 
The decimation, data jitter, and aliasing problems become easily 
audible. Audiophiles know this, but the average consumer is happy 
with their boombox and automobile playback. The insanely popular MP3 
format uses heavy compression of CD quality digital and throws away 
scads of data, but most listeners never notice.

There are now two "high-bit" audio formats being introduced that 
exploit the greater capacity of the DVD format. DVD-Audio is 24 bit, 
96Khz, and SACD is a 1 bit, ultra high sampling system that 
esentially mimics analog. The improvement with either is instantly 
heard as a more "natural" presentation, not as more volume, detail, 
or violins. Few consumers have ever seen one of these discs, let 
alone heard one. This despite the fact that most recording today is 
actually mastered at 20-24 bit resolution (often from analog session 
tapes). The high resolution recording is instead dumped to 16 bit for 
release. Because the market for a "better than good enough" product 
is minescule, the record companies have little incentive to release 
titles in the high-bit format, despite the fact the superior original 
sits in their vaults.

I suspect Adobe's perspective on bit depth is similar. The standard 
was based on what was possible 5 or 10 years ago. As a pre-press 
application, 8 bit is fine for the vast majority of users. The fact 
that today's scanners can provide 12, 14, or 16 bit data seems 
largely wasted. Yes, it permits more extensive manipulation without 
artifacts, but 16 bit may make no other difference in offset printing 
or web publishing.

I come to digital from a photographic background. The 8x10 AZO in 
Amidol contact print is a standard I'm familiar with. (So's the 12x20 
contact, but that's a bit much:-). Kodachrome 25, Ektar 25, platinum 
prints, Leica, Nikon, Schneider and Goerz glass, Tech Pan and Gigabit 
film. Ciba developed in Beers. I've even made a dye transfer print 
(once was enough!). I have high photographic standards. I like what I 
see from digital ink-jet printing, but my references have not been 
eclipsed. I'll be pleased when they are, because I don't want to rely 
on 19th century technology indefinitely.The few people (like myself) 
concerned with working directly with photographic images may not see 
the true benefit of 16 bit scans until "continuous tone" 16 bit 
printer technologies emerge. Or a hot-rod version of PS that 
streamlines the workflow and options for photographic output rather 
than print or web. And because I'm sure such products will eventually 
materialize, I will scan at the highest qualities I can, and archive 
the raw scans along with the negatives. For the same reason I kept my 
LPs.




--

Hi Bruce,

> I can't help wondering if there's a corollary here to digital audio,
> if I can drift off-topic.

I've delved into that quite a bit...having designed a lot of digital audio
gear...

> Problem was (and is) that the 16/44 CD never sounded as "good" as
> analog tape (or the venerable LP) in a high resolution audio system.
> The decimation, data jitter, and aliasing problems become easily
> audible.

Initially and up until about 7 years ago, I completely agree...the
oversampling filters were really pretty bad, and they did sound bad.  I
believe you mean interpolation, since decimation is decreasing data, and
interpolation is used to increase the data to smooth out the data to be more
analog (less distance between voltage steps), so a lower order filter can be
used which decreases aliasing.

With proper design, all the problems you mention can all be severely reduced
so they are not audible at all.  I have a very high end CD player (ML39),
that used a very very well designed analog front end, as well as every
conceivable method to keep analog noise down...and it easily sounds better
than any album, and certainly far better than any tape.  You're welcome to
come listen any time ;-)  This is due to painstaking engineering, and
considerable expense.

You make a good point, but I do believe that static visualization
(photography) is very different than dynamic audiblization (audio....ha, I
made a new word, I think ;-).

Regards,

Austin

Austin,

I think your off-list message was right on the head - Dan's position is one taken in dealing with color photographs.  When working in the B&W world tonality becomes much, much more important - in fact it is primary.  In the color world, color differences and shades would easily overcome or mask any banding problems excepting in broad expanses of the same color.

Maris

----- Original Message ----- 
From: "Austin Franklin" <darkroom@...tcom.com>
To: <DigitalBlackandWhiteThePrint@yahoogroups.com>
Sent: Wednesday, December 05, 2001 8:30 AM
Subject: RE: [Digital BW] 16-bit Scanning: Why?


| Todd,
| 
| > >> As to the argument that you lose tonal values, his response is
| > >> that theoretically you do but in the real world the result is
| > >> identical.  I am not an expert so I will not take sides one way
| > >> or the other.
| > >
| > > The fact that you lose tonal values isn't theoretical, it's
| > just a plain old
| > > fact...kind of like 1 apple + 1 apple = 2 apples.  It's a
| > simple experiment
| > > you can try your self ;-)
| >
| > Dan's was speaking about photographs, which contain a lot of noise from
| > grain and the like, as opposed to "perfect" computer generated gradients.
| > (Remember, adding noise is the conventional cure for banding). His point
| > isn't that tones don't get dropped, it's that in all his years
| > experience he
| > hasn't seen an instance where that shows up as detrimental in
| > print, and if
| > someone would show him an instance where it did he'd eat his words and
| > publish it in one of his books or magazine articles.
| 
| But I DO see it all the time...most any time I try to do moves to 8 bit
| files, and then print them using Piezo, I end up with posterization.  If I
| need to do moves, I re-scan.
| 
| > He knows histograms, and he knows the math, he knows tones get dropped. He
| > feels between output screening methods, the limited ability of humans to
| > differentiate anywhere near 256 tones, the amount of colors that can be
| > combined from as few as say 50 tones in each channel, etc, add up to
| > deteriorated 8-bit images that might look *slightly* different than their
| > 16-bit counterparts, but not necessarily worse, and sometimes better.
| >
| > No one has stepped up to the plate as yet.
| 
| Well, I know that I see exactly what he claims doesn't with my B&W images
| printed with Piezo...
| 
| Regards,
| 
| Austin
| 
| 
| Please visit the Group Homepage to check the Files, Bookmarks, Polls and other resources as they are often being updated. The page is at:
| 
| http://groups.yahoo.com/group/DigitalBlackandWhiteThePrint
| 
| Please follow these basic guidelines:
| - Include your full name with your message.
| - Include the address of your website, if you have one.
| - As threads develop, trim off excess portions of earlier messages to keep them short.
| - As the topic of a thread changes remember to change the subject header.
| - Good manners are required at all time. No personal attacks or "flames."
| - Complete your Yahoo profile.
| - Before posting a question, search the message archives and the various resources on the homepage. 
| 
| 
|  
| 
| Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/ 
| 
| 
|

Yes - that is precisely the cause of the 'combed' histogram.

