More Dubai disclosed K20D features

[lol101]

Originally Posted by Falcons

I just hope it takes as good photos as I have seen produced by *ist D, *istDS, *istDL, K110D, K100D, K100D Super & K10D - of course with a little help from those enlightened Pentaxians behind the viewfinder.

Anyway to all those frantically waiting to see what bells & whistles are soon to be available - make sure you still check into the "Post your Photos" threads - give yourself a little reminder of what it is all about.

Go on check out a few photos, make a comment and brighten up someone's day

No, no, no, no, no... wrong method...

Look it's easy:

Hey guys, have you seen those K20D pictures hidden in the "Post your Photos" thread?

There very well hidden but once you find them: WOW, it's GIGANORMOUS!!

[FotoPete]

Hey probably a crazy thought but what *IF* the 14 MP sensor is actually the Foveon X3 Pro CMOS found in the Sigma SD14??

The field of view crop would jump to 1.7X but you think its a possibility?

[Falcons]

Originally Posted by lol101

No, no, no, no, no... wrong method...

Look it's easy:

Hey guys, have you seen those K20D pictures hidden in the "Post your Photos" thread?

There very well hid but once you find them: WOW, it's GIGANORMOUS!!

LOL
Yes you have caught me out, I was trying to be subtle. Well now you have blown the lid on it - everyone check out the photos on the "post Your Photos" thread see if you can tell the difference between an *ist series and a K(whatever)D series and perhaps you may even see the photo taken by the K20D - or I could be making it all up, see for yourself.

[mawz]

Well, for starters the Nikon D3 does 5.4MP in DX mode at 11fps, it'll do 11fps in 5:4 crop mode at 10MP as well. So there's 1 oops (1080p is 2.1MP, far less than what the D3 puts out in DX mode).

If the K20D is 16:9, I'm flat out not interested. 3:2 is already too wide for me. Something along the lines of the Nikon D3's 5:4 crop mode is far more interesting (I'd buy a 4/3rds body if the noise performance was competitive and the UI wasn't so bad).

Don't get me wrong, I love my K10D. But if the K20D is more of a still/video convergence device like this rumour hints at, I'll be sucking it up and going for a Sony A700 instead.

[mawz]

Originally Posted by FotoPete

Hey probably a crazy thought but what *IF* the 14 MP sensor is actually the Foveon X3 Pro CMOS found in the Sigma SD14??

The field of view crop would jump to 1.7X but you think its a possibility?

Very doubtful, that's not a 14MP sensor, no matter what Foveon and Sigma claim. It's got 14 million sub-pixels.

[mattdm]

Originally Posted by mawz

Very doubtful, that's not a 14MP sensor, no matter what Foveon and Sigma claim. It's got 14 million sub-pixels.

Sure, and the K10D has 10 million subpixels. They're just on top of each other in the Foveon sensor so the interpolation comes out differently.

But I agree it's pretty doubtful -- Foveon has enough disadvantages that I'd be surprised to see Pentax go that way.

[mawz]

Originally Posted by mattdm

Sure, and the K10D has 10 million subpixels. They're just on top of each other in the Foveon sensor so the interpolation comes out differently.

But I agree it's pretty doubtful -- Foveon has enough disadvantages that I'd be surprised to see Pentax go that way.

Umm, no. The Bayer sensors used in most digitals do not use sub-pixels. They use full pixels entirely(1 sensor site= 1pixel), but interpolate colour information (Which degrades detail over the same number of full-colour pixels). Foveon sensors use 3 sub-pixels per pixel site, providing better colour renditions at the cost of significantly worsened high ISO performance and very poor resolution compared to a bayer sensor with the same number of sensor sites (but 3x the pixels as it's a direct mapping in bayer sensors). The Fuji SuperCCD sensors use the same sort of math as the Foveon's (but with 2 sites per pixel), but gain resolution over standard bayer sensors with the same amount of pixels due to their more interpolation-friendly layout.

Calling the Sigma SD14 a 14MP camera is an exercise in marketing, no more. In reality it's a 4.6MP body which delivers slightly more resolution than a 6MP bayer sensor (as the bayer interpolation costs you resolution in return for getting usable colour information).

