Sony NEX9 - wild speculation!

[ErikKaffehr]

Hi,

According to the "Sony Alpha Rumors" site there will be three full frame models 2012. They used to be well informed. I guess that leaking information is part of Sony's marketing.

Best regards
Erik

This all sounds likely in theory. But in practice, the first available image samples of the NEX-7 suggest quite the contrary, don't they? In his first report here on Luminous Landscape, there is even mention of how *weak* the AA filter seems to be on this camera, and how good fine detail and texture is rendered by it. There is a bit of a penalty regarding high ISO, but as the quality at lower ISO seems to be so excellent, you could just argue that there are enough photographers who will really like that sensor as it is now: Offering so superior resolution at lower to mid ISO.

Regarding FF sensor, I recall that some Sony rep has once said that it would be possible with the NEX mount. However, I do not know if that was a really technically informed source. Of course a FF NEX body would be very nice. I rather doubt it will be like this in the near future. Rather they will have a traditional A900 SLR replacement for the Alpha mount system, and/or one with SLT mirror in it.

Thomas

[Fine_Art]

The site (sonyalpharumors.com) that got all of the information about these Sony cameras absolutely right (including the 24MP APS-C sensor info by October, 2010), are stating (quoting from the same reliable Sony sources) that there would be 3 Full-frame cameras to be announced in 2012. 

I think 2 SLR/SLT Full-frames make sense but the 3rd one is not going to be an SLT/SLR at all and is going to be a NEX.

It could be a Nex format (size) camera with an alpha mount for the Zeiss/ G lenses.

[fotometria gr]

This all sounds likely in theory. But in practice, the first available image samples of the NEX-7 suggest quite the contrary, don't they? In his first report here on Luminous Landscape, there is even mention of how *weak* the AA filter seems to be on this camera, and how good fine detail and texture is rendered by it. There is a bit of a penalty regarding high ISO, but as the quality at lower ISO seems to be so excellent, you could just argue that there are enough photographers who will really like that sensor as it is now: Offering so superior resolution at lower to mid ISO.

Regarding FF sensor, I recall that some Sony rep has once said that it would be possible with the NEX mount. However, I do not know if that was a really technically informed source. Of course a FF NEX body would be very nice. I rather doubt it will be like this in the near future. Rather they will have a traditional A900 SLR replacement for the Alpha mount system, and/or one with SLT mirror in it.

Thomas

Actually its the opposite, the first impressions for the same performance a77 seem to saw that it is noticeably worst than the 16.2mpx cameras, there are things that it does superbly but not photographic image quality. Regards Theodoros www.fotometria.gr

[ErikKaffehr]

Well,

Michael shows samples he has shot with the A65 and it is definitively sharp at the pixel level (looking at the tattoed arm shots). He doesn't mention sharpening, and that can affect sharpness a lot.

Noise performance is much less impressive in the same evaluation.

In a way a 24 MP camera is more demanding on lenses than a 16 MP camera because we all pixel peep at actual pixels, but pictures are never seen at actual pixels. Either we see them on a low res screen, full HD is only 2 MPixels! Or we print. In print the image will be interpolated to whatever resolution the printer has. A larger image will interpolate better.

So I guess we need to wait with conclusions until:

There are real production samples available, controlled tests are done and real world raw converters are available.

Best regards
Erik

Actually its the opposite, the first impressions for the same performance a77 seem to saw that it is noticeably worst than the 16.2mpx cameras, there are things that it does superbly but not photographic image quality. Regards Theodoros www.fotometria.gr

[memento]

Actually its the opposite, the first impressions for the same performance a77 seem to saw that it is noticeably worst than the 16.2mpx cameras, there are things that it does superbly but not photographic image quality. Regards Theodoros www.fotometria.gr

The A77 does have similar, but not really the same performance than the NEX-7 because of

(1) semi-transparent mirror in the light path that might, in fact, cost some of the very finest details (there are reports in the net that nicely show the subtle image quality differences between a Sony SLT camera with the mirror removed)

(2) most shots around have been taken with the new 16-50/2.8 zoom lens which, very likely, just won't perform to the same super-high standard than the new fixed focal length lenses

(3) regarding high ISO, the SLT mirror eats 1/2 f-stop of light, so the high ISO performance of the NEX-7 will be better by this 1/2 f-stop

I agree with Erik that we should be careful with the evaluation of the NEX-7 image quality as there are only very few pre-series cameras around and even the testers are just getting the very first results. However, there ARE a few shots from the NEX-7 out there (and I am not talking A77 samples!) and they surely look very, very promising.