Maris

----- Original Message ----- 
From: "mtucker508" <mark@...>
To: <DigitalBlackandWhiteThePrint@yahoogroups.com>
Sent: Wednesday, December 05, 2001 9:25 AM
Subject: Re: [Digital BW] 16-bit Scanning: Why?


| --- In DigitalBlackandWhiteThePrint@y..., "Austin Franklin" 
| <darkroom@i...> wrote:
| > What happens when you make tonal adjustments with 8 bits is 
| you loose tonal
| > values. 
| 
| 
| Would these tones "moving around" and getting shifted be 
| responsible for a histogram that had holes in it? (Not that I've 
| ever had one, but a friend of a friend told me that *they* had 
| some missing areas in *their* histograms...
| 
| If this is the case, then you certainly have my full-on attention.
| 
| Mark
| 
| 
| 
| Please visit the Group Homepage to check the Files, Bookmarks, Polls and other resources as they are often being updated. The page is at:
| 
| http://groups.yahoo.com/group/DigitalBlackandWhiteThePrint
| 
| Please follow these basic guidelines:
| - Include your full name with your message.
| - Include the address of your website, if you have one.
| - As threads develop, trim off excess portions of earlier messages to keep them short.
| - As the topic of a thread changes remember to change the subject header.
| - Good manners are required at all time. No personal attacks or "flames."
| - Complete your Yahoo profile.
| - Before posting a question, search the message archives and the various resources on the homepage. 
| 
| 
|  
| 
| Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/ 
| 
| 
|

First, some questions. By 'combed' histogram do you mean only a 
histogram that contains gaps all the way from bottom to top or do you 
include those those histograms which are slightly 'combed', meaning 
those that have a base throughout most of the histogram but have very 
narrow (comblike) peaks and valleys? Also, very roughly speaking, how 
bad does this combing effect have to be before it manifests itself in 
output? Second, an observation. I took a greyscale tiff file scanned 
from my SS4000, opened the file as 16 bit in PS, duplicated the image 
and then loaded the same two curves into each image. I got the same 
histogram changes for both images. Won't hazard to guess the 
significance of this. Also, I've noticed that sometimes if hit the 
Auto Levels button I get regularly spaced bottom-to-top gaps in the 
histogram. 

--- In DigitalBlackandWhiteThePrint@y..., "Maris V. Lidaka, Sr." 
<mlidaka@a...> wrote:
> Yes - that is precisely the cause of the 'combed' histogram.
> 
> Maris
> 
> ----- Original Message ----- 
> From: "mtucker508" <mark@m...>
> To: <DigitalBlackandWhiteThePrint@y...>
> Sent: Wednesday, December 05, 2001 9:25 AM
> Subject: Re: [Digital BW] 16-bit Scanning: Why?
> 
> 
> | --- In DigitalBlackandWhiteThePrint@y..., "Austin Franklin" 
> | <darkroom@i...> wrote:
> | > What happens when you make tonal adjustments with 8 bits is 
> | you loose tonal
> | > values. 
> | 
> | 
> | Would these tones "moving around" and getting shifted be 
> | responsible for a histogram that had holes in it? (Not that I've 
> | ever had one, but a friend of a friend told me that *they* had 
> | some missing areas in *their* histograms...
> | 
> | If this is the case, then you certainly have my full-on attention.
> | 
> | Mark
> | 
> | 
> | 
> | Please visit the Group Homepage to check the Files, Bookmarks, 
Polls and other resources as they are often being updated. The page 
is at:
> | 
> | http://groups.yahoo.com/group/DigitalBlackandWhiteThePrint
> | 
> | Please follow these basic guidelines:
> | - Include your full name with your message.
> | - Include the address of your website, if you have one.
> | - As threads develop, trim off excess portions of earlier 
messages to keep them short.
> | - As the topic of a thread changes remember to change the subject 
header.
> | - Good manners are required at all time. No personal attacks 
or "flames."
> | - Complete your Yahoo profile.
> | - Before posting a question, search the message archives and the 
various resources on the homepage. 
> | 
> | 
> |  
> | 
> | Your use of Yahoo! Groups is subject to 
http://docs.yahoo.com/info/terms/ 

> | 
> | 
> |

I forgot to point out the very important fact that my duplicate file 
was 8-bit.

