Hello all,
not related to my previous post, but when I scan at 6400dpi with the V500, I
get very strong banding: http://www.gdargaud.net/BugFest/Epso...sk-Banding.jpg
Can anyone comment on it ? Is this normal ?!?
I hope I can improve it, but doing multiple passes don't improve it much.
--
Guillaume Dargaud http://www.gdargaud.net/
If that is a print, and it looks like it is, that is a RIDICULOUS
resolution. There is no reason to ever scan a print at more than about
600 dpi. The higher resolutions are for FILM, not prints. And when
using film, stop increasing the resolution when you can see the film
grain (normally this is somewhat in the range of 2,400 to 4,200 dpi).
NOTE: A 35mm film scanned at 2,700 dpi yields a 10 megapixel image.
That is more information that most 35mm images actually contain, and
with most films will begin to capture the film grain. The resolution
capabilities of most film scanners are ridiculous .... they are FAR
higher than the information storage capacity of the film being scanned,
at which point they not only provide no further benefit, they actually
begin to degrade the result (or, at least, make it FAR larger).
Guillaume Dargaud wrote:
> Hello all,
> not related to my previous post, but when I scan at 6400dpi with the V500, I
> get very strong banding:
> http://www.gdargaud.net/BugFest/Epso...sk-Banding.jpg
> Can anyone comment on it ? Is this normal ?!?
> I hope I can improve it, but doing multiple passes don't improve it much.
No. It's a Kodak disk negative.
6400dpi may be a bit too high for that purpose, but scanning at 3200 yields
only a small 1050x1400 scan. You can actually see the borders of the
negatives on the sample image I provided. And I still do get strong banding
at 3200dpi or by downsampling a 6400 scan.
I bought the scanner last week and I did use it first to scan large format
B&W. Worked great for that purpose.
**** Kodak to to 5th circle of hell for having ruined generations of family
images with their stupid disk format.
--
Guillaume Dargaud http://www.gdargaud.net/
I think I got it: in order to scan those disks without white/black point
issues, I've printed a neutral gray on a transparent A4 sheet, with 3 holes
left through for the disks. The scanner must do some calibration on the edge
of the image, because as soon as I remove the sheet, the problem goes away.
Now I only have to figure out how to get a proper exposure with 40% of the
scan completely black and 40% completely white...
And why does the software resample down my scans ?!?
--
Guillaume Dargaud http://www.gdargaud.net/
> If that is a print, and it looks like it is, that is a RIDICULOUS
> resolution. There is no reason to ever scan a print at more than about
> 600 dpi. The higher resolutions are for FILM, not prints. And when
> using film, stop increasing the resolution when you can see the film
> grain (normally this is somewhat in the range of 2,400 to 4,200 dpi).
>
> NOTE: A 35mm film scanned at 2,700 dpi yields a 10 megapixel image.
> That is more information that most 35mm images actually contain, and
> with most films will begin to capture the film grain. The resolution
> capabilities of most film scanners are ridiculous .... they are FAR
> higher than the information storage capacity of the film being scanned,
> at which point they not only provide no further benefit, they actually
> begin to degrade the result (or, at least, make it FAR larger).
Actually when scanning film it's not always possible to tell when what
appears to be film grain is what it really appears to be - there is a known
problem with resolution interacting with the grain to make it worse than it
really is, when that happens increasing the resolution can actually make
the film look less grainy (due to aliasing effects similar to placing fine
netting over a TV or monitor screen).
I've noticed the effect myself where scanning some film looks grainier on
low resolution preview scans than it does on the final 2700dpi scan. On my
nikon scanner I've also noticed grain reduce when doing multipass scans,
presumably due to microscopic differences in scan area between scans acting
as subpixel filtering.
I know you have several film scanners so it's possible for you to prove or
disprove this on your own kit by scanning the same film at different
resolutions - easier done with 400ASA or greater speed film as grain would
be larger.
There is also a potential problem of using modern high resolution scanners
at a lower resolution than the CCD/CMOS sensor - if a 7200DPI scanner is
used to make a 2700dpi scan then it will be scanning without using all the
elements of the sensor and stepping the motor over areas so creating
unscanned holes in the image (Digital cameras with 3 CCD's can produce
better pictures than normal mosaic filtered sensor cameras for the same
reason).
Because of this Scanning the same image at full resolution with a 2700dpi
scanner (or scanning 7200dpi on the high res scanner then scaling the image
to 2700dpi) can produce better results. Then if you have to scan lower than
sensor resolution it's better to use resolution thats the full res divided
by a whole number (e.g 1/2, 1/3 or 1/4 of the sensor resolution is easier
on the hardware than an odd division - assuming the driver lets you select
anything other than even divisions).
Banding with Epson V500 
Linear Mode