[mattdm]

Originally Posted by mawz

Umm, no. The Bayer sensors used in most digitals do not use sub-pixels. They use full pixels entirely(1 sensor site= 1pixel), but interpolate colour information (Which degrades detail over the same number of full-colour pixels).

This is semantics, and basically marketing from the other side. 1 sensor site does not by itself provide all the information for a pixel, but rather a portion of it. That's is to say, 1 sensor site = 1 subpixel, and the subpixels are color-interpolated into full RGB pixels.

This is exactly the same as subpixels on an LCD monitor, which can be used to provide better shapes for TrueType fonts ("ClearType") at the expense of color artifacts. The subpixels are still called subpixels even though they're used to provide higher resolution. (Little LCD screens like those used on the back of digital cameras are in fact sold/marketed by the number of subpixels, because higher numbers are more impressive. But thank goodness it hasn't caught on to say that the monitor I'm using has a 3840x1024 resolution.)

[ogl]

Originally Posted by mawz

Calling the Sigma SD14 a 14MP camera is an exercise in marketing, no more. In reality it's a 4.6MP body which delivers slightly more resolution than a 6MP bayer sensor (as the bayer interpolation costs you resolution in return for getting usable colour information).

YOU ARE RIGHT!
All tests are confirmed that real resolution of SD14 is not above 6.5-7 MP bayer sensor.

[mattdm]

Originally Posted by ogl

YOU ARE RIGHT!
All tests are confirmed that real resolution of SD14 is not above 6.5-7 MP bayer sensor.

I wouldn't say "all tests" -- Popular Photography puts it more on par with 8mpix bayer sensors in JPEG and almost like 10mpix in RAW. They're usually pretty generous in their testings, but then, that generosity goes to the competition too.

Anyway, that's not really the point -- both sensor types use a subpixel-based design, just differently. The bayer array gets more resolution by interpolating color into final files which use one pixel per subpixel, but again, that's just what font rendering on laptop screen does and doesn't actually mean you have any more "real" pixels in the sensor.

...

I just came across an interesting paper on rendering raw Bayer data to LCD screens in a way that takes advantage of this. Basically, does the color interpolation with translation from bayer subpixels (whatever you want to call them) to LCD subpixels, rather than bayer->"naïve" whole pixels. Pretty cool:

http://www.cs.ubc.ca/~atcheson/f/atc...d%20images.pdf

Which underscores another point I should make: I'm not saying the Bayer approach is bad (in fact, I think it or something like it has more of a future than Foveon). It just does use an interpolated subpixel approach, which can even be exploited for useful display improvements.

[mawz]

Originally Posted by mattdm

I wouldn't say "all tests" -- Popular Photography puts it more on par with 8mpix bayer sensors in JPEG and almost like 10mpix in RAW. They're usually pretty generous in their testings, but then, that generosity goes to the competition too.

Anyway, that's not really the point -- both sensor types use a subpixel-based design, just differently. The bayer array gets more resolution by interpolating color into final files which use one pixel per subpixel, but again, that's just what font rendering on laptop screen does and doesn't actually mean you have any more "real" pixels in the sensor.

Bayer does NOT use a sub-pixel-based design. Bayer sensors use colour interpolation to eliminate the need for sub-pixels. You're seriously stretching the definition of sub-pixels to fit bayer sensors into a technology that they do not use.

And PopPhoto is in no way a usable reference for any sort of testing. They can't even get basic specs right (See their 3.2fps Rebel XT review) let alone actual performance. Everything always tests well in their tests.

[ogl]

I heard that 16:9 ratio of sensor in K20D is not true. Maybe mask to 16:9.
But sensor is 3:2.

[ogl]

Originally Posted by mattdm

I wouldn't say "all tests" -- Popular Photography puts it more on par with 8mpix bayer sensors in JPEG and almost like 10mpix in RAW. They're usually pretty generous in their testings, but then, that generosity goes to the competition too.

Popular Photo is not serious site and magazine, it's simply POPULAR like POP-MUSIC.
Their reviews are funny. Easy-reading.
It's pity, but SD14 has no resolution above 7 MP bayer even in RAW.
Maybe 8 MP. OK. Let it be.
But forget about 10 MP.