Thomas

[fotometria gr]

The A77 does have similar, but not really the same performance than the NEX-7 because of

(1) semi-transparent mirror in the light path that might, in fact, cost some of the very finest details (there are reports in the net that nicely show the subtle image quality differences between a Sony SLT camera with the mirror removed)

(2) most shots around have been taken with the new 16-50/2.8 zoom lens which, very likely, just won't perform to the same super-high standard than the new fixed focal length lenses

(3) regarding high ISO, the SLT mirror eats 1/2 f-stop of light, so the high ISO performance of the NEX-7 will be better by this 1/2 f-stop

I agree with Erik that we should be careful with the evaluation of the NEX-7 image quality as there are only very few pre-series cameras around and even the testers are just getting the very first results. However, there ARE a few shots from the NEX-7 out there (and I am not talking A77 samples!) and they surely look very, very promising.

Thomas

I'm not sure I understand, are you trying to say that its a breakthrough of performance? I assure you that if it was a thrilling performance camera, you would have got the word spread by now! The usual procedure is of press being enthusiastic for everything that hits the market on when it does hit the market, then it starts fading.... IMO (and others) Sony got right the pulse of the market that requires a superb HYBRID camera, that is a camera that shoots superb video and is up (or exceeds) the photo IQ of the average market (even a ...D3100 does that). On my view, 24mpx on an APS-c sensor was chosen simply because 1920x3=almost6000, that means a sensor of 6000x4000=24k for 3:2, thus you get a real 9 pixel definition for each one of the 1920x1080 unit of analysis. This means that a 24mpx sensor will have no interpolation to decode the info on 1920x1080, it will have better sharpness, it will have better color because you have plenty (9!) of R, G and B poled pixels for each of the 2K and you keep the "35mm cinema standard" feeling! In accordance you have an "only" good stills camera. Now thats not very bad for a magnificent videocamera that I bet both cameras (a77 & Nex7) will prove to be! I think it should be obvious for photographers to understand that if the pentaprism is abandoned from a DSLR, its because the camera has priorities for which its not necessary. What needs 2 EVFs is a videocamera not a photographers camera! ...and I can use a superb videocamera that can make good stills at that price! Cheers Theodoros, www.fotometria.gr

[memento]

No, I am not saying it is a breakthrough performance camera.

But I believe it has the potential to make similar-quality pictures (at the lower ISO) to the now current 35 mm DSLR cameras (Canon 5D2, Sony A900). Yes this of course is because of the 24 MP sensor and the apparently very weak OLP filter. Yes I see the potential of true 24 MP resolution with this camera in the first shots that have been made available on the internet – and these are all only JPG samples so far! It should make really great 20x30" prints or even 24x36". (I know how good such prints can look even with an old 12 MP DSLR, no question – but its easier to get them in better quality with a 24 MP camera!)

Now the 5D2 has the same resolution since 2008! So the NEX-7 cant be a breakthrough performance camera. Obviously.

But its much smaller than the 5D2, yes, including the lenses (the supposedly high quality 24/1.8 is only 225 grams, the 50/1.8 another 200 grams. Especially the 24 is much smaller and more light weight than any other similar spec lens for any DSLR!). Its a great size for a camera that you can still hold in your hand properly but on the other hand will make your bag cosiderably lighter, if you are travelling or hiking, for example.

Now there are other cameras that are already on the market that are even smaller and have less weight than this. So the NEX-7 again seems no breakthrough.

It's just the combination of very high quality (even though not the absolutely record-highest ever) and nice light weight that make it so appealing.

Oh and there is the electronic viewfinder. In contrast to the good old days, I have come to actually dislike optical SLR viewfinders in the recent time, to be honest. Also I dislike the concept of the phase detection autofocus with the constant hassle with fine adjustment and all its related problems. Truth is that todays SLR cameras' viewfinders (probably exept the very high end bodies, such as your D3x) are just not made with the same quality as it used to be. They are not so much designed for perfect manual focusing. This even more applies to the crop sensor DSLR crowd out there. With all my latest DSLR cameras, I have come to actually prefer the live view screen to the OVF whenever possible. The NEX just takes it to the logical next step, does away with any mirror, and shows the full live view image even in the viewfinder.

