[kenyee]

O
USA (E = <$800, M = $801-$2600, P = >$2600)
2005 - E = 50.5%, M = 48.4%, P = 1.1%
2006 - E = 50.1%, M = 48.7%, P = 1.2%

LOL. That answers how many consumers can afford FF. That does have to be put in price perspective though...the profit/unit of that 1.1% is higher than the cameras in the M range I'd bet.
Anyways, when they eventually get FF into the 40D/D300 class is when Pentax needs to do a FF camera. That isn't now though :-)

[distudio]

MF cameras using FILM yes. Can't you get the difference? A given film emulsion had the SAME grain density whether it was MF of 35mm therefore a MF negative has far more grains overall. But we are talking about 2 sensors with the SAME number of photosites, therefore the smaller sensor has a higher resolution! At the end of the day you still get 12MP from both cameras assuming lenses that are decently sharp.

If you have two sensors of 12MP one FF and another APS then yes the APS has a higher absolute resolution but both have the same resolution as a ratio of image width. Disregarding AOV the if using the same lens on both sensors the FF image may show a higher final resolution per pixel due to the lower absolute resolution required of the system as a whole (assuming the a real lens).

[distudio]

Okay, here are some stats from a recognised international market survey comapny (I can't reveal their name, it's supposed to be confidential) for 2005 and 2006, I don't have 2007, probably not available yet.

Europe (Entry = <800 Euro, Mid = 801-2000 Euro, Pro = >2000 Euro)
2005 - E = 62.2%, M = 34.2%, P = 3.6%
2006 - E = 76.4%, M = 21%, P = 2.6%

USA (E = <$800, M = $801-$2600, P = >$2600)
2005 - E = 50.5%, M = 48.4%, P = 1.1%
2006 - E = 50.1%, M = 48.7%, P = 1.2%

From that, I guess you can see why Pentax are doing what they are doing.

Thanks Richard, very interesting stats, I wonder what the percentage would have been if the P market was split into 2600-10000 Euro and >10000 Euro? ;-)

[*isteve]

Hi thePiRaTE,

you are totally right. Of course, we must assume both the same pixel density for the sensor and the same MTF (resolution) for the lenses. Otherwise, we would compare apples with oranges. The very best 35mm lenses (ZEISS, Leica) are able to deliver about >100 MPixels while most lenses in use are inferior to 20MPixels (assuming full frame).

Now take as resolution the smaller of the two measures, (1) lens resolution and, (2) sensor resolution. Then, a 300mm FF image gives you 125% more information (more pixels or whatever) than a 200mm APS-C image.

If both sensors have 12 MP, then both images with have 12MP. Where did the extra 125% come from?

Perhaps a browse through this
Understanding resolution and MTF

Would be useful some time. When you find the bit about calculating system resolution and working out the linear resolution for a sensor let me know.

Well your formula is not quite correct is it. A 12MP APS sensor has more resolution (more pixels per linear mm) which exactly cancels out the shorter length (22 vs 36 mm) to get the same number of horizontal pixels.....

OK lets compare actual 12MP APSC and a 12MP FF camera then, from DP Review, and perhaps you can explain.

Absolute res Extinction res

H Canon 5D 2300 lppm 2500 lppm
V Canon 5D 2000 lppm 2500 lppm

H Sony A700 2200 lppm 2900 lppm
V Sony A700 2100 lppm 2800 lppm

Almost an exact match. Hmm. How can that be? By your "theory" the 5D should have quite a lot more resolution. In fact the tests were performed using a FF 50mm lens! If a dedicated APSC lens were fitted to the Minolta it may beat the Canon by more still.

Hey let go REALLY crazy and compare it with a 12MP digicam with a titchy 7.2mm side sensor...a Canon G9...

Absolute Extinction
H 1950 2400
V 1900 2400

Wow - not bad huh - its got a wonky 4X zoom lens stuck on the front and its almost matching a Canon 5D for resolution. Guess Phil must be lying huh!

When the tests of the Nikon D3 and D300 come out it will be interesting to compare wont it.

Hi thePiRaTE,
The images will look the same until you need maximum detail. Denying this would mean denying that middle format cameras can deliver better IQ than 35mm.

MF cameras using FILM yes. A given film emulsion had the SAME grain density whether it was MF of 35mm therefore a MF negative has far more grains overall. But we are talking about 2 sensors with the SAME number of photosites, therefore the smaller sensor has a higher pixel density/resolution! At the end of the day you still get 12MP from both cameras assuming lenses that are decently sharp.

In fact lens resolution for an APS dedicated lens can be higher than for a FF lens because it does not have to cover such a wide circle and requires less correction for spherical aberation. Check Canon EF-S 18-55mm f/3.5-5.6 IS - Test Report / Review for resolution figures for Canons new APSC kit lens vs one of their top FF primes. Its actually higher even though its a cheap little zoom.

Youll also find the same generally happens if you compare the newer Pentax DA zooms against the older FA zooms. More resolution. Its not because they are higher quality its just easier to design lenses for smaller image circles.

