Filmscanners mailing list archive (email@example.com)
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[filmscanners] Re: 12 vs 16-bit A/D conversion. Was: WinXP andDimageScan Dual II and more
Austin Franklin wrote:
> Hi Arthur,
>>The main area where the extra A/D bit depth is helpful is in defining
>>dark areas as image versus noise.
> Possibly, but that depends on many other factors. The CCD has noise, and
> the analog front end has noise. Only so many bits are actually useful, and
> because of that, it's highly unlikely in this case that there really are 16
> bits of noise free data.
I would say its nearly impossible that there are 16 bits of noisefree
data. Not at the price that scanner is selling for.
>>I have not compared the two, but a
>>review I read claimed considerable benefit the extra bits the A/D
> I would have to see the review, but more than likely the "considerable
> benefit" had nothing to do with the "extra bits". You just can't add bits
> and get better results, you have to have a better CCD and/or analog front
> end that will allow these extra bits to be data that is not simply noise.
No argument. Just because A/D converters with wider depth have become
cheap enough to put them into even a low end scanner doesn't guarantee
anything. Hopefully, there's more to it than just that change, or, as
you imply, its just specsmanship and a way to sell more scanners. I
don't have the URL for the review handy. I read it about a month ago.
> This is one of the issues that was discussed a while ago, is that
> manufacturers can make a claim, like number of bits of the A/D, with
> absolutely no substantiation as to whether it is meaningful or not...and
> most customers have no idea that more does not mean better...and simply
> believe that more bits must be better, heck, it's more.
I agree again. The proof is in the scan. Otherwise the numbers are
meaningless. I've been advocating truth in scanner specs for years now.
>>It is probably the principal weakness of the Dual
>>Scan II, particularly the blue channel. It is, in fact, that extra A/D
>>bit conversion that should, in theory, help fix the problem you are
>>claiming, that the dark areas are ill defined from the noise threshold.
> The way to tell how many bits are useful (to at least a usable degree) is to
> check the CCD spec and see what it has for a saturation point and a noise
> floor, and calculate the dynamic range of that.
> The Scan Dual III uses a three line (tri-linear) 2700 pixel/line sensor,
> according to their spec. Kodak doesn't make a sensor in that configuration,
> so I'm not sure whose sensor they use. If you find out what CCD these
> scanners use, let me know, and I'll look up the specs.
It would be interesting. Avision makes the Minolta scanners, as I
understand it. I suspect the CCD is one of the same ones other
companies refer to as 2700 ppi, but Minolta calls it 2820 ppi, because
the 35mm frame is slightly under one inch.
Unsubscribe by mail to firstname.lastname@example.org, with 'unsubscribe
or 'unsubscribe filmscanners_digest' (as appropriate) in the message title or