Chris


--- In DigitalBlackandWhiteThePrint@y..., "tzinzunzan2000" 
<ldmr@c...> wrote:
> First, some questions. By 'combed' histogram do you mean only a 
> histogram that contains gaps all the way from bottom to top or do 
you 
> include those those histograms which are slightly 'combed', meaning 
> those that have a base throughout most of the histogram but have 
very 
> narrow (comblike) peaks and valleys? Also, very roughly speaking, 
how 
> bad does this combing effect have to be before it manifests itself 
in 
> output? Second, an observation. I took a greyscale tiff file 
scanned 
> from my SS4000, opened the file as 16 bit in PS, duplicated the 
image 
> and then loaded the same two curves into each image. I got the same 
> histogram changes for both images. Won't hazard to guess the 
> significance of this. Also, I've noticed that sometimes if hit the 
> Auto Levels button I get regularly spaced bottom-to-top gaps in the 
> histogram. 
> 
> --- In DigitalBlackandWhiteThePrint@y..., "Maris V. Lidaka, Sr." 
> <mlidaka@a...> wrote:
> > Yes - that is precisely the cause of the 'combed' histogram.
> > 
> > Maris
> > 
> > ----- Original Message ----- 
> > From: "mtucker508" <mark@m...>
> > To: <DigitalBlackandWhiteThePrint@y...>
> > Sent: Wednesday, December 05, 2001 9:25 AM
> > Subject: Re: [Digital BW] 16-bit Scanning: Why?
> > 
> > 
> > | --- In DigitalBlackandWhiteThePrint@y..., "Austin Franklin" 
> > | <darkroom@i...> wrote:
> > | > What happens when you make tonal adjustments with 8 bits is 
> > | you loose tonal
> > | > values. 
> > | 
> > | 
> > | Would these tones "moving around" and getting shifted be 
> > | responsible for a histogram that had holes in it? (Not that 
I've 
> > | ever had one, but a friend of a friend told me that *they* had 
> > | some missing areas in *their* histograms...
> > | 
> > | If this is the case, then you certainly have my full-on 
attention.
> > | 
> > | Mark
> > | 
> > | 
> > | 
> > | Please visit the Group Homepage to check the Files, Bookmarks, 
> Polls and other resources as they are often being updated. The page 
> is at:
> > | 
> > | http://groups.yahoo.com/group/DigitalBlackandWhiteThePrint
> > | 
> > | Please follow these basic guidelines:
> > | - Include your full name with your message.
> > | - Include the address of your website, if you have one.
> > | - As threads develop, trim off excess portions of earlier 
> messages to keep them short.
> > | - As the topic of a thread changes remember to change the 
subject 
> header.
> > | - Good manners are required at all time. No personal attacks 
> or "flames."
> > | - Complete your Yahoo profile.
> > | - Before posting a question, search the message archives and 
the 

> various resources on the homepage. 
> > | 
> > | 
> > |  
> > | 
> > | Your use of Yahoo! Groups is subject to 
> http://docs.yahoo.com/info/terms/ 
> > | 
> > | 
> > |

Sorry for the long post, but I thought I might add what I've discovered 
with my images. After reading through this thread about high bit image 
editing, I thought it was worth noting that one important distinction 
needs to be made. I've spent some time experimenting with 8 and 16 bit 
BW scans and trying to determine if there is a perceptible difference 
in the output image.  Also have had numerous discussions with Antonis 
on this, who's knowledge of PhotoShop is truly phenomenal, and we've 
recently done some testing with hi-res piezo's output on the 7000. Here 
is what my current conclusions are, based on what I've been able to 
see:

There is no doubt that when you drop from 65,000 levels of gray (16 
bit) to 256 (8 bit), there are pixels that shift in tonality, since 
they are moved to the nearest 256 level gray value. But whether or not 
you will see this, either as a slight compromising of the tonal range 
or as flat-out posterization has to do with the resolution of your 
scan, if it is a B&W film neg we are talking about.  Computer generated 
images are a different matter.  The key seems to me to be the 
distribution of pixels and their relative values to each other.  If, in 
16 bit, you have a large region of pixels that are the same value next 
to another large region of pixels that is one value lighter, you 
probably won't see a difference in a print, as they are only 1/65,000th 
of the tonal spectrum apart.  But if you convert to 8 bit, either PS 
will assign them to the same gray value OR to adjacent values depending 
on the choices it makes, and the two regions then become only 1/256th 
apart, and the border between these areas will be visible in print, ie. 
posterization.

However, and this is the important part, if your scan is high enough in 
resolution, at the pixel level you will see very few regions where 
there are many pixels very close together in value.  In the scan, each 
silver grain in the film is made up of numerous pixels, and these will 
vary substantially in value depending on whether they are located in 
the middle of the grain, the edge, the clear film base between, etc. 
You will see this if you look closely down at pixel level, the image 
actually looks "noisy."  But the perception of tone these pixels render 
is effectively an averaging effect, since we are seeing them from very 
"far away" in a sense, since our prints are not such severe 
enlargements.  So if you take a 16 bit scan and drop to 8 bit, those 
pixels will each shift marginally lighter or darker, but since the 
viewer's impression of tonality is made up of their "average," one 
usually cannot perceive this. And presumably the number of pixels that 
shift lighter and the number that shift darker will be relatively equal 
as this is a random process. Posterization will only occur if you have 
a predominance of one color pixel in a region that is visible at the 
viewer's level of resolution.  This can, BTW, develop in smooth tone 
areas of a hi-res image if you res down, or print _very_ small, as this 
may force pixels that were once only slightly different in value to 
become the same, again producing posterization.

I print a lot of skies, and in low-res scans I have trouble with 
posterizing if I use severe levels or curve tweaks in 8 bit. Or if I 
blur to smooth out grain. But I can get away with the contrast tweaks 
just fine in hi-res scans.  There is no visible difference in the 
output of an image that is 16 bit going to the printer and an 8 bit 
one.  Even though there is combing in the histo's of some of the layers 
in my PS files I can see no difference in print, which is what matters 
to me.

Hopefully this is coherent enough, it's a bit hard to explain using 
just words. Also hope it sheds some light...

Phil
http://philbard.com

> I print a lot of skies, and in low-res scans I have trouble with
> posterizing if I use severe levels or curve tweaks in 8 bit. Or if I
> blur to smooth out grain. But I can get away with the contrast tweaks
> just fine in hi-res scans. 

Isn't it also so, that you get more posterization problems if you apply
several adjustments at a "pixel level" through Image -> Adjust than what
you get if you apply your adjustments in adjustment layers. In the
latter case the shifts on the "pixel level" will be those of one
"resultant" adjustment instead of several.

Martin

Comments interspersed below.

----- Original Message ----- 
From: "tzinzunzan2000" <ldmr@...>
To: <DigitalBlackandWhiteThePrint@yahoogroups.com>
Sent: Wednesday, December 05, 2001 11:06 AM
Subject: Re: [Digital BW] 16-bit Scanning: Why?


| First, some questions. By 'combed' histogram do you mean only a 
| histogram that contains gaps all the way from bottom to top or do you 
| include those those histograms which are slightly 'combed', meaning 
| those that have a base throughout most of the histogram but have very 
| narrow (comblike) peaks and valleys?

I mean gaps all or part of the way but not just peaks and valleys.  You will always have peaks and valleys.

