On 5 Jun 2003 at 13:07, T Rittenhouse wrote: > Very interesting... > > Now, when they say a digicam is 6 megapixel, do they mean 6 million > blue-green-green-red pixels (24 million sensor pixels), or do they mean 1.5 > million blue-green-green-red pixels (6 million sensor pixels)?
Indeed it is interesting.. My E-10 is spec'd as a 4M pixel camera. The raw files consist of 2256x1684 pixels laid out in the RGBG bayer pattern as you saw on my page. The processed files are 2240x1680 pixels as processed by the camera or 2248x1676 as processed by an external application (varying amounts of edge pixels have to be discarded dependant upon the demosaicing algorithm utilized). > Now, on the Foveon, when they say 3 megapixels, do they mean 3 million, > equivalent, blue-green-red pixels, or 1 million, equivalent, blue-green-red > pixels? And, what about the reduced green sensitivity of the Foveon sensor > compared to a Bayer sensor? As Herb mentioned, each pixel location features three colour sensors unlike the fixed matrix of single primary colour sensors on a bayer sensor. I do have problems with the hype surrounding the "human green sensitivity" given that in any case regardless of how much original information is sampled for each colour the interpolated image has a finite colour depth for each RGB component in the composite image. I've seen no proof that the green component is any more linear. Obviously though the majority of lumimance component is derived from the green channel given that there are 2 pixels for ever one of green and blue. > Sometimes I think digital folks are all bankers at heart (If we went back on the > gold standard would the dollar be worthe 1/350 of an ounce, or 1/7500 of an > ounce of gold?). If you wan to delve into marketing speak and tweaking the truth look into the controversy surrounding Fujifilms claims regarding it's proprietary SuperCCD sensor technology :-( http://home.fujifilm.com/products/digital/sccd/faq.html BTW Did you have a glance over the pdf that I provided the link to at the bottom of the page? If not take a peek, it's pretty interesting stuff and quite revealing in that there is no one best demosaicing algorithm. http://www4.ncsu.edu:8030/~rramana/Research/demosaicking-JEI-02.pdf Cheers, Rob Studdert HURSTVILLE AUSTRALIA Tel +61-2-9554-4110 UTC(GMT) +10 Hours [EMAIL PROTECTED] http://members.ozemail.com.au/~distudio/publications/ Pentax user since 1986, PDMLer since 1998