And finally, because the sensor resolution is so much higher, APSC sensors can generally get away with slightly weaker AA filter which makes them sharper on a per-pixel basis.

So before you go on about FF sensors at least know your facts and learn a bit about the physics.

The real advantage of a FF sensor compared to an APSC one is that it should have less noise at high ISO for the same number of MP (2X the pixel area = about a stop) or double the resolution (2X as many pixels) for the same noise. However this will present more of a challenge because you are using full image circle lenses. I would lay bets that the 24MP Nikon D3X does not double the res of the D3.

However there are real disadvantages too. Larger sensors with larger pixels suffer from more aliasing, moire, light falloff and edge softness. An FF image taken with a less than stellar lens (or any lens wide open) will not look too hot around the edges. It will also get hotter faster and live-view will be harder to implement as a result. It may even cause low ISO noise to be higher than an APSC equivalent.

Also, advances in sensor design have also delivered a 1 stop improvement in noise for APS sensors. They are just a year or so behind.

So, the advantage at low ISO is minimal, the advantage at high ISO is about a stop (using identical sensor design). On the other hand you have a bigger mirror box (slower shutter) and around a $1500 premium over price of an equivalent APS camera (Canon 30D --> 5D or Nikon D2X --> D3). Assuming most of that premium is the sensor and mirror box upgrade, then any FF camera is likely to cost around $1500 more than a similarly specced APS version. Thats a lot to pay for a stop less noise which will be wiped out in a year....

[*isteve]

If you have two sensors of 12MP one FF and another APS then yes the APS has a higher absolute resolution but both have the same resolution as a ratio of image width. Disregarding AOV the if using the same lens on both sensors the FF image may show a higher final resolution per pixel due to the lower absolute resolution required of the system as a whole (assuming the a real lens).

Using a decent lens the difference will be minimal, and using a dedicated APSC lens will cancel out any abvantage.

However there is not 125% more information in the FF image now is there?

Plus check DPReview for comparison of Canon 5D and Sony A700. Pretty much the same, and this is using an FF lens on the APSC camera. To quote Pentax, the standard 50mm will not be as sharp as the new digitally optimised lenses we will be announcing soon....

[distudio]

Using a decent lens the difference will be minimal, and using a dedicated APSC lens will cancel out any abvantage.

However there is not 125% more information in the FF image now is there?

Resolution no, but if you want to apply it to overall image quality then yes, it would be safe to assume that for a sensor of similar age and technology capture latitude would be far higher providing for greater overall fidelity. I suppose we could draw upon test images from a top end 12MP P&S to give the fidelity vs resolution argument some clarity.

Plus check DPReview for comparison of Canon 5D and Sony A700. Pretty much the same, and this is using an FF lens on the APSC camera. To quote Pentax, the standard 50mm will not be as sharp as the new digitally optimised lenses we will be announcing soon....

Firstly to be fair they're not technologically matched but I'll concede that the A700 at low ISO has very similar res to the 5D, however hike the ISO up to 3200 and do a comparison again and the 5D kills it. Tests will prove just how capable the new lenses are, until then I'd prefer to reserve judgment as I've found flaws with their assessment of new lenses before.

Cheers,

[*isteve]

Resolution no, but if you want to apply it to overall image quality then yes, it would be safe to assume that for a sensor of similar age and technology capture latitude would be far higher providing for greater overall fidelity.



Firstly to be fair they're not technologically matched but I'll concede that the A700 at low ISO has very similar res to the 5D, however hike the ISO up to 3200 and do a comparison again and the 5D kills it. Tests will prove just how capable the new lenses are, until then I'd prefer to reserve judgment as I've found flaws with their assessment of new lenses before.

Cheers,

Not technologically matched, no, but in this case the lens SHOULD have favoured the 5D. High ISO on the A700 looks a lot worse than the K20D, but I concede a stop of noise using similar architecture, consistent with 2X the sensitivity. But thats always been the case, and frankly thats about the only real advantage of FF and even though people refuse to accept it sometimes, there are some nasty downsides too in addition to the cost. Looking at the 40D the noise profile is very similar to the 5D and the resolution is only slightly lower.

[distudio]

Not technologically matched, no, but in this case the lens SHOULD have favoured the 5D. High ISO on the A700 looks a lot worse than the K20D, but I concede a stop of noise using similar architecture, consistent with 2X the sensitivity. But thats always been the case, and frankly thats about the only real advantage of FF and even though people refuse to accept it sometimes, there are some nasty downsides too in addition to the cost. Looking at the 40D the noise profile is very similar to the 5D and the resolution is only slightly lower.

All the per format res/noise arguments aside I'd still far prefer to have the creative flexibility of my 31mm LTD on a FF body than have to contend with a 21/3.2 to cover the same AOV.

Cheers,

[*isteve]

All the per format res/noise arguments aside I'd still far prefer to have the creative flexibility of my 31mm LTD on a FF body than have to contend with a 21/3.2 to cover the same AOV.