 Also, very roughly speaking, how 
| bad does this combing effect have to be before it manifests itself in 
| output?

I know of no predictable formula - very roughly speaking, a gap of 1/4 of the length of the histogram would result in output problems but less than that it depends on the image.  As Dan Margulis suggests, look at the image and not the histogram.

 Second, an observation. I took a greyscale tiff file scanned 
| from my SS4000, opened the file as 16 bit in PS, duplicated the image 
| and then loaded the same two curves into each image. I got the same 
| histogram changes for both images. Won't hazard to guess the 
| significance of this. 

I would guess it was a fairly detailed image without broad expanses of identical tonality.

Also, I've noticed that sometimes if hit the 
| Auto Levels button I get regularly spaced bottom-to-top gaps in the 
| histogram. 

It indicates that Auto Levels has changed the white and black points, but not the middle gray point, so the new spacing between white and black points would be expected to be evenly distributed.

Again, don't place too much reliance on the histogram - rely on the image.

Maris


| 
| --- In DigitalBlackandWhiteThePrint@y..., "Maris V. Lidaka, Sr." 
| <mlidaka@a...> wrote:
| > Yes - that is precisely the cause of the 'combed' histogram.
| > 
| > Maris
| > 
| > ----- Original Message ----- 
| > From: "mtucker508" <mark@m...>
| > To: <DigitalBlackandWhiteThePrint@y...>
| > Sent: Wednesday, December 05, 2001 9:25 AM
| > Subject: Re: [Digital BW] 16-bit Scanning: Why?
| > 
| > 
| > | --- In DigitalBlackandWhiteThePrint@y..., "Austin Franklin" 
| > | <darkroom@i...> wrote:
| > | > What happens when you make tonal adjustments with 8 bits is 
| > | you loose tonal
| > | > values. 
| > | 
| > | 
| > | Would these tones "moving around" and getting shifted be 
| > | responsible for a histogram that had holes in it? (Not that I've 
| > | ever had one, but a friend of a friend told me that *they* had 
| > | some missing areas in *their* histograms...
| > | 
| > | If this is the case, then you certainly have my full-on attention.
| > | 
| > | Mark

what is your high re scan? 400odpi?

> From: "pbard2002" <phil@...>
> Reply-To: DigitalBlackandWhiteThePrint@yahoogroups.com
> Date: Wed, 05 Dec 2001 17:19:22 -0000
> To: DigitalBlackandWhiteThePrint@yahoogroups.com
> Subject: Re: [Digital BW] 16-bit Scanning: Why?
> 
> Sorry for the long post, but I thought I might add what I've discovered
> with my images. After reading through this thread about high bit image
> editing, I thought it was worth noting that one important distinction
> needs to be made. I've spent some time experimenting with 8 and 16 bit
> BW scans and trying to determine if there is a perceptible difference
> in the output image.  Also have had numerous discussions with Antonis
> on this, who's knowledge of PhotoShop is truly phenomenal, and we've
> recently done some testing with hi-res piezo's output on the 7000. Here
> is what my current conclusions are, based on what I've been able to
> see:
> 
> There is no doubt that when you drop from 65,000 levels of gray (16
> bit) to 256 (8 bit), there are pixels that shift in tonality, since
> they are moved to the nearest 256 level gray value. But whether or not
> you will see this, either as a slight compromising of the tonal range
> or as flat-out posterization has to do with the resolution of your
> scan, if it is a B&W film neg we are talking about.  Computer generated
> images are a different matter.  The key seems to me to be the
> distribution of pixels and their relative values to each other.  If, in
> 16 bit, you have a large region of pixels that are the same value next
> to another large region of pixels that is one value lighter, you
> probably won't see a difference in a print, as they are only 1/65,000th
> of the tonal spectrum apart.  But if you convert to 8 bit, either PS
> will assign them to the same gray value OR to adjacent values depending
> on the choices it makes, and the two regions then become only 1/256th
> apart, and the border between these areas will be visible in print, ie.
> posterization.
> 
> However, and this is the important part, if your scan is high enough in
> resolution, at the pixel level you will see very few regions where
> there are many pixels very close together in value.  In the scan, each
> silver grain in the film is made up of numerous pixels, and these will
> vary substantially in value depending on whether they are located in
> the middle of the grain, the edge, the clear film base between, etc.
> You will see this if you look closely down at pixel level, the image
> actually looks "noisy."  But the perception of tone these pixels render
> is effectively an averaging effect, since we are seeing them from very
> "far away" in a sense, since our prints are not such severe
> enlargements.  So if you take a 16 bit scan and drop to 8 bit, those
> pixels will each shift marginally lighter or darker, but since the
> viewer's impression of tonality is made up of their "average," one
> usually cannot perceive this. And presumably the number of pixels that
> shift lighter and the number that shift darker will be relatively equal
> as this is a random process. Posterization will only occur if you have
> a predominance of one color pixel in a region that is visible at the
> viewer's level of resolution.  This can, BTW, develop in smooth tone
> areas of a hi-res image if you res down, or print _very_ small, as this
> may force pixels that were once only slightly different in value to
> become the same, again producing posterization.
> 
> I print a lot of skies, and in low-res scans I have trouble with
> posterizing if I use severe levels or curve tweaks in 8 bit. Or if I
> blur to smooth out grain. But I can get away with the contrast tweaks
> just fine in hi-res scans.  There is no visible difference in the
> output of an image that is 16 bit going to the printer and an 8 bit
> one.  Even though there is combing in the histo's of some of the layers
> in my PS files I can see no difference in print, which is what matters
> to me.
> 
> Hopefully this is coherent enough, it's a bit hard to explain using
> just words. Also hope it sheds some light...
> 
> Phil
> http://philbard.com
> 
> 
> 
> 
> Please visit the Group Homepage to check the Files, Bookmarks, Polls and other
> resources as they are often being updated. The page is at:
> 
> http://groups.yahoo.com/group/DigitalBlackandWhiteThePrint
> 
> Please follow these basic guidelines:
> - Include your full name with your message.
> - Include the address of your website, if you have one.
> - As threads develop, trim off excess portions of earlier messages to keep
> them short.
> - As the topic of a thread changes remember to change the subject header.
> - Good manners are required at all time. No personal attacks or "flames."
> - Complete your Yahoo profile.
> - Before posting a question, search the message archives and the various
> resources on the homepage.
> 
> 
> 
> 
> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
> 
>

on 12/5/01 12:50 PM, Martin Glader wrote:

> Isn't it also so, that you get more posterization problems if you apply
> several adjustments at a "pixel level" through Image -> Adjust than what
> you get if you apply your adjustments in adjustment layers. In the
> latter case the shifts on the "pixel level" will be those of one
> "resultant" adjustment instead of several.