Of course all this sounds great, but still is subject to how the actual cameras will perform. We all have not really photographed with the NEX-7 so far, even though some of us had it in their hands. So, for now, I just take the word of those who have already worked with it a bit – such as Michael's first report at Luminous Landscape – that it really lives up to the expectations.

No wonder, of course! But still a really nice, modern and innovative camera with image quality for serious photography.

cheers,
Thomas

[fotometria gr]

No, I am not saying it is a breakthrough performance camera.

But I believe it has the potential to make similar-quality pictures (at the lower ISO) to the now current 35 mm DSLR cameras (Canon 5D2, Sony A900). Yes this of course is because of the 24 MP sensor and the apparently very weak OLP filter. Yes I see the potential of true 24 MP resolution with this camera in the first shots that have been made available on the internet – and these are all only JPG samples so far! It should make really great 20x30" prints or even 24x36". (I know how good such prints can look even with an old 12 MP DSLR, no question – but its easier to get them in better quality with a 24 MP camera!)

Now the 5D2 has the same resolution since 2008! So the NEX-7 cant be a breakthrough performance camera. Obviously.

But its much smaller than the 5D2, yes, including the lenses (the supposedly high quality 24/1.8 is only 225 grams, the 50/1.8 another 200 grams. Especially the 24 is much smaller and more light weight than any other similar spec lens for any DSLR!). Its a great size for a camera that you can still hold in your hand properly but on the other hand will make your bag cosiderably lighter, if you are travelling or hiking, for example.

Now there are other cameras that are already on the market that are even smaller and have less weight than this. So the NEX-7 again seems no breakthrough.

It's just the combination of very high quality (even though not the absolutely record-highest ever) and nice light weight that make it so appealing.

Oh and there is the electronic viewfinder. In contrast to the good old days, I have come to actually dislike optical SLR viewfinders in the recent time, to be honest. Also I dislike the concept of the phase detection autofocus with the constant hassle with fine adjustment and all its related problems. Truth is that todays SLR cameras' viewfinders (probably exept the very high end bodies, such as your D3x) are just not made with the same quality as it used to be. They are not so much designed for perfect manual focusing. This even more applies to the crop sensor DSLR crowd out there. With all my latest DSLR cameras, I have come to actually prefer the live view screen to the OVF whenever possible. The NEX just takes it to the logical next step, does away with any mirror, and shows the full live view image even in the viewfinder.

Of course all this sounds great, but still is subject to how the actual cameras will perform. We all have not really photographed with the NEX-7 so far, even though some of us had it in their hands. So, for now, I just take the word of those who have already worked with it a bit – such as Michael's first report at Luminous Landscape – that it really lives up to the expectations.

No wonder, of course! But still a really nice, modern and innovative camera with image quality for serious photography.

cheers,
Thomas

Similar quality prints to the A900 or 5D2? According to physics, the larger the pixel the less the induced noise and that is irrelevant to ISOs, those of us that print really large (much larger than you state) everyday, will assure you that "you print better from better pixel definition", if it was for res to guaranty a quality print, manufacturers would have exceeded 100mpx by now simply by multiplying the area of their smaller size sensors. My quote was not for a camera of "poor IQ" but rather for a sensor that can't be better from it's 16.2mpx counterparts on STILLS and I'm sure it will prove not to be, as well as that Sony judged the market needs correctly and produced a HYBRID camera that has superb low price video as a priority! Cheers Theodoros www.fotometria.gr
P.S: There are are many that think that the higher the dpi, the better the print. That would only be true if the recorded per pixel info was of the same quality (which is rarely the case), to give you an example the D700 prints a (200iso) 100x150cm image at 72dpi much better than a D7K at the same size, with the same ISO, although the later will be at 83dpi, with a 24mpx you would be at 100dpi for the same size, my suspicion is that the APS-c 24mpx will prove to be worst than both.

[ErikKaffehr]

Hi!

