Re: [PanoToolsNG] New Pentax K10
- On Tue, 05 Dec 2006 01:32:03 +0900, Roger Howard
> On Sun, December 3, 2006 9:14 pm, Roger D. Williams wrote:Thanks, Roger, that's very clear and makes perfect sense. I think I've got
>> And is that 12-bit
>> for RAW really right? Or a mistype? Excuse my ignorance. I assumed that
>> because you could create 16-bit TIFF files developed straight from RAW
>> images, the latter must themselves be 16-bit. I see that this isn't
>> necessarily so...
> I haven't heard of any sensors short of a few on the MF/LF digital backs
> that can capture true 16bits per pixel of luminosity. But using a 22bit
> pipeline just means less accumulated quantization error within the
> internal processing.
> Think of it this way - if you were stuck to only using integer math, many
> calculations would be rounded and would accumulate error quickly. By
> allowing a few extra decimal places, you can work in a higher precision,
> and then only round back down at the end, minimizing quantization
> (rounding) error.
> We process our RAW files into 16bit simply because most imaging software
> doesn't allow working in 10, 11, or 12 bit modes (for instance) - so your
> 12bit data is padded out to 16bits.
it straight, now. Fuji's unique sensor appears to fall between the standard
12-bit and the MF/LF backs.