Yes and no.

There is a misconception that if you have a stack of adjustment layers they
are somehow combined first, and then applied to the image as one grand move.
That is not the case. Adjustment layers are applied to the file one by one,
from the bottom to the top.

The scenario by which they are less destructive is that you can reopen your
adjustment layers and re-manipulate it, change it's opacity, it's apply
mode, it's blend mode, over and over again,  but whatever your final choice
for that layer, the manipulation is only applied once. So, it's possible to
tweak and experiment ad infinitum before committing the move to the image.

Todd

I'm using 2400 dpi scans from 4x5 TMax 100.  It's about a 100MB 
grayscale file...


Phil
http://philbard.com


--- In DigitalBlackandWhiteThePrint@y..., ternahan <ternahan@s...> 
wrote:
> what is your high re scan? 400odpi?

Austin, Todd, Anybody Else,

Would it be correct to think about it this way:

An 8-bit file is like a 1/8" thick piece of glass. A 16-bit file is like a 
1/2" thick piece of glass. As in, you can toss and work with and 
manipulate the 1/2" thick piece without doing much damage to it, 
whereas the 1/8" thick piece is very fragile, as in -- It's much 
easier to mess up, or "comb", a histogram in an 8-bit file, 
whereas you can tone, or Level, or Curve, a 16-bit much more, 
before introducing any degree of "combing"?

(Would somebody care to diagram that prior sentence?)

I've never known what I've done to a file when I check the 
Histogram and see the combing. I work with most all files the 
exact same way. Some stay nice and smooth, others get pretty 
ugly. Maybe I'd run into much less chance of ugly ones if I'd do 
most of that initial burning/doding/leveling/curving in 16-bit?

-Mark

> Maybe I'd run into much less chance of ugly ones if I'd do
> most of that initial burning/doding/leveling/curving in 16-bit?

If you do your tone curves in the scanner driver, it will be done on high
bit data, no matter whether you are getting 8 bit or high bit data out.  My
workflow is to do ALL my tonal corrections in the original scan, and save as
8 bit data, and do NO tonal corrections in PS.

> > The fact that you lose tonal values isn't theoretical, it's just a
> plain old
> > fact...kind of like 1 apple + 1 apple = 2 apples.  It's a simple
> experiment
> > you can try your self ;-)
> >
> > Regards,
> >
> > Austin
>
> Not taking sides, I just think he was trying to say, that yes you do
> lose tonal value (numerically if you do the math), but with the
> quality of printers available, it won't show. Again, I don't have
> enough experience to say, one way or the other.

Hi,

Quality of printer doesn't matter.  I believe he (being Dan Margulis) is
talking about color, and WE, as in this list, are talking about B&W ONLY.
You do see it in the print, that's my point!

Regards,

Austin

> Again, don't place too much reliance on the histogram - rely on the image.
>
> Maris

Maris,

For color, I won't dispute that at this point, but for B&W, which is
exclusively what we are talking about on this list, I DO dispute that.  I
HEAVILY rely on my histogram...and I get perfect prints.  When I have gaps,
I almost always have a problem when I print that image.  When I re-scan and
get a great histogram, my prints are always great.

I WOULD suggest relying HEAVILY on the histogram for B&W.

Regards,

Austin

Austin,

But if your scanner only scans at 10, 12, or 14 bit, you would surely be
much better to do the scans at 16bit in PS, even if it means going from 8bit
output to 16 bit in PS, before doing them. And your method depends on the
scanner software being as good as PS in doing the tonal corrections - many
aren't from what I read. You also can't save the various stages of tonal
correction if you do them in the scanner.

In PS I can save each stage in high-bit tonal correction, in case I want to
go back and change something. Vuescan allows some tonal changes to be made
to the saved raw scan, but I still believe that most are best done in PS,
and Bruce Fraser recommends sending images to the printer in 16bit, thus
allowing PS to convert to the printer color space in 16bit, rather than
8bit.

Bob Frost

----- Original Message -----
From: "Austin Franklin" <darkroom@...>

If you do your tone curves in the scanner driver, it will be done on high
bit data, no matter whether you are getting 8 bit or high bit data out.  My
workflow is to do ALL my tonal corrections in the original scan, and save as
8 bit data, and do NO tonal corrections in PS.

Hi Phil,

> There is no doubt that when you drop from 65,000 levels of gray (16
> bit) to 256 (8 bit), there are pixels that shift in tonality, since
> they are moved to the nearest 256 level gray value.

ALL that happens when you convert from 16 bit to 8 bits is the bottom 8 bits
is lopped off and the top 8 is used...NO "real" "processing" at all is done,
it's simply truncation.  It is shifting tonality, in a sense, but it is
really just truncation.

> But if you convert to 8 bit, either PS
> will assign them to the same gray value OR to adjacent values depending
> on the choices it makes,

See above, that isn't really what happens...

> In the scan, each
> silver grain in the film is made up of numerous pixels

Well, possibly if you have very grainy film, or are scanning at a very high
resolution...typically, with decent exposure, film and processing, you will
have many film grains represented by a single pixel.

> ...But the perception of tone these pixels render
> is effectively an averaging effect, since we are seeing them from very
> "far away" in a sense...

Exactly how halftoning works...

> Posterization will only occur if you have
> a predominance of one color pixel in a region that is visible at the
> viewer's level of resolution.

Correct.  Technically, you ALWAYS have posterization...but whether it is
visible or not what is important.