More complex:

- Larger sensor collects more photons. Less noise.
- Larger sensor would have higher MTF on a given feature size than a smaller sensor assuming similar lens
- A denser sensor will resolve more detail than a less dense sensor and it will interpolate better
- Pixel size is by large irrelevant for shot noise
- Design rules matter. A large part of the sensel area is taken up by wiring and transistors. Using narrower design rules increases physical fill factor.

A well designed 24 MP sensor will be somewhat better than a similarly well designed  16 MP sensor but will be at disadvantage to a well designed 24 MP sensor that is physically larger.

Finally, the strength of OLP (AA) filter is reduced with reduced pixel size. So a significant part of the MTF advantage with a large sensor is lost due to stronger AA-filtering.

Decreasing pixel size leads to a small loss in DR, but that can be compensated for by better readout noise and also by better physical fill factor.

Best regards
Erik

Similar quality prints to the A900 or 5D2? According to physics, the larger the pixel the less the induced noise and that is irrelevant to ISOs, those of us that print really large (much larger than you state) everyday, will assure you that "you print better from better pixel definition", if it was for res to guaranty a quality print, manufacturers would have exceeded 100mpx by now simply by multiplying the area of their smaller size sensors. My quote was not for a camera of "poor IQ" but rather for a sensor that can't be better from it's 16.2mpx counterparts on STILLS and I'm sure it will prove not to be, as well as that Sony judged the market needs correctly and produced a HYBRID camera that has superb low price video as a priority! Cheers Theodoros www.fotometria.gr
P.S: There are are many that think that the higher the dpi, the better the print. That would only be true if the recorded per pixel info was of the same quality (which is rarely the case), to give you an example the D700 prints a (200iso) 100x150cm image at 72dpi much better than a D7K at the same size, with the same ISO, although the later will be at 83dpi, with a 24mpx you would be at 100dpi for the same size, my suspicion is that the APS-c 24mpx will prove to be worst than both.

[fotometria gr]

Hi!

More complex:

- Larger sensor collects more photons. Less noise.
- Pixel size is by large irrelevant for shot noise
Decreasing pixel size leads to a small loss in DR, but that can be compensated for by better readout noise and also by better physical fill factor.

Best regards
Erik

Hi Erik, I believe that you have sentences that oppose each other up there, "Larger pixel size also collects more photons. ie: less noise". I mean the second is wrong for the same reasons that the first is right. Also... "loss in DR" is because of increased (per pixel) noise! Regards, Theodoros www.fotometria.gr

[ErikKaffehr]

Hi,

No contradiction! The number of photons collected is still the same. For shot noise it doesn't matter how many pixels the photons are distributed over.

The article below explains it very well: http://www.luminous-landscape.com/essays/dxomark_sensor_for_benchmarking_cameras.shtml

DR (in the technical sense) is affected by pixel size. Smaller pixels used to have smaller full well capacity, and DR (in the technical sense) is defined as FWC/read noise. So smaller pixels will have less DR, unless readout noise is reduced.

Best regards
Erik

Hi Erik, I believe that you have sentences that oppose each other up there, "Larger pixel size also collects more photons. ie: less noise". I mean the second is wrong for the same reasons that the first is right. Also... "loss in DR" is because of increased (per pixel) noise! Regards, Theodoros www.fotometria.gr

[fotometria gr]

Hi,

No contradiction! The number of photons collected is still the same. For shot noise it doesn't matter how many pixels the photons are distributed over.

The article below explains it very well: http://www.luminous-landscape.com/essays/dxomark_sensor_for_benchmarking_cameras.shtml

DR (in the technical sense) is affected by pixel size. Smaller pixels used to have smaller full well capacity, and DR (in the technical sense) is defined as FWC/read noise. So smaller pixels will have less DR, unless readout noise is reduced.

Best regards
Erik

I'm not sure my British BEng in MechEng allows me to agree with the above! A smaller pixel is always noisier than a larger one (provided that the well depth is the same), its because the smaller pixel has more depth in proportion since well entrance is narrower especially towards the edges of the frame where the photons have to enter the well at a much greater angle, the problem is even more evident with WA lenses that increase the photons bouncing angle even further! Cheers Theodoros. www.fotometria.gr

[ErikKaffehr]

Hi,

That is part of the reason they have micro lenses.

