Prev: Nikon Coolscan V Refurbished
Next: Epson v750 for print scanning
From: Barry Watzman on 14 Jan 2008 00:14
If you have pixels spaced 1/4,800th of an inch apart (that's what 4800
dpi means), you can use every pixel, or every other pixel (every other
pixel giving only 2,400 dpi). There are no resolution choices in
between 2,400 dpi and 4,800 dpi that don't use interpolated pixels
("calculated" pixels where no actual image data exists). Below 2,400
dpi, your next choice not using interpolated pixels would be every 3rd
pixel (1,600 dpi).

Bits does not have anything to do with resolution but rather color
depth. The human eye can see (depending on which source you cite)
somewhere between 1 million and 10 million colors. Well, 24-bit color
depth is 16 million colors. More can be helpful in certain situations
involving slides with really dark areas that are still discernable as
not being solid black, but the benefit is marginal and usually of
interest only to professionals [also note that the JPEG format is only
capable of saving files in 24-bit color].

Can't help you with the glass mounts. clearly reflections from the
glass are possible and may distort your results.


Peter D wrote:
> Thank you for your helpful information. I decided I wouldn't use any
> interpolated resultions, staying with optical resolution only. From what
> I've read 2700 x 2700 seems to be the 'sweet spot' for 35mm. I didn't
> realise that I have to select an optical resolution that has to be divide
> into the highest optical resolution. I would have liked to go with 2700 or
> 3000. Are you certain I have to go with 2400 if I don't want to go as high
> as 4800? 2400 is fine for these slides, but I would prefer to go to 2700 or
> 3000, but if that would actually be worse or no better than 2400 then I
> don't see the point (longer scan time, larger file size).
>
> How about the option of using 96 bit? Would that really be better than 48
> bit or is overkill?
>
> Also, any tips on scanning slides in glass mounts?
>
> Thanks for all your help.
>
> "Barry Watzman" <WatzmanNOSPAM(a)neo.rr.com> wrote in message
> news:478991db$0$5194$4c368faf(a)roadrunner.com...
>> 4800 x 9600 implies that the resolution in one direction (vertical or
>> horizontal) is 4800 dpi and the resolution in the other direction is 9600
>> dpi. Optical means that this is the "true", "actual" resolution ... you
>> get a "real" pixel (a point actually imaged by the sensor) at those
>> resolutions in those directions. On the contrary, "interpolated"
>> resolution is resoulution where there are pixels between "real" pixels
>> that are "fabricated" by interpolating between the nearest "real" pixels.
>>
>> The resolution in one direction is determined by the actual sensor element
>> (CCD, CIS element, etc.). The resolution in the other direction is
>> determined by the stepper motor that moves the scan head across the
>> material being scanned ... how fine it's steps are (how many points the
>> head stops at to image).
>>
>> 4800 x 4800 is almost too high ... that is going to give you a scanned
>> image (for 35mm images) of almost 32 megapixels. I'm going to go out on a
>> limb and say that there isn't that much detail present in ANY 35mm images.
>> 2700 dpi give 10 megapixels, which is a good value. In SOME images, a
>> SLIGHTLY higher value might be better, but for most images, 10 megapixels
>> (2700 dpi) is going to pretty much capture everything. The problem you
>> have is that with a hardware resolution of 4800 x 4800, the next lower
>> choice that you have that uses only optical resolution is going to be 2400
>> x 2400 (e.g. use every other pixel), and that is getting to be too low (it
>> will give about an 8 megapixel image ... nothing to sneeze at, but more
>> would probably be better)).
>>
>>
>>
>> Peter D wrote:
>>> I decided to go with the HP G4050. Mostly because of price ($180),
>>> convenience, and the ability to do 16 slides or 30 negatives at a time
>>> without any additional equipment. That's enough to make it worth loading
>>> and walking away (semi-automatic). I'll play with it once I get it, and
>>> pass on what I learn. As I go through, I'm going to single out
>>> "spectacular" slides/negs and send them away for professional scanning
>>> (89c each).
>>>
>>> I wonder if anyone can advice me on the following:
>>>
>>> 1) Scanning slides in glass mounts - do I need to scan differently or do
>>> I treat just like the other slides?
>>>
>>> 2) Best scanning resolution and colour depth - I can go to 96 bit 4800
>>> x4800 optical ( HP says "4800 x 9600 Hardware Scanning Resolution" -
>>> whatever that means).
>
>
From: Barry Watzman on 16 Jan 2008 23:33
I think you are incorrect in your statement about flatbed scanners. I
think that the true optical resolution of most of the current and recent
past crops of flatbed scanners are in the range of 2,400 to 4,800 dpi.
My HP 5470C from 2002 has a maximum true optical resolution of 2,400 dpi.