Regards,

Austin

Hi Bob,

> But if your scanner only scans at 10, 12, or 14 bit, you would surely be
> much better to do the scans at 16bit

I don't know what you mean.  No matter what the scanner bit depth, if it is
above 8, it returns 16 bit data...you even though you get 16 bits from the
scanner, you really only get 10/12/14 bits of VALID image data.

> in PS, even if it means
> going from 8bit
> output to 16 bit in PS, before doing them.

Well, I don't know that that will work as you think it will.  Let me think
this through  When you convert from 8 bit to 16 bit, you are just using the
top 8 bits of data, and the bottom are now all zeros.  Now there are 255
"gaps" between each of the valid image data points...and when you make the
moves, you then re-map them to new 16 bit values, so the lower 8 bits will
now be being used...BUT you have only 256 discrete values in the first
place.  When you go back to 8 bits though, you are just lopping off the
bottom 8 bits and any of the values that were re-mapped into values that the
top 8 bits are the same, will now not be distinct...so I believe this method
does not necessarily work, and can cause losing some of your 256 original
values.  Let me think more about it...unless you or someone else has a
thought on this...

> And your method depends on the
> scanner software being as good as PS in doing the tonal corrections - many
> aren't from what I read.

I can't imagine ANY scanner software NOT doing the tonal adjustments on high
bit data, but possibly some do...  Tonal adjustments are really nothing more
than a simple look-up table and a re-mapping, they really aren't complex
algorithms, so one being "better or worse" would be hard to do...what would
make one better or worse was if they did the mapping on low bit data instead
of high bit data.

> You also can't save the various stages of tonal
> correction if you do them in the scanner.

What do you mean by that?  I can save my tonal correction curves...if that's
what you mean?  The tonal corrections are only done in the preview window,
not to the actual scanned data, at least as far as you are seeing when
making the corrections.  What you "settle on" in the preview window is then
downloaded to the LUTs in the scanner (Look Up Tables) and applied to the
actual scanned data.  Typically preview is done at a much lower resolution
than the actual scan.

> In PS I can save each stage in high-bit tonal correction, in case
> I want to
> go back and change something. Vuescan allows some tonal changes to be made
> to the saved raw scan, but I still believe that most are best done in PS,
> and Bruce Fraser recommends sending images to the printer in 16bit, thus
> allowing PS to convert to the printer color space in 16bit, rather than
> 8bit.

Er, we're talking B&W here, and unless you have the Piezo Pro or what ever
it's called, you can't send 16 bit data to the printer...and if you aren't
using SOME form of quad tome printing, well, 8 bits of 16 bits isn't going
to matter...

Regards,

Austin

on 12/5/01 5:11 PM, Austin Franklin wrote:

>> in PS, even if it means
>> going from 8bit
>> output to 16 bit in PS, before doing them.
> 
> Well, I don't know that that will work as you think it will.  Let me think
> this through  When you convert from 8 bit to 16 bit, you are just using the
> top 8 bits of data, and the bottom are now all zeros.  Now there are 255
> "gaps" between each of the valid image data points...and when you make the
> moves, you then re-map them to new 16 bit values, so the lower 8 bits will
> now be being used...BUT you have only 256 discrete values in the first
> place.  When you go back to 8 bits though, you are just lopping off the
> bottom 8 bits and any of the values that were re-mapped into values that the
> top 8 bits are the same, will now not be distinct...so I believe this method
> does not necessarily work, and can cause losing some of your 256 original
> values.  Let me think more about it...unless you or someone else has a
> thought on this...


I don't know what really happens numerically when you do an 8-bit to 16-bit
conversion, but I don't think it helps. When you do a rotation, or an
upsampling, or gaussian blur, you smooth your histogram, but what you don't
do is add real discrete values, values which have meaning as detail. Those
operations cause some averaging (Austin, keep me honest on my terminology
;-), but they do not increase dynamic range, or detail (detail defined as
the ability to distinguish between discrete tones). In fact, they increase
the likelihood of banding.

I still don't know exactly why that's true, but, I remember earlier on I was
printing smooth gradients and getting banding, both onscreen and in print. I
kept trying to blur the gradient to smooth it out, but the more I did the
worse it got. Finally I asked for help on some PS lists and was shocked to
learn that noise was the common cure for banding.

I guess it's that something needs to fill in between those missing tones and
the randomness of noise breaks the edges. Somehow the purity of averaging
doesn't hide as well. But that's why images with lots of detail fare better
in 8-bits than very smooth images.

Having just said that Austin, it occurs to me that possibly that is one
reason your images tend to posterize. You say your images are virtually
grainless, you don't sharpen, and I see your tonalities are very smooth, so
your type of image might just be prone to that sort of thing. They always
look good though.

Nice picture of your kid BTW. He's a good looking kid.

Todd

Hi Todd,

> Having just said that Austin, it occurs to me that possibly that is one
> reason your images tend to posterize.

Well, my images DON'T posterize, since I know what to do so I don't get
posterization ;-)

> You say your images are virtually
> grainless

Most are, but my 800ASA+ images usually have visible grain...

> you don't sharpen

Nope.

> and I see your tonalities are very
> smooth,

Why thank you.

> so
> your type of image might just be prone to that sort of thing.

What sort of thing, posterizing?  I don't know if my images are or are not
prone to it, since I don't get it...and I scan for full histograms.

> They always
> look good though.

The difference between a good photographer and a great photographer is great
photographers only let people see the great images...or so I hear.

> Nice picture of your kid BTW. He's a good looking kid.

That would be his mother's fault, not mine, I assure you ;-)

Thanks!

Austin

HI Austin-

>Hi Bruce,
>
>>  I can't help wondering if there's a corollary here to digital audio,
>>  if I can drift off-topic.
>
>I've delved into that quite a bit...having designed a lot of digital audio
>gear...