Cheers,

Ah well thats a personal creative choice and thats fine, but would not sell many cameras for Pentax since not many of us have the 31mm. I find it a great substitute for a standard lens, but I am surprised there is not a fast non-pancake 24 or 20mm on the horizon. They made both before so they have the "pattern" as it were. The 15 looks interesting as well.

But I was originally reponding to a claim that FF would have 125% more resolution, and this is clearly not true from a theoretical or evidential POV.

[distudio]

But I was originally reponding to a claim that FF would have 125% more resolution, and this is clearly not true from a theoretical or evidential POV.

Spacial resolution maybe not, however the resolution of DR will generally be greater, the percentage depends on the camera but it should generally be larger for a FF camera.

[*isteve]

Spacial resolution maybe not, however the resolution of DR will generally be greater, the percentage depends on the camera but it should generally be larger for a FF camera.

Not that much though. An extra stop in the shadows is not going to be that noticeable once the image is processed and a tone curve applied.

[distudio]

Not that much though. An extra stop in the shadows is not going to be that noticeable once the image is processed and a tone curve applied.

Noise may be better by a stop or more but for me it's the highlights where they shine. I've been playing with some 6400ISO D3 RAW files and I can tell you from practical experience that the amount of headroom in them is close to what the K10D provides at 200ISO, quite extraordinary for a small format digital camera.

My K10D is on the block, it's never been a body that I've been particularly enamored with IQ wise, lets hope that the K20D is a vast improvement for now.

[falconeye]

Perhaps a browse through this
Understanding resolution and MTF
Would be useful some time. When you find the bit about calculating system resolution and working out the linear resolution for a sensor let me know.

Hi *isteve,

you are a true evangelist, aren't you? Maybe, I should confess that I hold a Ph.D. in Physics. I may have used simple words but I think I 100% understood what I wrote...

If both sensors have 12 MP, then both images with have 12MP. Where did the extra 125% come from?
[... ]
Well your formula is not quite correct is it. A 12MP APS sensor has more resolution (more pixels per linear mm) which exactly cancels out the shorter length (22 vs 36 mm) to get the same number of horizontal pixels.....
[...]
OK lets compare actual 12MP APSC and a 12MP FF camera then, from DP Review, and perhaps you can explain.

The word "resolution" is a physical measure (such as dpi for computer screens). Two sensors built with equal technology will have equal resolution. Therefore, the larger of the two will sport more sensor cells. The size of a sensor cell is the technology constant, not their number.

This is why current typical numbers are 12MP for APS-C, 24MP for 35mm FF and ~40MP for 645 FF.

If you make sensor cells smaller, you get higher ISO noise etc. This happens for P&S cameras, not for DSLRs though. Fortunately, because more than 6 MP for P&S cameras deliver worse IQ, cf. Best picture quality with 6 megapixels!...

Wow - not bad huh - its got a wonky 4X zoom lens stuck on the front and its almost matching a Canon 5D for resolution.

Resolution tests aren't standardized. A comparison of sample photos is quite telling, though. Or find a scientific measurement of MTF.

In fact lens resolution for an APS dedicated lens can be higher than for a FF lens because it does not have to cover such a wide circle and requires less correction for spherical aberation.

Resolution of a lens will always be weaker near the borders and corners. It is not becoming better in the center just because you make it weaker at borders. It could be, theoretically, I agree. But this is not covered by MTF measurements of existing lenses. I see a certain amount of wishful thinking in here

[*isteve]

Noise may be better by a stop or more but for me it's the highlights where they shine. I've been playing with some 6400ISO D3 RAW files and I can tell you from practical experience that the amount of headroom in them is close to what the K10D provides at 200ISO, quite extraordinary for a small format digital camera.

My K10D is on the block, it's never been a body that I've been particularly enamored with IQ wise, lets hope that the K20D is a vast improvement for now.

Well we will see. The K10D was worse than the K100D, but I think it was the internal processing, because the Nikon D80 had less of an issue.

Its not really a sensor issue though. A sensor is a linear recording device. I can absorb 2X as many photons with a photosite of 2X the area which allows me more scope with the tine curve, but at the high end, once its full its full. The "rolloff" in phils DR tests are more a result of the way the internal processing works and how the tone curve is applied.

The K20D is better as far as I could tell. How much better is hard to say as the firmware I had was not final and the DR expansion feature was not working.

[distudio]

Its not really a sensor issue though. A sensor is a linear recording device. I can absorb 2X as many photons with a photosite of 2X the area which allows me more scope with the tine curve, but at the high end, once its full its full. The "rolloff" in phils DR tests are more a result of the way the internal processing works and how the tone curve is applied.

The K20D is better as far as I could tell. How much better is hard to say as the firmware I had was not final and the DR expansion feature was not working.

The tone curve should affect the gamma but should not affect the black and white points (overall DR) assuming they aren't designed to clip either end. The well saturation (related to the sensors natural virtual ISO) also has to be factored into the DR equation, it's far easier to build a deeper well over a larger pixel area.