Velocity Reviews - Computer Hardware Reviews

Velocity Reviews > Newsgroups > Computing > Digital Photography > Claimed high scanned film "information" is mostly garbage

Reply
Thread Tools

Claimed high scanned film "information" is mostly garbage

 
 
Kennedy McEwen
Guest
Posts: n/a
 
      04-02-2009
In article <gr0k9k$5gv$(E-Mail Removed)>, John McWilliams
<(E-Mail Removed)> writes
>dan c. wrote:
>> On Apr 1, 8:28 am, Kennedy McEwen <(E-Mail Removed)> wrote:

>
>>> Firstly, if you save both images uncompressed then the 200 ISO and the
>>> 3200 ISO files will be exactly the same size. Ie. it is the compression
>>> that is generating the size difference.

>
>This part is incorrect. Higher ISOs will have more noise, other things
>being equal.
>

If the files are uncompressed then they will be the same size - even if
one of them is ONLY noise. What I said is CORRECT. It is the
compression that generates the file size difference
--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's ****ed.
Python Philosophers (replace 'nospam' with 'kennedym' when replying)
 
Reply With Quote
 
 
 
 
Scott W
Guest
Posts: n/a
 
      04-02-2009
On Apr 2, 7:21*am, Bob Larter <(E-Mail Removed)> wrote:
> John McWilliams wrote:
> > David J Taylor wrote:
> >> John McWilliams wrote:
> >>> dan c. wrote:
> >>>> On Apr 1, 8:28 am, Kennedy McEwen <(E-Mail Removed)> wrote:

>
> >>>>> Firstly, if you save both images uncompressed then the 200 ISO and
> >>>>> the 3200 ISO files will be exactly the same size. *Ie. it is the
> >>>>> compression that is generating the size difference.

>
> >>> This part is incorrect. Higher ISOs will have more noise, other things
> >>> being equal.

>
> >> How will having more noise affect the size on an uncompressed image?

>
> >> Agreed that, when compressed, a noisier image may have a larger file
> >> size.

>
> > My RAW files, as they come from the camera, are of different sizes, not
> > by much, but a few percentage points up or down. Are you saying camera
> > compression of the RAW data causes this?

>
> Yes. RAW files are losslessly compressed, & the file size will increase
> with increases in detail or noise.
>


And on at least some cameras the camera takes this into account when
estimating how many more shots can fit on a memory card, go from iso
100 to 1600 and the number of shot left will show less. A quick
test on my 350D shows 407 left at iso 100 but only 352 left at iso
1600.



 
Reply With Quote
 
 
 
 
Scott W
Guest
Posts: n/a
 
      04-02-2009
On Apr 2, 12:19*pm, Alan Browne <(E-Mail Removed)>
wrote:

> At scan time there are 13, maybe 14 bits of scan info. *The scan s/w
> might curve the data but that does nothing to areas which are near black
> (0 valued). *Examination of .tif images from 16 bit Minolta and Nikon
> scanners shows no detail in the darkest areas. *It's dead black.


If you are scanning a negative then the dead black areas is where the
scanner over exposed the negative, once it is clips to the max value
there is no more shadow detail posible.

I use a Minolta, which has pretty brain dead software to it tends to
over expose negatives all the time. I had to scan as a positive and
then invert the image and play with the curves and color to get it to
look right. VueScan is a bit better but even there it is not prefect.


 
Reply With Quote
 
Scott W
Guest
Posts: n/a
 
      04-02-2009
On Mar 30, 1:04*pm, RichA <(E-Mail Removed)> wrote:
> We've often heard the claim from filmists that a film image contains a
> lot more information than a digital image. *This is true, but the
> information is useless junk. *When colour film is scanned at 4800 dpi
> and 48 bits, it generates huge files. *But much of this is worthless
> (even detrimental) as far as the actual image is concerned. *What the
> scan is recording is mostly information about the grain of the film
> that does not contribute (except what we'd call noise) to the image.
> A high resolution scan records every aspect of the grain and colour
> clouds, even the info for the garbage. *For proof of this from a
> digital perspective, take two shots of a subject, one at 200 ISO and
> one at 3200 ISO. *Now, crop them down to equal sized areas from the
> image and save them. *Take a look at the file size. *The grainy, high
> ISO image can be as much as twice as large because there was more
> information to save, but it certainly did nothing to contribute to the
> image's quality, in fact, because the information represented mostly
> noise, it hurt the image as high ISO does.


I don't know why this matters so much to you, but for what it is worth
I think you are correct.