Yeah, attracts similar types:-)

>
>>  Problem was (and is) that the 16/44 CD never sounded as "good" as
>>  analog tape (or the venerable LP) in a high resolution audio system.
>>  The decimation, data jitter, and aliasing problems become easily
>>  audible.
>
>Initially and up until about 7 years ago, I completely agree...the
>oversampling filters were really pretty bad, and they did sound bad.  I
>believe you mean interpolation, since decimation is decreasing data, and
>interpolation is used to increase the data to smooth out the data to be more
>analog (less distance between voltage steps), so a lower order filter can be
>used which decreases aliasing.

No, I meant 48 or 96 to 44, and 20-24 to 8. Oversampling and 
upsampling has always seemed like just more bandaids.

>
>With proper design, all the problems you mention can all be severely reduced
>so they are not audible at all.  I have a very high end CD player (ML39),
>that used a very very well designed analog front end, as well as every
>conceivable method to keep analog noise down...and it easily sounds better
>than any album, and certainly far better than any tape.  You're welcome to
>come listen any time ;-)

Sure, would love to. But the ML is not a mass market product. At a 
very steep price, it mitigates problems that cannot be avoided in 
affordable gear. And I wonder how it can sound better than a master 
tape:-)

>  This is due to painstaking engineering, and
>considerable expense.

Yes, and I haven't been able to deal with the 1160 because I've been 
busy scanning 5x7s on a Scitex Eversmart, also painstaking and 
considerable:-)

>
>You make a good point, but I do believe that static visualization
>(photography) is very different than dynamic audiblization (audio....ha, I
>made a new word, I think ;-).

Well there are a lot of speaker designers who only look at impulse 
data. But looking at art-or a photograph-is an active, Minor White 
used to say creative, activity. You can learn to look at an 8x10 
contact differently than a disposable CVS 4x6. Or a platinum print, a 
Daguerreotype, or a piezo. Just like learning to listen to a 
symphony-the composition, not the orchestra or the radio. The 
carrier, audio or visual, has to be transparent to the source, the 
art, to be effective. Cinema does it by not letting you see a frame 
long enough. DVD doesn't, because it compresses data in areas that 
don't move. Drives me to distraction.

I'd suspect the reason you don't sharpen is because you know what a 
good lens looks like, and USM doesn't always look kosher.

Regards,

Bruce
--

HI Austin-

>Hi Bruce,
>
>>  I can't help wondering if there's a corollary here to digital audio,
>>  if I can drift off-topic.
>
>I've delved into that quite a bit...having designed a lot of digital audio
>gear...


Yeah, attracts similar types:-)

>
>>  Problem was (and is) that the 16/44 CD never sounded as "good" as
>>  analog tape (or the venerable LP) in a high resolution audio system.
>>  The decimation, data jitter, and aliasing problems become easily
>>  audible.
>
>Initially and up until about 7 years ago, I completely agree...the
>oversampling filters were really pretty bad, and they did sound bad.  I
>believe you mean interpolation, since decimation is decreasing data, and
>interpolation is used to increase the data to smooth out the data to be more
>analog (less distance between voltage steps), so a lower order filter can be
>used which decreases aliasing.

No, I meant 48 or 96 to 44, and 20-24 to 8. Oversampling and 
upsampling has always seemed like just more bandaids.

>
>With proper design, all the problems you mention can all be severely reduced
>so they are not audible at all.  I have a very high end CD player (ML39),
>that used a very very well designed analog front end, as well as every
>conceivable method to keep analog noise down...and it easily sounds better
>than any album, and certainly far better than any tape.  You're welcome to
>come listen any time ;-)

Sure, would love to. But the ML is not a mass market product. At a 
very steep price, it mitigates problems that cannot be avoided in 
affordable gear. And I wonder how it can sound better than a master 
tape:-)

>  This is due to painstaking engineering, and
>considerable expense.

Yes, and I haven't been able to deal with the 1160 because I've been 
busy scanning 5x7s on a Scitex Eversmart, also painstaking and 
considerable:-)

>
>You make a good point, but I do believe that static visualization
>(photography) is very different than dynamic audiblization (audio....ha, I
>made a new word, I think ;-).

Well there are a lot of speaker designers who only look at impulse 
data. But looking at art-or a photograph-is an active, Minor White 
used to say creative, activity. You can learn to look at an 8x10 
contact differently than a disposable CVS 4x6. Or a platinum print, a 
Daguerreotype, or a piezo. Just like learning to listen to a 
symphony-the composition, not the orchestra or the radio. The 
carrier, audio or visual, has to be transparent to the source, the 
art, to be effective. Cinema does it by not letting you see a frame 
long enough. DVD doesn't, because it compresses data in areas that 
don't move. Drives me to distraction.

I'd suspect the reason you don't sharpen is because you know what a 
good lens looks like, and USM doesn't always look kosher.

Regards,

Bruce

-- 

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

Hi Bruce,

> Oversampling and
> upsampling has always seemed like just more bandaids.

Not a bandaid at all.  ALL digital output requires some kind of filter to
smooth out the stepping, it's just part of digital to analog conversion....
I can explain it in detail if you like.  Digital upsampling is simply to
reduce the requirements on the analog output filters, which as you correctly
noted, have a "skew" problem with different frequencies traveling through
the filter at different timing, and therefore distorting the phase
relationship between frequencies.  This is solved by oversampling, which
does allows the use of first order filters, which inherently don't have near
the phase issue third order filters do.  I tend to use Butterworth filters
;-)

> Sure, would love to.

Any time.  Wish we had spoken more when you were up here!

> But the ML is not a mass market product. At a
> very steep price, it mitigates problems that cannot be avoided in
> affordable gear. And I wonder how it can sound better than a master
> tape:-)

Tape doesn't have as high a dynamic range as a CD does...most masters today
are not on analog tape, but on digital tape...which is an entirely different
beast.

> The
> carrier, audio or visual, has to be transparent to the source, the
> art, to be effective.

Yes, exactly!

> DVD doesn't, because it compresses data in areas that
> don't move. Drives me to distraction.

I agree with DVD...MPEG has always driven me crazy, because I can see the
glitches in it.

> I'd suspect the reason you don't sharpen is because you know what a
> good lens looks like, and USM doesn't always look kosher.

