Need more megapix? Future possibilities and no need for full frame!

[rburgoss]

Since several days ago, while checking some of my pictures, sometimes (on screen of course), I frequently have to check EXIF data to see with which of my bodies (cameras) such picture was taken with. Among this, I have plenty of pictures shot with my Optio Wpi (6 megapix) p&s.

Noticing the quality of the wpi pics is not as bad as they should be, when compared to pics shot with my K20D, I did the following math excersise:

Attachment 19784

If Pentax (or suppliers) could make a sensor with 340 pixels per millimiter (for the WPI), that means a sensor for an APS-C DSLR with the same pixel density would be a 42 megapix sensor.

I know for sure that the numbers given by manufacturers are not always true, because the real pixel count is one thing and effective pixels is another. Interpolation plays a big role here. Anyway, what limits sensor manufacturers to produce such a sensor?

BTW. At 14.6 megapix from my K20D, I can now draw the line between good glass and not so good glass. When compared to my istD, I feel like in the back film days, being the istD like shooting with Tri-X (400 ISO) and the K20D with Kodak Panatomic X (32 ISO). Those sure made a great difference in sharpness.

Considering this, I sometimes wonder if its worth it to have more megapixels or even a full frame sensor with the same density of the D20D, only to discover that image quality limitation would sit on the glass used.

I will try to do some real world tests with my Wpi, side by side with my K20D, trying to compare apples to apples (that is, to compare similar size pixels pictures with similar images). I will try to post later.

Robert B.

[Big Dave]

Robert,
That is a great exercise. I have often thought about this, but gone to the effort that you have. Good job. It really sheds light on the real situation.

Dave

[k100d]

the problem is to get 42 megapixels that dense, the demand on the lenses will be huge, especially if the photosites are so tiny. and to have the same performance as the Optio sensor wouldn't be very acceptable. hopefully advances in the future can help this


FF has other advantages
-one much bigger viewfinder. i just got a film SLR and the viewfinder is so much better than the tiny one on the K100D
-putting beause of the larger photosites you can get stuff like the D3 which has great high ISO performance

[Frank Fletcher]

No more megapixels please!!!

The images from my gfs K100D Super way exceed aither of my K10Ds imo. Of course, I prefer the ergonomics/handling/weather-proofing/etc of my K10s.

More DR, brighter viewfinder, and especially sharper local contrast please.

I can't currently afford a 35mm, 6x?cm or 4x5" digital camera - but if Pentax can bring it down to the masses - I am soooo there.

The best scenario financially is if a (real) 35mm FF comes out - I can use all my FF primes. If some MF like a 645d comes out ... I'd love to buy in but I suspect it'd be too expensive + lenses.

[Frank Fletcher]

It's been saaid before (many times), but another thing Pentax could do that would help me out is to make a <=10mp version of a k20d, using the same upgraded cmos technology.

BTW, I often like BW film comparisions. I used to prefer (MF) Kodak Technical Pan. developed in Technidol or Agfa Rodinal (preferred). If I wanted the grit look I'd push Kodak Tri-X or Ilford HP5 - developed in Rodinal or D76 depending on what feel I wanted. If (rarely) I wanted smooshie grain, I'd use a standard T-MAX100 in D76.

My favourite was TechPan in Rodinal - if that tells you anything about my desired sensor technology.

[rburgoss]

Originally Posted by k100d

the problem is to get 42 megapixels that dense, the demand on the lenses will be huge, especially if the photosites are so tiny. and to have the same performance as the Optio sensor wouldn't be very acceptable. hopefully advances in the future can help this


FF has other advantages
-one much bigger viewfinder. i just got a film SLR and the viewfinder is so much better than the tiny one on the K100D
-putting beause of the larger photosites you can get stuff like the D3 which has great high ISO performance

Agree on the viewfinder. I still mis the IMAX sensation from my old MX body when looking into the viewfinder. About the density, its there. The Optio WPI has that density. Its sensor performance is quite good and I'm affraid to say its better than the one on the K20D, because it has more pixels in the same capture area. See the following test:

Test goal: To have the same subject captured by two different sensors (K20D and Optio Wpi), with the same magnificacion, using the same focal length. For this I Used the WPI at 18.9 mm setting -its longest f/l- and the K20D with my DA12-24, set at 18 mm. The crop factor between each other is what gives the angle of view).

Attachment 19791

Both images are basic JPGS from both cameras, both full frames captured, without any PP. Wpi set at 6 mpix and K20D at 14.6 mpix. I guess this image proves my point. The WPI shows a lot better resolution than the K20D, given same subject, same distance, same focal length, even same ISO (BOTH at 200 ISO). Of course, the difference comes from the pixel density which is about 60% higher than of the K20D.

Are you hearing this PENTAX? How about giving us a DSLR with the same pixel density as the Optio's.... Or at least, implant the same interpolation alogarithms from the OPTIO to a new DSLR.... Or could it be that the Optio has better glass than the DA 12-24... at the very center of the frame? Nahhhh.....

Robert B.

[deejjjaaaa]

Originally Posted by rburgoss

The WPI shows a lot better resolution than the K20D, given same subject, same distance, same focal length, even same ISO (BOTH at 200 ISO).

