D300 noise and Capture NX

Started Nov 29, 2007 | Discussions thread
Arash Hazeghi Contributing Member • Posts: 967
Re: Here's a note to Phil!

Iliah Borg wrote:

Arash Hazeghi wrote:

Different sensors have different characteristics,

Agree 100%

thus each have an optimized demosaic

Demosaicing for Bayer sensor does not depend on sensor properties,
pre-demosaicing routines depend mostly on the lens, AA filter, sensor
pitch, actual geometry of the sensitive element, and well capacity.
Demosaicing can be optimized for those parameters based on spec
sheets and little experimentation. Nikon demosaicing is far from
being state of the art.

yes I was talking about whole demosaic+predemosaic+postdemosaic routines that are applied to data for each pixel location right after ADC. you are right, Bayer demosaic algorithm by itself is a well defined code and is the same for all Bayer cameras, but the pre-demosaic and even poat-demosaic (like sharpening etc.) rountines are not the same and remember no one has the sensor data sheet except for Nikon and Sony.

and NR algorithm which only the manifacturer of
the sensor knows.

Same, digital part of NR is nothing really tricky.

it's not tricky but you could use a more effective NR algorithm if you know more about the corrolation of the samples (i.e. history of the data) which of course you know better if you know what the actual sensor is. NR in Canon DPP or ACR for example create completely different images.

ACR and others use a generic demosaic algorithm and
a generic NR/Anti Moire. Thus picrures from most cameras look the

Picture look in ACR/LR is due to design decisions: methods of white
balance, saturation control, colour conversion, tone curve, fast
imprecise demosaicing and clay-producing (for the lack of better
term) NR.

-- hide signature --

no text

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow