Sigma/Foveon Questions
in [Digital Cameras]
|
Prev: SI Facescape
Next: FF camera with mirrorless design
From: nospam on 26 Apr 2010 16:11 In article <YAdBn.167851$0N3.24456(a)newsfe09.iad>, Martin Brown <|||newspam|||@nezumi.demon.co.uk> wrote: > A monochrome pixel is a pixel, but only if the image can be interpretted > as an image in its own right. A Bayer mask image looks pretty strange if > displayed as colours without demosaicing. If you split a 12Mpixel Bayer > mask image into the four separate colours of 3Mpixel each then you can > call them pixels in the various monochrome representations of colour > images. When they are still all mixed up it is misleading to call them > pixels because that has an implication of all being channels measured. technically they aren't pixels on the sensor but very few people make the distinction. the number is the same either way. > By any reasonable definition a photosite on Bayer sensor array is only > measuring one colour out of R,G,B or possibly two in the case of low > light modified C,M,Y,G filters. That is necessarily losing information > about the true luminance of the image. a little, but not as much as some people think. > There are 12M sensor sites, but each one is measuring only a part of the > image data. In particular there is a difference in the spatial > resolution of the red and blue sensors compared to green. they're not taken independently of the green. > And of all the > colours when compared to an unfiltered sensor array of the same > dimensions. Demosaicing can fix most of this up which is why the Bayer > pattern is useed. It works because the human eye is ropey at colour > resolution. yes, and it works well. > In common computer imaging usage a pixel is generally taken to mean a > monochrome image of 8, 16 or 32 bits, or a colour pixel with either > palletted 8 bits, 16bits (R,G,B = 5,6,5), 24bits (8bits for each of > R,G,B), 48bits (16 bits each for R,G,B). or cmyk or rgba or numerous other variants. > The Bayer mosaic is none of these as there are a lot of implicit zeroes > in the measurement grid and a distinct green bias due to the twofold > excess of green pixels in the grid. it's a single spatial element of an image, whether it's on the sensor or in the image.
From: nospam on 26 Apr 2010 16:12 In article <hr3qdv$jcl$1(a)news.eternal-september.org>, David J Taylor <david-taylor(a)blueyonder.co.uk.invalid> wrote: > No, I'm describing what happens in the typical 12MP DSLR, which has 3 > million red-sensitive photo-sites, and delivers 12 million full-colour > output pixels in its JPEG file. it doesn't upsize the red and blue. each pixel is processed with its neighbors.
From: nospam on 26 Apr 2010 22:19 In article <7v6BpuB8vh1LFwKH(a)kennedym.demon.co.uk>, Kennedy McEwen <rkm(a)nospam.demon.co.uk> wrote: > We start with 4Mp of red and blue, we end up with 12Mp of red and blue. > Irrespective of what else is used in the interpolation that is *STILL* > interpolation and upsizing. interpolation yes. no upsizing since it's already the correct size.