DxO-mark released new data on the Alpha 77. I enclose plots comparing it to the Alpha 55. The reason I choose that camera was that both have the same translucent mirror design, which steals some light.

So actual measurements show that your statement is not valid in this case.

An Australian publication had some Imatest data on the A77 and I would say it was quite comparable regarding resolution to an Alpha 850 (full frame) hey tested earlier. So it seems that the megapixels actually work. http://www.photoreview.com.au/reviews/cameraaccessories/sony-dt-1650mm-f28-ssm-lens.aspx

Best regards
Erik

I'm not sure my British BEng in MechEng allows me to agree with the above! A smaller pixel is always noisier than a larger one (provided that the well depth is the same), its because the smaller pixel has more depth in proportion since well entrance is narrower especially towards the edges of the frame where the photons have to enter the well at a much greater angle, the problem is even more evident with WA lenses that increase the photons bouncing angle even further! Cheers Theodoros. www.fotometria.gr

[fotometria gr]

Hi,

That is part of the reason they have micro lenses.

DxO-mark released new data on the Alpha 77. I enclose plots comparing it to the Alpha 55. The reason I choose that camera was that both have the same translucent mirror design, which steals some light.

So actual measurements show that your statement is not valid in this case.

An Australian publication had some Imatest data on the A77 and I would say it was quite comparable regarding resolution to an Alpha 850 (full frame) hey tested earlier. So it seems that the megapixels actually work. http://www.photoreview.com.au/reviews/cameraaccessories/sony-dt-1650mm-f28-ssm-lens.aspx

Best regards
Erik

Microlenses help but don't solve the problem. DXO is using its own converter to measure things, lets wait until we try it our shelves, I'm gonna buy the bloody thing for videocamera and of course try it in stills, mainly against D7K which I also own. According to DXO, my D7K has better DR than both my Fuji s5 or my D700, it even beats a P1 P45+!! I can assure you its far from achieving this, its nearly a stop down from D700, an additional one from the S5pro and definitely 2.5 stops worst than my Imacon 528c when used in single shot. Mind you that the 528c is not up to the P45+ (Which I've a lot of experience with) in single shot mode. Regards, Theodoros. www.fotometria.gr

[ErikKaffehr]

Hi,

Just a few comments. Marc McCalmont has written that his Pentax K5 has better  DR than his P45+. Now he has an IQ180 (Camera). http://www.luminous-landscape.com/forum/index.php?topic=55756.msg453739#msg453739

The definition that DxO uses for DR is FWC/read noise, or SNR=1. I'm not sure how relevant this figure is. High DR means that read noise is low.

The Fuji S5 has two set of sensels and does essentially something like on chip HDR. So it gains some advantage.

The Nikon D700 is not an APS-C camera, it's full frame, and it is built for high ISO performance. It's readout is noisy, therefore it gets less than very good DR rating at DxO.

According to DxO mark the DR of the D7000 is higher than that of D700 at low ISO. If you look at "tonal range", which is based on different SNR criteria you see that the D700 is better at all ISOs. The tonal range is about shot noise, and here the larger size of the sensor comes into play.

I never stated anything else than that shrinking pixel size, within reasonable limits will not increase noise. It will reduce DR, however. But DR (at SNR=1) does probably not matter that much to photography.

DxO measurements are not based on converted data, AFAIK, so they are not using their raw-converter. You can read raw data without conversion.

Look here, folks with higher exams than you or me (I happen to have an MSc in mechanical engineerting, too) have spent a lot of effort getting insight in how sensors work. Why don't you read up on the stuff?

I would start with this one: http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/index.html

Also, I would remind you that you started this discussion claiming that the new sensors would have extreme OLP (AA-filtering) for video, that this is not the case has been clearly shown by Michael Reichmann's image samples and also by the test in the Australian paper I referred to earlier. HD-Video is about two megapixels so that would require extreme AA-filtering. Sony has actually a NEX size sensor built for a videocamera with 15 micron pixels, but that has nothing to do with the NEX-7.


