Re: Is the EPSON 4990 really 16 bit?

On 03 Feb 2006 18:06:05 GMT, Marjolein Katsma <nobody@xxxxxxxxxxx>
wrote:

Noise, by contrast, looks much "sharper". It's individual pixels
(rather than grain "lumps" which are clusters of pixels).

That assumes pixels are smaller than grains - but that will depend on a
number of factors (at least the resolution of the scanner in proportion
to the size of the grains).

That's why I wrote "depending on resolution, of course". Also, grain
is analog and therefore not of uniform size like pixels.

Noise will be demonstrated by occasional single pixels which are
bright red or green, etc. Radically brightening up a dark noisy scan
will show the noise much better. Now, scan again with boosted AG, say,
+2.0 or +3.0 and then look at the same area and you'll see it's
smooth, without any "random" pixels. For a better comparison, brighten
up the dark scan (the one with the noise) and you'll see the
difference quite clearly.

I think you are referring to scanning a positive (slide) here.

Yes, because that's where the noise is the easiest to spot. This is
due to slides having a much wider dynamic range than negatives. (The
image on the negative has a compressed dynamic range.)

Therefore it's more likely that a slide will have a part of the
dynamic range which the scanner will not cover adequately. Don't
believe the "theory" because according to that a scanner with about
12.5-bits of dynamic range should cover all possible dynamic range of
slides. But it doesn't. Even your 14-bit scanner with 1.5 bits of
overhead for noise still can't penetrate dark areas without noise
(anything below about 32 on the histogram after gamma is applied).

I've commented on this before - but the same "dark" area in a negative
image is actually light (a "thin" area in the negative) although the
scanner's preview will show it as dark; the scanner will also show the
same values and controls for AG but actually does the *opposite* (less
light rather than more); so boosted AG will/should lead to *more* noise
rather than less on a negative.

In addition to compressed dynamic range mentioned above, reversal is
another problem why noise is more complicated to explain with
negatives. It can be done, but it's much more straightforward to take
a dense slide and experiment with that.

Don.
.

Relevant Pages

  • Re: CMOS vs. CCD -- Link to Article
    ... dynamic range, and deliver more pixels, then it would be something. ... Phil ... data on the rate the noise grows with time for each pixel, ...
    (sci.astro.amateur)
  • Re: Is the EPSON 4990 really 16 bit?
    ... Grain, or 'averaged' JPG artifacts? ... pixels. ... Noise, by contrast, looks much "sharper". ... at a dark area of an image which is, say, dirty and dark black/brown. ...
    (comp.periphs.scanners)
  • Re: Is the EPSON 4990 really 16 bit?
    ... (rather than grain "lumps" which are clusters of pixels). ... I agree though that what "grain" I've seen so far doesn't look like ... etc. Radically brightening up a dark noisy scan ... will show the noise much better. ...
    (comp.periphs.scanners)
  • Re: Sonys new prosumer: Canon going to cringe?
    ... If you had a sensor that had a billion pixels, and one that had one million pixels, both the same size, and they had the same quantum efficiency, the gigapixel sensor would not have any more shot noise, and could get away with 31.6x as much read noise per pixel, without increasing image read noise, all the while with 31.6x the potential resolution. ... try tracking dynamic range too ...
    (rec.photo.digital.slr-systems)
  • Re: And again, digital vs (LF) film...
    ... exposition of digital is 20x smaller than of film;-). ... Dynamic Range and Transfer Functions of Digital Images ... spectrum of noise). ... So let's say you average all pixels in a 10mm diameter ...
    (rec.photo.digital)
Loading