Kennedy McEwen wrote:
> In article <478aefae$0$4985$4c368faf(a)roadrunner.com>, Barry Watzman
> <WatzmanNOSPAM(a)neo.rr.com> writes
>> If you have pixels spaced 1/4,800th of an inch apart (that's what 4800
>> dpi means), you can use every pixel, or every other pixel (every other
>> pixel giving only 2,400 dpi). There are no resolution choices in
>> between 2,400 dpi and 4,800 dpi that don't use interpolated pixels
>> ("calculated" pixels where no actual image data exists). Below 2,400
>> dpi, your next choice not using interpolated pixels would be every 3rd
>> pixel (1,600 dpi).
>>
> Since none of the flatbeds mentioned in this thread so far have a TRUE
> optical resolution (as opposed to an optical sampling density, which
> says nothing about true resolution) better than 2000ppi, and most are
> abysmally lower, it doesn't matter one iota what resolution you use
> between 2400 and 4800ppi. "No actual image data exists" above 2000ppi
> for any of them. They are all interpolated - its just a question of
> whether they are analogue interpolated or digitally interpolated.
>
> Stepping 1/4800th of an inch between samples doesn't give any additional
> information if the optics themselves can't resolve more than 1/2000th of
> an inch! Its just analogue interpolation.
>
> So use whatever resolution you like between 2400 and 4800ppi - you
> aren't going to get any more detail off the film since you have already
> exceeded the true resolution of any flatbed scanner.
From: Peter D on 17 Jan 2008 10:28
Unless I assume that HP is faking the results or playing clever games with
the facts, the optical res of the G4050 is at least 4800x4800. The HP site
at lists the resolutions as follows:
Optical Scanning Resolution - Up to 4800 dpi
Hardware Scanning Resolution* - Up to 4800 x 9600 dpi
enhanced scanning resolution* - Unlimited

*I don't know what "Hardware Scanning Resolution" is, but I think "enhanced
scanning resolution" is another name for "interpolated" and thus "Hardware
Scanning Resolution" isn't. I presume that the stated Optical Scanning
Resolution is true unless someone can show me hard facts otherwise.

Test slides at 4800 x4800 96 bit are incredibly detailed, far beyond what I
expected, but they do take a long time -- aobut 40-50 minutes to do 16 at a
time.

As for my situation, I've decided to scan the slides at 2400, 48 bit color
depth and save them in a non-'lossy' format because I can store them
compressed as "rar" (WinRAR) and use "par" files to protect them against
loss due to damage. I'll then create jpg high quality but smaller 300 dpi
versions for ordinary printing and viewing use. If anyone thinks I should do
something different please tell me.

"Barry Watzman" <WatzmanNOSPAM(a)neo.rr.com> wrote in message
news:478edaaf$0$4928$4c368faf(a)roadrunner.com...
>I think you are incorrect in your statement about flatbed scanners. I
>think that the true optical resolution of most of the current and recent
>past crops of flatbed scanners are in the range of 2,400 to 4,800 dpi. My
>HP 5470C from 2002 has a maximum true optical resolution of 2,400 dpi.
>
>
> Kennedy McEwen wrote:
>> In article <478aefae$0$4985$4c368faf(a)roadrunner.com>, Barry Watzman
>> <WatzmanNOSPAM(a)neo.rr.com> writes
>>> If you have pixels spaced 1/4,800th of an inch apart (that's what 4800
>>> dpi means), you can use every pixel, or every other pixel (every other
>>> pixel giving only 2,400 dpi). There are no resolution choices in
>>> between 2,400 dpi and 4,800 dpi that don't use interpolated pixels
>>> ("calculated" pixels where no actual image data exists). Below 2,400
>>> dpi, your next choice not using interpolated pixels would be every 3rd
>>> pixel (1,600 dpi).
>>>
>> Since none of the flatbeds mentioned in this thread so far have a TRUE
>> optical resolution (as opposed to an optical sampling density, which says
>> nothing about true resolution) better than 2000ppi, and most are
>> abysmally lower, it doesn't matter one iota what resolution you use
>> between 2400 and 4800ppi. "No actual image data exists" above 2000ppi
>> for any of them. They are all interpolated - its just a question of
>> whether they are analogue interpolated or digitally interpolated.
>>
>> Stepping 1/4800th of an inch between samples doesn't give any additional
>> information if the optics themselves can't resolve more than 1/2000th of
>> an inch! Its just analogue interpolation.
>>
>> So use whatever resolution you like between 2400 and 4800ppi - you aren't
>> going to get any more detail off the film since you have already exceeded
>> the true resolution of any flatbed scanner.


