Olympus-OM
[Top] [All Lists]

[OM] Re: Canon 9950F question

Subject: [OM] Re: Canon 9950F question
From: Moose <olymoose@xxxxxxxxxxxxxxx>
Date: Sat, 25 Jun 2005 00:00:46 -0700
Andrew Dacey wrote:

>I agree about the sharpness, that's pretty fine detail but it's nice
>if you can bring out all the detail that's in the image. One thought
>ocurred to me though that the photo-i review scanned everything at
>4000dpi, which is the max resolution for the Nikon 4000. However, the
>9950F will go up to 4800dpi. How much extra detail you'd get is
>debatable but it's possible that 4000dpi isn't a "native" resolution
>for the 9950F and so the sharpness might be suffering because of that.
>I would've liked to have seen at least 1 sample of a scan done at
>4800dpi and then resampled to 4000dpi to test if this was the case.
>  
>
Comparisons like that are really difficult. Human vision seems 
unalterably biased in favor of the larger image, so any comparison needs 
to be at the same effective ppi. Yet, as you point out, the native 
resolutions of scanners differs. But downsampling (and up) inevitably 
introduce artifacts, especially at anything other than even integer 
divisions of the native resolution. 4000 is 83.3333% of 4800. How is one 
ever to know whether the comparison is fair to both scanners?

Perhaps more to the point is the question of what one will be scanning. 
My purpose for a 9950F would be to scan all the old film I have, from 
the family album with way too many, often fuzzy, 6x9 pics of the 
firstborn, me, with a Kodak folder through those I have taken over 
decades. Test scans of some of these leads me to the conclusion that the 
vast majority of this cache is not fine grain enough for 4000 dpi to be 
of any use. I know I've read from those who test scanners and/or do a 
lot of scanning for a living that anything over 3200 dpi is wasted on 
normal film. I think that number will be alot lower for much of my older 
stuff. Grain used to be a lot larger than it is now, so easy to forget. 
I think 1200 dpi will be more than sufficient for the old 6x6 and 6x9 
B%W and color.

Also, no matter what the specs say, dpi of flim and flatbed scanners 
just aren't comparable. My guess is that the 9950F and the comparable 
performance Epsons at 4800 dpi are about equivalent in ability to 
resolve detail to a 3200 dpi film scanner.

>As for the colour, I'm assuming that the reference image at the top of
>the page has been adjusted to reflect how the image looks on a slide.
>While my monitor might then display that image differently all the
>images would be "out" by the same amount so relative comparisons can
>still be made. As well, the reviewer comments about the colour as
>well. 
>
Sure, but remember, this is a guy with a day job doing his interactive 
reviews in his one time. I think he does an astonishing job... But to 
expect him to get the best possible color results out of several pieces 
of equipment and software all at the same time is unrealistic.

>I'm not as concerned with that though as I always scan in higher
>bit-depth so that I have the leeway to adjust the colour to be
>accurate. It could also be a software or profile issue.
>  
>
Exactly.

>My former scanner was an FS2710 as well. I tried the demo of Vuescan
>and with one test I was convined that I had to buy the full version.
>I've heard of some issues with Vuescan and the 9950F so I'm a little
>put off by that but I'm running a mac and the issues I've heard of
>were on PC's so that may or may not affect me.
>  
>
I think those were all resolved some time ago. I helped Ed a bit with 
some issues with the Can*n 5000F and he is almost umbelieveably 
responsive in keeping at it until an issue is resolved.

>I'm curious as to what the process is for creating icc profiles for
>different films? I'm guessing that shooting a colour target is
>involved? 
>
Yes.

>Could you supply details? I think this was a feature that's
>been added in more recent versions of vuescan as when I started with
>it there was only a very limited selection of film profiles, which
>weren't icc profiles.
>  
>
Same here. Ed does so many revisions that I also didn't notice when it 
appeared. The process is pretty straightforward. You shoot an IT8 target 
and scan the film with Profile=>Profile film, and it generates an icc 
profile. It's all explained, in Ed's terse manner, in the Help/User's 
Guide under"Film Profiling with IT8 Targets". The source for target he 
recommends, Wolf Faust, is the best, as far as I can determine. Quick 
and easy. You pay with PayPal and he ships them right away.

>>So why don't you just buy one, do all the testing and pass the results
>>on to us? :-)
>>    
>>
>
>I may have to go that route. Obviously I can't go for the scanner
>comparisons that I'm asking for because then I'd have to own them all
>and that defeats the purpose of wanting to know which one to get. So
>if I do get one I could only compare it with the FS2710 and that would
>be on different systems with different monitors so not really a valid
>test. So I suspect any review I'd do would simply be along the
>"pleased" or "displeased" line.
>  
>
What? You don't want to spend all your time playing with scanners?

My experience with going to the FS4000 from the FS2710 is about what I 
expected in resolution. About all the higher resolution provides for 
most scans is bigger files. Certainly it can pull more out of the film, 
but not all that much in many cases. And for many uses, it doesn't make 
any real difference. Here's some comparisons I posted before:
____________________________________________________________
The bottom row of samples are:

1. 5000F using CanoScan software at 2400 dpi, full pixel sample, no 
processing.
2. 5000F using VueScan software at 2400 dpi, full pixel sample, no 
processing.
3. FS2710 using VueScan at 2700 dpi, downsampled to same size as 2400 
dpi scans. I think I sharpened this, not sure.
4. Scan #3 with grain reduction applied with NeatImage.
________________

> I scanned the same slide on the FS4000 as in the examples I posted 
> from the can*n 5000F and FS2710. There is clearly less obvious 
> and smaller graininess in the 4000 dpi scan.
>
> I have  added to http://www.moosemystic.net/Gallery/ComboA.jpg a 
> sample of the same slide scanned on the FS4000 and downsized to match 
> the 2720 dpi scan. Order of the top row is now raw 2720 dpi scan, raw 
> 4000 dpi scan downsized and 2720 dpi scan processed in NeatImage. Its 
> very interesting to me that the higher resolution actually resulted in 
> apparently smaller and less prominent grain that the lower dpi scan.  
> I'm not sure if more detail is resolved, but more is visible, at least 
> partly because it was obscured by the 'grain', and shadow detail is a 
> bit better.
>
> I seem to recall examples of higher dpi resulting in grain aliasing on 
> some fine grained film. Here is is clearly striking at lower dpi with 
> a grainier film, mid 90s AGFA 200RS.
>
Be sure to view ComboA.jpg at 100%.
______________________________________________________________

Moose



==============================================
List usage info:     http://www.zuikoholic.com
List nannies:        olympusadmin@xxxxxxxxxx
==============================================

<Prev in Thread] Current Thread [Next in Thread>
Sponsored by Tako
Impressum | Datenschutz