I am not sure why you
claim the 18.9mm mark on WPI's lenses
is the same as 18.9mm on 12-24 ("...with my DA12-24, set at 18 mm...") ? 12-24 are FF's distances... you need to set it @ 12mm if you want to have APS-C's 18mm... and then to make it fair WPI's 18.9 = FF's 114 = APS-C's 76

[Ash]

Very interesting exercise.
I get your point - must be a logistical reason for that...

[Ash]

Originally Posted by deejjjaaaa

I am not sure why you
claim the 18.9mm mark on WPI's lenses
is the same as 18.9mm on 12-24 ("...with my DA12-24, set at 18 mm...") ? 12-24 are FF's distances... you need to set it @ 12mm if you want to have APS-C's 18mm... and then to make it fair WPI's 18.9 = FF's 114 = APS-C's 76

Looks like what's been done is an equivalent f/l (and thus perspective) between the 2 cams provides the two different images, with the WPI image cropping only the small area in the centre of the K20D's image by virtue of its smaller sensor and viewfinder.

I'm also unsure that this is a fair comparison, but it challenges the idea that bigger is better...

[Stevopedia]

Originally Posted by deejjjaaaa

I am not sure why you
claim the 18.9mm mark on WPI's lenses
is the same as 18.9mm on 12-24 ("...with my DA12-24, set at 18 mm...") ? 12-24 are FF's distances... you need to set it @ 12mm if you want to have APS-C's 18mm... and then to make it fair WPI's 18.9 = FF's 114 = APS-C's 76

I think the point he's trying to make here is that if the K20D had the same pixel density as the WPI, then the two (WPI shot and K20D crop) should be identical, because the amount of sensor that captured the image is exactly the same if the focal length is exactly the same.

[dosdan]

The size of the chain links seems to indicate the the equiv. FL is the same for both shots. I'm surprised that the quality of the lens in the the P&S didn't pole-axe it in this test. How did they produce a small lens with such resolving power?

Dan.

[rburgoss]

Originally Posted by dosdan

The size of the chain links seems to indicate the the FL is the same for both shots. I'm surprised that the quality of the lens in the the P&S didn't pole-axe it in this test. How did they produce a lens with such resolving power?

Dan.

That is exactly my point. Remember the good ol' film days? When we, minor beings using 35 mm cameras drooled over the Medium Format results?

During those days, the film was the same, just cutted differently (for 35 mm or medium format), but MF results were stunning in matters of resolution. Why? Because there was more film to capture the same image. If I remember correctly too, MF lenses even though had (still have) great resolution, but the 35 mm lenses had a lot better resolution. Things being equal (film), then the final images were a lot better in MF, just because there was more film.

Film grain was the same for both formats, but in 35 mm, film enlargment had to be a lot bigger to get the same size as mf negatives.

In the digital era, film no longer exists. Just sensors, which technically do not have "grain", just pixels. But as my test proves it, it really doesn't matter how many pixels there are. What matters is the pixel density on which to capture the image.

Consider this, the K200D is a 10 mpix camera. The K20D is a 14.6 megapix one. Both have the same sensor size (APS-C), which means, the K20D has tighter pixel density in order to fit that many pixels in the same sensor size. Think about the istD with 6 mpix? (I have one ) Which also means its sensor has bigger pixels.

In short words, what matter is pixel density to get detail. Of course, total pixel count matters a lot.

Full frame DSLR's with 12 to 14 mpix sensors will not give me more detail and sharpness as my K20D. The only pro is that will allow me to use my old FF glass to its full potential, but nothing more.

OTOH, give me an APS-C sensor with 42 mpix, then we are talking business. Technology exists to produce such pixel density (The optio wpi has it! )

I rest my case. I am sure the guys at Pentax already know this. There is no need for FF to get the results we want. And BTW, I've been recently comparing some shots from my K20D with some old negative drum scans (Fuji Superia) (24 mpix scans), and to be honest I can see better detail from my K20D.

Robert B.

[kenyee]

Originally Posted by rburgoss

what matter is pixel density to get detail. Of course, total pixel count matters a lot.

The other side of this is that high pixel density = smaller photosites = crappy low ISO performance...

Other companies are jumping to FF because they get better ISO performance and there is a large existing base of FF lenses and it seems to be the best compromise between body size vs. image quality if you're using an SLR.

You can say MF is the best image quality, but it seems to be a big jump in camera size compared to FF vs. APS-C (the D300/D700 use the same body except for the viewfinder bulge)...

[deejjjaaaa]

ok, I was wrong about P&S lens marking - mea cupla, need to brush
off all that dust from my old 2002 nikon p&s... however, the picture posted
seems strange

K20D has 23.4mm x 15.6mm sensor and WPi has 5.75mm x 4.31mm sensor, the ratio is ~14:1, but
full frames from K20D and WPi (not crops w/ a piece of keychain) have ratio ~9:1

[Marc Sabatella]

Originally Posted by kenyee

The other side of this is that high pixel density = smaller photosites = crappy low ISO performance...

Excactly! ou might not be able to tell pictures from the Optio from those of the K100D at ISO 100, but try again at 1600...

On the other hand, while I keep hear we're running up against laws of physics here and you just *can't* get noise levels down much more with smaller pixels, I'm fortunately ignorant enough about the physics involved to remain hopeful that someday, we'll find a way :-)