From: Peter D on 17 Jan 2008 10:31
I'm soorry. I forgot the URL to the HP G4050 scanner ->
http://tinyurl.com/2kjlot

"Peter D" <please@.sk> wrote in message
news:13out1rm6kq6cbb(a)corp.supernews.com...
> Unless I assume that HP is faking the results or playing clever games with
> the facts, the optical res of the G4050 is at least 4800x4800. The HP site
> at lists the resolutions as follows:
> Optical Scanning Resolution - Up to 4800 dpi
> Hardware Scanning Resolution* - Up to 4800 x 9600 dpi
> enhanced scanning resolution* - Unlimited
>
> *I don't know what "Hardware Scanning Resolution" is, but I think
> "enhanced scanning resolution" is another name for "interpolated" and thus
> "Hardware Scanning Resolution" isn't. I presume that the stated Optical
> Scanning Resolution is true unless someone can show me hard facts
> otherwise.
>
> Test slides at 4800 x4800 96 bit are incredibly detailed, far beyond what
> I expected, but they do take a long time -- aobut 40-50 minutes to do 16
> at a time.
>
> As for my situation, I've decided to scan the slides at 2400, 48 bit color
> depth and save them in a non-'lossy' format because I can store them
> compressed as "rar" (WinRAR) and use "par" files to protect them against
> loss due to damage. I'll then create jpg high quality but smaller 300 dpi
> versions for ordinary printing and viewing use. If anyone thinks I should
> do something different please tell me.
>
> "Barry Watzman" <WatzmanNOSPAM(a)neo.rr.com> wrote in message
> news:478edaaf$0$4928$4c368faf(a)roadrunner.com...
>>I think you are incorrect in your statement about flatbed scanners. I
>>think that the true optical resolution of most of the current and recent
>>past crops of flatbed scanners are in the range of 2,400 to 4,800 dpi. My
>>HP 5470C from 2002 has a maximum true optical resolution of 2,400 dpi.
>>
>>
>> Kennedy McEwen wrote:
>>> In article <478aefae$0$4985$4c368faf(a)roadrunner.com>, Barry Watzman
>>> <WatzmanNOSPAM(a)neo.rr.com> writes
>>>> If you have pixels spaced 1/4,800th of an inch apart (that's what 4800
>>>> dpi means), you can use every pixel, or every other pixel (every other
>>>> pixel giving only 2,400 dpi). There are no resolution choices in
>>>> between 2,400 dpi and 4,800 dpi that don't use interpolated pixels
>>>> ("calculated" pixels where no actual image data exists). Below 2,400
>>>> dpi, your next choice not using interpolated pixels would be every 3rd
>>>> pixel (1,600 dpi).
>>>>
>>> Since none of the flatbeds mentioned in this thread so far have a TRUE
>>> optical resolution (as opposed to an optical sampling density, which
>>> says nothing about true resolution) better than 2000ppi, and most are
>>> abysmally lower, it doesn't matter one iota what resolution you use
>>> between 2400 and 4800ppi. "No actual image data exists" above 2000ppi
>>> for any of them. They are all interpolated - its just a question of
>>> whether they are analogue interpolated or digitally interpolated.
>>>
>>> Stepping 1/4800th of an inch between samples doesn't give any additional
>>> information if the optics themselves can't resolve more than 1/2000th of
>>> an inch! Its just analogue interpolation.
>>>
>>> So use whatever resolution you like between 2400 and 4800ppi - you
>>> aren't going to get any more detail off the film since you have already
>>> exceeded the true resolution of any flatbed scanner.
>
>


From: Raphael Bustin on 17 Jan 2008 21:07
On Thu, 17 Jan 2008 09:28:58 -0600, "Peter D" <please@.sk> wrote:


>Test slides at 4800 x4800 96 bit are incredibly detailed, far beyond what I
>expected, but they do take a long time -- aobut 40-50 minutes to do 16 at a
>time.

Here's a very simple test to see if that 4800 x 4800 is "real."

Downsample a copy of the image to half that resolution,
ie. 2400 x 2400.

Then upsample back up to 4800 x 4800, and compare to
the original.

If there's no difference (which I'm fairly certain will be the case)
then there wasn't any information in the scan that couldn't be
captured at the lower resolution. Meaning that you were
carrying around 4x more pixels than were justified.

At some point (in the fist downsampling step) as you lower
that target resolution, you *will* hit a point where real detail
is lost. That is your effective resolution. Anything much
above that is wasted pixels -- data with no information.


rafe b
www.terrapinphoto.com
First  |  Prev  |  Next  |  Last
Pages: 1 2 3 4
Prev: Nikon Coolscan V Refurbished
Next: Epson v750 for print scanning