It would seem that some industry standard definition of "pixel" might help.
IF a pixel is a discrete location on a 2D surface, then the Foveon only has 3.4MP, but they are 3.4 TRUE, full color pixels.
A 6 "MP" bayer has 6 million discrete locations on a 2D plane, but they aren't TRUE pixels, not full color. Detail and color information is not fully recorded at each site.
Perhaps it's time for an industry body to step in and regulate the marketting and definition of pixels.
Could there be a fair distinction between bayer pixels and X3 pixels -- "bPixels" (for bicubic pixels ? ) and 'full color" or "full spectrum" pixels for X3 ???
Then a Foveon would be 3.4MP or 10 million bPixels.
and a bayer would be 6 Million bPixels or somewhere between 2.5-3.5 million full color pixels, equivalent?
Since when does the Z-axis not count in measurement?
A 10 mp mosaic sensor is a true 10 mp sensor.
A 10 mp X3 sensor (producing a 3.3 mp image) is a true 10 mp sensor.
My Olympus C-2500L advertised a 2.5 mp sensor even though it only created 2.34 mp images. You ever wonder why there's a discrepancy like that? It's because the edge pixels in a mosaic sensor are useless. They need their neighboring pixels to help in the Bayer interpolation.
Your bPixels suggestion is ridiculous. Bayer interpolation deserves better than an artificial equivalency ranking. How can you argue against Fuji's 6 mp SuperCCD which creates 12 mp images? The images really are 12 mp! IMO, one of it is false advertising.
The Sigma SD10 has a 10 million pixel Foveon X3 CMOS sensor. The Polaroid has a 4.5 million pixel sensor which outputs 1.5 megapixel images. I don't see why the SD10 would get an update.
Ooooops, I got the terms mixed up ... I thought the polaroid had 4.5 X3 sensor. DAMN YOU MARKETING FUKKERS!
I can't believe there are still people out there who think a camera with "only" 4.5 MP is inadequate. Unless these consumers are going to make wall-sized posters with their images, realistically 2 MP is enough. MP-whores need to see the light of day and realize that images that are properly exposed and in focus are more important than how many pixels captured them and the possibility for enlarging it to exorbitant proportions.
Anyone that "is" concerned about "megapixels" that much wouldn't even be looking at this camera anyway -- they'd be using medium or large format film or a professional medium or large format Leaf or Kodak back.
But the idea that the Foveon technology is spreading outside Sigma is important. However, the Bayer tech has just gotten really good -- I don't know if the Foveon will ever make appreciable inroads into the market ('s kind of like Macintosh, except Foveon didn't happen first). I mean, take a look at the quality you get out of a Nikon D1x or D2h -- noiseless, perfectly exposed, indistinguishable from print film on paper. But competition is wonderful.
The z axis doesn't count because it isn't a 3d image. It counts for overall fidelity
So what? Megapixels are not directly correlated to final output resolution anyway...not in mosaic sensors, Foveon sensors or SuperCCD sensors. It's a lame benchmark for image quality, so Foveon should absolutely be allowed to advertise its sensors as being three times the grid size.
As for the bPixel thing again, what happens when you downsize the RAW output a 6 megapixel mosaic sensor to ~3 megapixels? What about the interpolated output? Neither resulting truly represents what you'd get with your so called "full-color pixels."
I admit that it's clumsy, but both Foveon and the bayer boys mis-state the resolution of their devices, if only because there isn't a consistent convention about what constitutes a "pixel" across usage and technology.
For years a video camera with a beam splitting prism and three 640x480 CCDs was considered a 300K device, not a 900K device. Although, some video devices have been advertised with this more flattering convention in recent years. The TFT I'm looking at right now is sold as a 1024x768 device, not a 1024x768 X3 device.
4.5MP is still inadequate for many uses if the general goal is to equal the performance of film. For high ISO color work, 4.5MP (x-y) is about the equal of what you can get from film before grain gets in the way. Scientists have put the resolution of 35mm, in a best case scenario, as high as 30MP. In reality, we know that an 8000dpi film scan only returns about a 10-15% more resolution than a 4000dpi film scan, so the truth is somewhere around 4000-4500 dpi, and grain will be a nuissance even before that becomes an issue. Out in the real world, in everyday use, because you can get a cleaner image from (large sensor) digital cameras, we need about 10MP to "equal" the resolution performance of film under most circumstances.