As I said in another post, typically, the issue isn't the scanner...it's the
exposure/development/lense, and unless all those are at their best, scanning
at 4000+ DPI is going to just reproduce those defects...kind of like playing
an 8 track through a $10k stereo...it isn't going to sound very good ;-)

Regards,

Austin

>> Oversampling and
>> upsampling has always seemed like just more bandaids.
> 
> Not a bandaid at all.  ALL digital output requires some kind of filter to
> smooth out the stepping, it's just part of digital to analog conversion....
> I can explain it in detail if you like.

Isn't this similar to the argument for sharpening?

I've always heard the "need" for sharpening explained something like, "all
CCD scans require some kind of sharpening to compensate for the softening
effects of scanning, it's just part of the analog to digital conversion, as
optics are involved which can soften and image, and CCDs are prone to blur
and smear."

However I totally agree the need for it is an aesthetic choice, and all
scanners are not created equal.

Todd

Hi Todd,

> >> Oversampling and
> >> upsampling has always seemed like just more bandaids.
> >
> > Not a bandaid at all.  ALL digital output requires some kind of
> filter to
> > smooth out the stepping, it's just part of digital to analog
> conversion....
> > I can explain it in detail if you like.
>
> Isn't this similar to the argument for sharpening?

No.  Actually, it's the opposite!  Oversampling does the OPPOSITE of
sharpening, it smooth out the transitions between samples...where sharpening
decreases the width of the transitions...sharpness is basically
contrast...and to increase contrast, you have to increase the difference
between values.

> I've always heard the "need" for sharpening explained something like, "all
> CCD scans require some kind of sharpening to compensate for the softening
> effects of scanning, it's just part of the analog to digital
> conversion, as
> optics are involved which can soften and image, and CCDs are prone to blur
> and smear."

Well, I obviously disagree with that ;-)

Regards,

Austin

> I've always heard the "need" for sharpening explained something like, "all
> CCD scans require some kind of sharpening to compensate for the softening
> effects of scanning, it's just part of the analog to digital
> conversion, as
> optics are involved which can soften and image, and CCDs are prone to blur
> and smear."

Todd,

I thought about this a bit.  Remember, the Leafscan (which is the scanner I
have) only uses a neutral density filter for B&W scans, therefore
eliminating the smear and bloom characteristics of using red and blue
filters with CCDs.  That may be the real reason why I don't sharpen, aside
from the fact my images are shot with high quality glass, exposed and
developed for sharpness and tonality.

I've always said the Leaf will be a B&W "cult" scanner because of this
unique ability to scan B&W without using color filters... ;-)

Regards,

Austin

>> I've always heard the "need" for sharpening explained something like, "all
>> CCD scans require some kind of sharpening to compensate for the softening
>> effects of scanning, it's just part of the analog to digital
>> conversion, as
>> optics are involved which can soften and image, and CCDs are prone to blur
>> and smear."
> 
> Todd,
> 
> I thought about this a bit.  Remember, the Leafscan (which is the scanner I
> have) only uses a neutral density filter for B&W scans, therefore
> eliminating the smear and bloom characteristics of using red and blue
> filters with CCDs.  That may be the real reason why I don't sharpen, aside
> from the fact my images are shot with high quality glass, exposed and
> developed for sharpness and tonality.

Austin,

I really don't disagree, per se. My sharper films and images do "require"
less sharpening. I guess it boils down to what we want from sharpening.

In the case of the Leaf, in my tests, which were by no means exhaustive,
(maybe 8 comparisons), my grayscale scans were always the 2nd sharpest
choice short of choosing the singularly sharpest of the three color
channels. IOW, usually just the green channel would be slightly sharper than
the grayscale scan (though sometimes blue, and once even red). But I'm
really talking on the level of grain sharpness - nothing you'd see at normal
magnification. I decided it wasn't a sufficient difference to merit the
longer scan time, and if I wanted the additional sharpness I could get it
with USM. But which channel is the right sharpness? I also look at dust
specks and decide how sharp they should be. It's really purely subjective as
to how sharp that little edge should be, no?

I'm guessing the tradition of sharpening has come about for two reasons:

A) Perhaps through a microscope they measure the performance of a particular
lens/film/dev combo, at reproducing a resolution test chart. They'd then
scan that film and measure the performance on the monitor, and decide that
in order to match the films performance on the monitor, or for a given
output, sharpening needs to be employed. That's my guess. For real images
more subjective reasons might prevail in deciding whether sharpening is
worthwhile or not.

B) As I understand it, traditional offset half tone screening breaks up
image sharpness, and printers have found some "over sharpening" of an image
prior to printing could counteract that effect. In my case I've found that
my 1160/Epson Driver prints actually come out looking sharper than my
monitor image! So again, different rules may apply here.

I think it's all subjective. USM is just another tool to be used and abused.
;-)

Todd

Hi Todd,

> I'm guessing the tradition of sharpening has come about for two reasons:

USM actually is a darkroom printing operation...I never used it.

> I think it's all subjective. USM is just another tool to be used
> and abused.
> ;-)

Agreed, but before believing you NEED to sharpen, you might want to find out
why, and it may be simply your cheapo FooYuck 16-340 f2.8 zoom, bad focus,
bad exposure or bad development...as in your images on the film ARE fuzzy in
the first place!  I think also that scanned images give people the ability
to REALLY zoom in on them, and see larger than typically viewed images...

Regards,

Austin

Actually the thread at Dan's ledet.com site was in 1999 and touched minimally on this issue - the I sent you in text form was this year's thread which covered it thoroughly but is not at the ledet.com site

Maris

----- Original Message ----- 
From: "Todd Flashner" <tflash@...>
To: <DigitalBlackandWhiteThePrint@yahoogroups.com>
Sent: Wednesday, December 05, 2001 12:12 PM
Subject: Re: [Digital BW] 16-bit Scanning: Why?

Austin I've combined two of your posts into one.


[snipped]

| If you want the thread from the color theory list Maris gave the
| link, or I could send you a text file Maris complied which covers a good bit
| of it. 

[remainder snipped]