From: Ray Fischer on 27 Apr 2010 01:47 Martin Brown <|||newspam|||@nezumi.demon.co.uk> wrote: >Ray Fischer wrote: >> Kennedy McEwen <rkm(a)kennedym.demon.co.uk> wrote: >>> Ray Fischer >>>> David J Taylor <david-taylor(a)blueyonder.co.uk> wrote: >>>>> "Ray Fischer" <rfischer(a)sonic.net> wrote in message >>>>> news:4bd3ec93$0$1610$742ec2ed(a)news.sonic.net... >>>>>> David J Taylor <david-taylor(a)blueyonder.co.uk> wrote: >>>>>>>> however, the total number doesn't change. there are 12 million on the >>>>>>>> sensor and 12 million in the image, or however many the sensor has. >>>>>>> There are 12 million monochrome pixels on the sensor, >>>>>> No, there are 4.6 million pixels. Any other claim is a lie. >>>>>> >>>>>>> interpolated to 12 >>>>>>> million colour pixels. The sensor only has 3 million red pixels, but >>>>>> Learn what "pixel" means. > >I suggest that you learn what a pixel means - that would help. You're too much of a stupid ashsole to be condescending. >The Bayer grid contains filtered photosensor sites. It takes the data >from several of these to construct any pixel in the final image. An outright lie. You're "confused" becuse you think that a pixel must have a certain amount of color information. In fat it need contain no color information at all. >>>>> This part of the thread had evolved to being about Bayer in a hypothetical >>>>> 12MP DSLR, and whether or not spatial interpolation was involved. >>>> A pixel is a picture element. You cannot split up the color >>>> componenets of a pixel in some arbitrary way and then claim that a >>>> single pixel is really three, four, or a thousand pixels. >>>> >>> So a Bayer CFA sensel, being an incomplete picture element, >> >> WRONG. >> >> It's not incomplete. There's no such thing as an "incomplete" pixel. >> A monochrome pixel is still a pixel. > >IA monochrome pixel is a pixel, but only if the image can be interpretted Wrong. No "if". > A Bayer mask image looks pretty strange if Non sequitur. The definition of pixel has nothing to do with any sensor type. >By any reasonable definition You're not reasonable. >There are 12M sensor sites, Irrelevant. >In common computer imaging usage a pixel is generally taken to mean a >monochrome image of 8, 16 or 32 bits, or a colour pixel with either >palletted 8 bits, 16bits (R,G,B = 5,6,5), 24bits (8bits for each of >R,G,B), 48bits (16 bits each for R,G,B). LOL! Tell us: Where did you get your degree in computer science? Where did you get your education in computer graphics? I got mine from Stanford and Cal Poly and from working in the graphics business. -- Ray Fischer rfischer(a)sonic.net
From: Ray Fischer on 27 Apr 2010 01:50
Kennedy McEwen <rkm(a)kennedym.demon.co.uk> wrote: >In article <4bd4da0d$0$1602$742ec2ed(a)news.sonic.net>, Ray Fischer ><rfischer(a)sonic.net> writes >>Kennedy McEwen <rkm(a)kennedym.demon.co.uk> wrote: >>> Ray Fischer >>>>David J Taylor <david-taylor(a)blueyonder.co.uk> wrote: >>>>>"Ray Fischer" <rfischer(a)sonic.net> wrote in message >>>>>news:4bd3ec93$0$1610$742ec2ed(a)news.sonic.net... >>>>>> David J Taylor <david-taylor(a)blueyonder.co.uk> wrote: >>>>>>>> however, the total number doesn't change. there are 12 million on the >>>>>>>> sensor and 12 million in the image, or however many the sensor has. >>>>>>> >>>>>>>There are 12 million monochrome pixels on the sensor, >>>>>> >>>>>> No, there are 4.6 million pixels. Any other claim is a lie. >>>>>> >>>>>>> interpolated to 12 >>>>>>>million colour pixels. The sensor only has 3 million red pixels, but >>>>>> >>>>>> Learn what "pixel" means. >>>>> >>>>>Ray, >>>>> >>>>>This part of the thread had evolved to being about Bayer in a hypothetical >>>>>12MP DSLR, and whether or not spatial interpolation was involved. >>>> >>>>A pixel is a picture element. You cannot split up the color >>>>componenets of a pixel in some arbitrary way and then claim that a >>>>single pixel is really three, four, or a thousand pixels. >>>> >>>So a Bayer CFA sensel, being an incomplete picture element, >> >>WRONG. >> >>It's not incomplete. There's no such thing as an "incomplete" pixel. >>A monochrome pixel is still a pixel. >> >A monochrome pixel is a pixel in a monochrome image. >For those lagging at the back with Ray, we are discussing colour images, >not monochrome images. You are pushing your religion. I'm trying to educate you. >As such a monochrome sensel is NOT a complete pixel, A sensel isn't ANY kind of pixel. >It is a complete pixel All pixels are "complete". Your claim is as absurd as claiming that 13 isn't a "complete" number because it's missing some factors. -- Ray Fischer rfischer(a)sonic.net |