Olympus-OM
[Top] [All Lists]

[OM] Re: Some new PEN FT photos

Subject: [OM] Re: Some new PEN FT photos
From: Moose <olymoose@xxxxxxxxxxxxxxx>
Date: Wed, 17 Nov 2004 12:16:43 -0800
Dean Tyler wrote:

>This may be a silly question, but I have an Espson 2450 and it is soft.  The
>4870 seems soft also.  The review of the new Can*n flatbed said it is soft
>compared to the 4000.  The question is, if all the scanners require heavy
>sharpening, what does the increased dpi really get you?
>
If you can get your head around the idea that sharpening is a part of 
the scanning process with virtually all scanners, and especially with 
all flatbed scanners, then the question partially resolves itself. I 
don't yet know about the 4000-5400 dpi film scanners, but some LCE and 
sharpening improves almost all images from the 2710. The software that 
comes with the scanners has sharpening available, but most reviewers 
show comparisons both with and without scanning so that the raw hardware 
performance can be compared and end up concluding that the sharpening in 
good image editors is better and more flexible than the sharpening in 
the scanner software.

So, what more dpi in a flatbed gives you first is more detail captured 
from the film. This is true whether both scans are compared unsharpened 
or both sharpened. Depending on the individual image and details of 
relative dpis and print size, a scan from a higher dpi scanner will 
sometimes look sharper even when downsized to equal dpi for printing.

When comparing flatbeds to film scanners, the dpi, which is a hardware 
spec about the sampling rate, rather than a real resolution spec, is 
simply not comparable. It appears that the new high end 4800 dpi 
flatbeds produce 35mm film images that, after sharpening, are about 
equivalent to a 2800 dpi film scanner, also properly sharpened. At a 
guess, the good 2400 dpi flatbeds produce something like the resolution 
of a 1200 to 1600 dpi film scanner. Those who make comparisons differ 
somewhat.

>I understand the increase in Dmax is most likely important with the new 
>scanners.
>
That's certainly one thing that has improved in flatbeds, but still 
doesn't quite reach the best film scanners.

One big thing that flatbeds offer is the ability to put some, to quite a 
few, depending on model, slides or negs on the platen at once and scan 
them without having to mess about with each frame. Even if it all takes 
some time, it is automated and one can be doing something else while the 
process goes on. The 4870 can do 8 slides, 4 strips of up to 6 35mm 
exposures each, various multiples of MF frames or 2 4x5 images at once. 
The 9950F is similar in that regard, but adds another stip of 35mm, for 
up to 30 images in one process. While not up to the quantity, speed and 
quality of the big Nikons with automated feeders, the are also much cheaper.

>It seems to me that you can get a high quality flatbed and you have to sharpen 
>a lot or a high quality film scanner that really brings out the grain.
>
You make it sound like the high res scanners are at fault for the grain 
of the film they scan. Norman Koren believes that many of the early 
complaints about the FS 4000 were really about its excellent 
sharpness/resolution, which accurately captured grain that was there all 
along, but not prominent in earlier prints from less accurate scans 
and/or lower resolution optical prints.

That's part of the reason I put an image with grain reduction applied in 
the samples I posted. Even at 2700 dpi, fast, grainy film is going to 
look grainy, simply because it IS grainy. It seems to me that that is 
down to me, for my film choice, not the scanner, for accurately 
reproducing what is there. In the end, the film scanner image with grain 
reduction applied is pretty good. I think it would make a technically 
nice 11x14 print at 240 dpi, as the whole image is 3660x2463.

>  When we discuss these scanners are we really just discussing being able to 
> print larger?
>
No, we are talking about actually capturing more detail from the film. 
Look at the hair and the molding detail behind the head in my sample. 
There is simply more detail resolved in the film scan, even after 
intentionally rather strong grain reduction, than in the flatbed scans.

>  With 4000 dpi you can get a good 11x14 equal to an 8x10 on a 2800 dpi.  It 
> this too simplistic?
>
Yes. You are assuming that dpi is a bigger limitation to print size than 
it actually is. If you don't need any more detail, you can upsample a 
smaller scan. This is what your printer driver is doing anyway when you 
print say a 300dpi image on a 1440 dpi printer. Do you imagine that high 
quality, high res printers are simply printing at the low res of a low 
dpi input?You may also get lower sized images up to the dpi you like for 
printing using stair interpolation in an image editor with remarkable 
results. Harald Johnson's tests show that it doesn't seem to matter 
whether you do the upsampling before or let the printer driver do it. 
Since the printer driver is optimized for the specific characteristics 
of the hardware it is driving, why not just let it do its job?

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