This of course has nothing to do with P&S models like the Polariod mentioned here, but it is something to note. MP matters, even if it is abused by the marketeers.
Comments
It would seem that some industry standard definition of "pixel" might help.
IF a pixel is a discrete location on a 2D surface, then the Foveon only has 3.4MP, but they are 3.4 TRUE, full color pixels.
A 6 "MP" bayer has 6 million discrete locations on a 2D plane, but they aren't TRUE pixels, not full color. Detail and color information is not fully recorded at each site.
Perhaps it's time for an industry body to step in and regulate the marketting and definition of pixels.
Could there be a fair distinction between bayer pixels and X3 pixels -- "bPixels" (for bicubic pixels ? ) and 'full color" or "full spectrum" pixels for X3 ???
Then a Foveon would be 3.4MP or 10 million bPixels.
and a bayer would be 6 Million bPixels or somewhere between 2.5-3.5 million full color pixels, equivalent?
Clumsy, iGuess, harrumpph
A 10 mp mosaic sensor is a true 10 mp sensor.
A 10 mp X3 sensor (producing a 3.3 mp image) is a true 10 mp sensor.
My Olympus C-2500L advertised a 2.5 mp sensor even though it only created 2.34 mp images. You ever wonder why there's a discrepancy like that? It's because the edge pixels in a mosaic sensor are useless. They need their neighboring pixels to help in the Bayer interpolation.
Your bPixels suggestion is ridiculous. Bayer interpolation deserves better than an artificial equivalency ranking. How can you argue against Fuji's 6 mp SuperCCD which creates 12 mp images? The images really are 12 mp! IMO, one of it is false advertising.
Originally posted by Eugene
The Sigma SD10 has a 10 million pixel Foveon X3 CMOS sensor. The Polaroid has a 4.5 million pixel sensor which outputs 1.5 megapixel images. I don't see why the SD10 would get an update.
Ooooops, I got the terms mixed up ... I thought the polaroid had 4.5 X3 sensor. DAMN YOU MARKETING FUKKERS!
Anyone that "is" concerned about "megapixels" that much wouldn't even be looking at this camera anyway -- they'd be using medium or large format film or a professional medium or large format Leaf or Kodak back.
But the idea that the Foveon technology is spreading outside Sigma is important. However, the Bayer tech has just gotten really good -- I don't know if the Foveon will ever make appreciable inroads into the market ('s kind of like Macintosh, except Foveon didn't happen first). I mean, take a look at the quality you get out of a Nikon D1x or D2h -- noiseless, perfectly exposed, indistinguishable from print film on paper. But competition is wonderful.
Originally posted by Matsu
The z axis doesn't count because it isn't a 3d image. It counts for overall fidelity
So what? Megapixels are not directly correlated to final output resolution anyway...not in mosaic sensors, Foveon sensors or SuperCCD sensors. It's a lame benchmark for image quality, so Foveon should absolutely be allowed to advertise its sensors as being three times the grid size.
As for the bPixel thing again, what happens when you downsize the RAW output a 6 megapixel mosaic sensor to ~3 megapixels? What about the interpolated output? Neither resulting truly represents what you'd get with your so called "full-color pixels."
For years a video camera with a beam splitting prism and three 640x480 CCDs was considered a 300K device, not a 900K device. Although, some video devices have been advertised with this more flattering convention in recent years. The TFT I'm looking at right now is sold as a 1024x768 device, not a 1024x768 X3 device.
4.5MP is still inadequate for many uses if the general goal is to equal the performance of film. For high ISO color work, 4.5MP (x-y) is about the equal of what you can get from film before grain gets in the way. Scientists have put the resolution of 35mm, in a best case scenario, as high as 30MP. In reality, we know that an 8000dpi film scan only returns about a 10-15% more resolution than a 4000dpi film scan, so the truth is somewhere around 4000-4500 dpi, and grain will be a nuissance even before that becomes an issue. Out in the real world, in everyday use, because you can get a cleaner image from (large sensor) digital cameras, we need about 10MP to "equal" the resolution performance of film under most circumstances.
This of course has nothing to do with P&S models like the Polariod mentioned here, but it is something to note. MP matters, even if it is abused by the marketeers.