Best regards
Erik

Microlenses help but don't solve the problem. DXO is using its own converter to measure things, lets wait until we try it our shelves, I'm gonna buy the bloody thing for videocamera and of course try it in stills, mainly against D7K which I also own. According to DXO, my D7K has better DR than both my Fuji s5 or my D700, it even beats a P1 P45+!! I can assure you its far from achieving this, its nearly a stop down from D700, an additional one from the S5pro and definitely 2.5 stops worst than my Imacon 528c when used in single shot. Mind you that the 528c is not up to the P45+ (Which I've a lot of experience with) in single shot mode. Regards, Theodoros. www.fotometria.gr

[fotometria gr]

Hi,

Just a few comments. Marc McCalmont has written that his Pentax K5 has better  DR than his P45+. Now he has an IQ180 (Camera). http://www.luminous-landscape.com/forum/index.php?topic=55756.msg453739#msg453739

The definition that DxO uses for DR is FWC/read noise, or SNR=1. I'm not sure how relevant this figure is. High DR means that read noise is low.

The Fuji S5 has two set of sensels and does essentially something like on chip HDR. So it gains some advantage.

The Nikon D700 is not an APS-C camera, it's full frame, and it is built for high ISO performance. It's readout is noisy, therefore it gets less than very good DR rating at DxO.

According to DxO mark the DR of the D7000 is higher than that of D700 at low ISO. If you look at "tonal range", which is based on different SNR criteria you see that the D700 is better at all ISOs. The tonal range is about shot noise, and here the larger size of the sensor comes into play.

I never stated anything else than that shrinking pixel size, within reasonable limits will not increase noise. It will reduce DR, however. But DR (at SNR=1) does probably not matter that much to photography.

DxO measurements are not based on converted data, AFAIK, so they are not using their raw-converter. You can read raw data without conversion.

Look here, folks with higher exams than you or me (I happen to have an MSc in mechanical engineerting, too) have spent a lot of effort getting insight in how sensors work. Why don't you read up on the stuff?

I would start with this one: http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/index.html

Also, I would remind you that you started this discussion claiming that the new sensors would have extreme OLP (AA-filtering) for video, that this is not the case has been clearly shown by Michael Reichmann's image samples and also by the test in the Australian paper I referred to earlier. HD-Video is about two megapixels so that would require extreme AA-filtering. Sony has actually a NEX size sensor built for a videocamera with 15 micron pixels, but that has nothing to do with the NEX-7.


Best regards
Erik

I'm sorry to say that I just realized that there is a completely different approach here that leads nowhere! Its obviously my fault not realizing the different approach. In my view (which is a fundamental to my approach), a  photo is only the printed thing in paper, this view is the same with Adams, Kappa, Kertez, Bresson, Koudelka, Man Ray, or any other PHOTOGRAPHER that I know! Since I consider myself as a photographer, I print my own photos in a lab THAT I OWN AND RUN which is considered by other photographers that choose it from other labs, to be a top quality one by any standard worldwide! But since I am also a photographer and have the responsibility to teach photography to newcomers, I also have the responsibility to test most of the new tech that is induced in these "Archaic" days we live with digital equipment. My procedure when testing is based on the above fundamentals and Is the one I found most effective for photography with printed images, than newer theories that prove things on screens (not even monitors sometimes) among "photographers" that never print but have an opinion on everything! The procedure is: 1. Shoot Raw, 2. Convert Raw to 16bit Tiff with as much DR latitude as possible and with the OPTIMUM Raw conversion engine FOR THE RAW COMPRESSION LOGIC that the camera's sensor was designed for. 3. Process the tiff for max. contrast that will withstand full "highlight DR" when printing. 4. Adjust "levels" and "Gamma" in PS for optimum LowlightDR/noise character (personal matter, opinions are as many as A-Ho..es). 5. Print with absolute profile. 6. Compare prints only.
  To do this I shoot with 17 different cameras (not all that I own.., only the ones I use) and 162 different lenses (both analog and digital of all formats and media) and use a full color analog lab with color analyzer, Thermostatic devise, top enlarger ...etc, scanner at 4000dpi of 4.9 D-max which is used at least on 8x scanning, 10bit monitor, dual processor multicore 32 Gb ram 4x Graf.card computer (only describe our main machine), profile calibrator for all our devises, and fully linearized 112cm 11 pigment ink plotter (also use tonner and chemical lasers) that prints at 2880dpi! (As I say in my site we can have more than 98% profile accuracy for the first -test- print, even if we print gold or silver.)
  Based on the above tests, the results are as I quoted earlier, both in DR and noise! I obviously think that there are all shorts of crap spread around the web and I am really sorry that people "bite" on those, it seems that crooks pretending to be photographers without pictures is a good business in the "corrupted" times we live on. Cheers, Theodoros. www.fotometria.gr
P.S. I'd rather not continue the above conversation unless you quote your own test or experience and not refer "dogmatically" all the time, to things that you've read and that have created false impressions and theories on you. In fact let theory creators to defend their theories by themselves(!), Don't let them use you.
  Ah.. and another thing... "I never stated anything else than that shrinking pixel size, within reasonable limits will not increase noise" that you state up there IS A CHANGE IN YOUR ORIGINAL POSITION because "within reasonable limits" is absent in what you was stating before, also... 24mpx on an APS-c sensor is "beyond reasonable limits" by definition because its todays... "upper limit"!!!!   