It was a shock when I started scanning film that I could not compress
the images nearly as small as I could from my digital. With too much
compressing the noise in the sky went from annoying random noise to
disastrous jpeg artifacts. This was at a time when hard disk space
was not cheap and so it did matter, now who cares how large a scanned
image file is.
 
Reply With Quote
 
Kennedy McEwen
Guest
Posts: n/a
 
      04-03-2009
In article
<(E-Mail Removed)>,
Scott W <(E-Mail Removed)> writes
>
>And on at least some cameras the camera takes this into account when
>estimating how many more shots can fit on a memory card, go from iso
>100 to 1600 and the number of shot left will show less. A quick
>test on my 350D shows 407 left at iso 100 but only 352 left at iso
>1600.
>

The 5D is similar.

My point is that this is not an intrinsic consequence of the
higher/lower ISO, but of the "compressability" of noise.
--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's ****ed.
Python Philosophers (replace 'nospam' with 'kennedym' when replying)
 
Reply With Quote
 
Kennedy McEwen
Guest
Posts: n/a
 
      04-03-2009
In article <(E-Mail Removed)>, Alan Browne
<(E-Mail Removed)> writes
>
>At scan time there are 13, maybe 14 bits of scan info. The scan s/w
>might curve the data but that does nothing to areas which are near
>black (0 valued). Examination of .tif images from 16 bit Minolta and
>Nikon scanners shows no detail in the darkest areas. It's dead black.
>

If its "dead black" then it certainly exceeds the scanner's dynamic
range. It is "dead black" with the much older 10-bit LS-20 scanner. By
your "analysis" this shows that there is no more than 10-bits of dynamic
range at scan time.

You are misinterpreting (at best) or misrepresenting (at worst) the
evidence!
--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's ****ed.
Python Philosophers (replace 'nospam' with 'kennedym' when replying)
 
Reply With Quote
 
Kennedy McEwen
Guest
Posts: n/a
 
      04-03-2009
In article
<(E-Mail Removed)>,
Scott W <(E-Mail Removed)> writes
>
>It was a shock when I started scanning film that I could not compress
>the images nearly as small as I could from my digital. With too much
>compressing the noise in the sky went from annoying random noise to
>disastrous jpeg artifacts. This was at a time when hard disk space
>was not cheap and so it did matter, now who cares how large a scanned
>image file is.


I think if you had assessed the effect of noise on compression and had
used something like Noise Ninja, even at its lowest levels, this would
never have been a problem for you.

All that RichA is stating is that conventional compression algorithms do
not compress noise. That is totally unrelated to his claim that digital
has more information content than film (though nobody argues on total
information, just resolution).
--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's ****ed.
Python Philosophers (replace 'nospam' with 'kennedym' when replying)
 
Reply With Quote
 
Kennedy McEwen
Guest
Posts: n/a
 
      04-03-2009
In article <(E-Mail Removed)>, David J.
Littleboy <(E-Mail Removed)> writes
>
>"Alan Browne" <(E-Mail Removed)> wrote:
>> Kennedy McEwen wrote:
>>>
>>>> Where the film itself does not resolve to the ability of the scanner is
>>>> further waste as well. Where a 4000 - 6000 dpi scan of high res film
>>>> does yield mainly useful information, that is not so of most ISO 100 and
>>>> higher films.
>>>>
>>> However, films like Provia can and do exceed the resolution of 4000ppi
>>> scanners, especially with high contrast images.

>>
>> I said "most" not "all".

>
>But Provia 100F with real world images rarely, if ever, shows detail
>significantly* beyond what a Nikon 9000 will get. This "desktop scanners
>don't get everything from film" mantra is almost always film types trying to
>insist, for example, that 35mm is better than 12MP FF digital, when it's not
>even close.
>

That is the issue David. Film runs out of performance at an MTF
dependent on the scene contrast at that limiting resolution. Digital
runs out of resolution at the Nyquist limit, where the MTF is typically
65%. Its horses for courses - there are requirements where resolution
is king and the scene contrast can be bumped up as high as you like (a
good example is the photo-lithography used for semiconductors, including
digital sensors themselves), in other applications scene contrast
seriously limits the achievable resolution (and the higher MTF of
digital at its limiting resolution wins).

I don't have any problem accepting that in most examples my EOS 5D
outperforms many of my Provia shots. But I do have Provia examples
which have much finer detail that could ever be achieved by the 5D.
Indeed, I have a link somewhere to one of your own images which
demonstrate the resolution limits of the 5D, with aliasing all over the
place. I bet you could get way beyond that limitation on Provia film!

This is basically the third point on my original response to RichA's
deliberately provocative post. I don't know anyone that claims film has
more information than digital, but plenty claim more resolution (while
the recent crop of digital cameras from Canon, Nikon and Sony challenge
even that, it wouldn't surprise me if film still shows an edge in some
sheer resolution tests).
--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's ****ed.
Python Philosophers (replace 'nospam' with 'kennedym' when replying)
 
Reply With Quote
 
John McWilliams
Guest
Posts: n/a
 
      04-03-2009
Kennedy McEwen wrote:
> In article <gr0k9k$5gv$(E-Mail Removed)>, John McWilliams
> <(E-Mail Removed)> writes


>> This part is incorrect. Higher ISOs will have more noise, other things
>> being equal.
>>

> If the files are uncompressed then they will be the same size - even if
> one of them is ONLY noise. What I said is CORRECT. It is the
> compression that generates the file size difference.


You are correct, Sir; I erred.

--
john mcwilliams

"My wife said, 'It's either me or the ham radio. There's not enough room
for both of us.' Over."
 
Reply With Quote
 
Kennedy McEwen
Guest
Posts: n/a
 
      04-03-2009
In article <(E-Mail Removed)>, David J.
Littleboy <(E-Mail Removed)> writes
>
>"Kennedy McEwen" <(E-Mail Removed)> wrote:
>> David J. Littleboy <(E-Mail Removed)> writes
>>>>>>
>>>>> However, films like Provia can and do exceed the resolution of 4000ppi
>>>>> scanners, especially with high contrast images.
>>>>
>>>> I said "most" not "all".
>>>
>>>But Provia 100F with real world images rarely, if ever, shows detail
>>>significantly* beyond what a Nikon 9000 will get. This "desktop scanners
>>>don't get everything from film" mantra is almost always film types trying
>>>to
>>>insist, for example, that 35mm is better than 12MP FF digital, when it's
>>>not
>>>even close.
>>>

>> That is the issue David. Film runs out of performance at an MTF dependent
>> on the scene contrast at that limiting resolution.

>
>Actually, it doesn't. Film's failure to render textures correctly at
>enlargements much over 4x or 5x means that all that gorgeous test chart
>resolution is completely meaningless for pictorial photography.


There you go again, illustrating my point whilst attempting to dispute
it - the texture you are rendering is low contrast while the test chart
is high contrast.

> Hit film
>with enough noise reduction to lose the grain, and you lose the detail too.
>
>> Digital runs out of resolution at the Nyquist limit, where the MTF is
>> typically 65%.

>
>Huh? Most Bayer cameras are doing OK at 70% or so of Nyquist


You are clearly talking about the luminance Nyquist, not the Nyquist of
the Bayer filter array!
>
>All the squawking about resolution and information and K25 and 8000 ppi drum
>scans is completely irrelevant to real life: when you enlarge any current
>practical film the 13x required to get you to 12x18, it looks like cr@p. But
>the 5D hangs in there just fine.
>

Utter *******s. I have a stack of 12x18" prints hanging on my walls,
some made with film (Provia and Velvia) and some made with the 5D. You
would be hard pressed to tell the difference. None of them are from K25
originals, but I don't expect that would be any worse, given the
performance of the emulsion.


>>Indeed, I
>> have a link somewhere to one of your own images which demonstrate the
>> resolution limits of the 5D, with aliasing all over the place. I bet you
>> could get way beyond that limitation on Provia film!

>
>Presumably, you are thinking of this image:
>
>http://www.pbase.com/davidjl/image/102380922/original
>

No - I'll post the link when I find it. I was sure it was yours.

--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's ****ed.
Python Philosophers (replace 'nospam' with 'kennedym' when replying)
 
Reply With Quote
 
 
 
Reply

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Re: adding exif data to scanned film nospam Digital Photography 4 11-01-2009 07:54 PM
Re: adding exif data to scanned film Keith Nuttle Digital Photography 1 10-30-2009 06:24 PM
Nikon S10 - zoom range less than claimed? Marty Fremen Digital Photography 4 11-21-2007 09:13 AM
Bizarre New Invention Claimed To Induce Psychic Powers. Aurum Solis HTML 2 09-23-2006 09:49 AM
After having 8mm film reels digitally archived, film looks very grainy/ filled with static. Is this digital-looking noise normal? + more 8mm film questions Phil Edry Digital Photography 11 10-10-2004 11:57 PM



Advertisments