[ErikKaffehr]

Hi,

Reasonable limits are simply not stretching technology beyond its capacity. For instance I'd say that going for 2 micron pixels may not be reasonable. The pixels would have very low full well capacity, or fill factor can sink below reasonable. My feeling is that going down to around 4 microns may be quite optimal with the technology we have.

Actually, I'm posting my findings on my website, http://echophoto.dnsalias.net/ekr/index.php/photoarticles . Obviously I have much smaller resources than you and also narrower interest.

This posting was originally about the possibility of a full frame NEX system which would be able to handle legacy lenses with an up to date sensor. There is a significant interest using the NEX series as an imager for legacy lenses, Leica M-lenses but perhaps also Leica-R, NEX-7 is not ideal for that because of the crop factor. You essentially stated that the image quality from the NEX7 will be crap, because it is mainly intended for video. Is this finding based on your testing?

Contrary to your views, I see nothing wrong with relying on other persons tests, as long as the tests are well documented.

Regarding DxO, they have built a large data base of sensor data. It's easy to draw erroneous conclusions from the data, but it seems that the data is well respected. For instance you state that your D7000 is better then your D700 according to DxO-mark, but that only applies to DR at low ISO. Tonal range is better on the D700 according to DxO. I'd really suggest that you check the paper by Emil Martinec on sensor noise, it explains the issues very well.

By the way, if you would take the time to check out the DxO-mark figures for the Sony Alpha 77 and the Sony Alpha 55 you would see that noise characteristics are very close at "print" setting, so there seems to be little disadvantage with the increased pixel count.

Best regards
Erik

I'm sorry to say that I just realized that there is a completely different approach here that leads nowhere! Its obviously my fault not realizing the different approach. In my view (which is a fundamental to my approach), a  photo is only the printed thing in paper, this view is the same with Adams, Kappa, Kertez, Bresson, Koudelka, Man Ray, or any other PHOTOGRAPHER that I know! Since I consider myself as a photographer, I print my own photos in a lab THAT I OWN AND RUN which is considered by other photographers that choose it from other labs, to be a top quality one by any standard worldwide! But since I am also a photographer and have the responsibility to teach photography to newcomers, I also have the responsibility to test most of the new tech that is induced in these "Archaic" days we live with digital equipment. My procedure when testing is based on the above fundamentals and Is the one I found most effective for photography with printed images, than newer theories that prove things on screens (not even monitors sometimes) among "photographers" that never print but have an opinion on everything! The procedure is: 1. Shoot Raw, 2. Convert Raw to 16bit Tiff with as much DR latitude as possible and with the OPTIMUM Raw conversion engine FOR THE RAW COMPRESSION LOGIC that the camera's sensor was designed for. 3. Process the tiff for max. contrast that will withstand full "highlight DR" when printing. 4. Adjust "levels" and "Gamma" in PS for optimum LowlightDR/noise character (personal matter, opinions are as many as A-Ho..es). 5. Print with absolute profile. 6. Compare prints only.
  To do this I shoot with 17 different cameras (not all that I own.., only the ones I use) and 162 different lenses (both analog and digital of all formats and media) and use a full color analog lab with color analyzer, Thermostatic devise, top enlarger ...etc, scanner at 4000dpi of 4.9 D-max which is used at least on 8x scanning, 10bit monitor, dual processor multicore 32 Gb ram 4x Graf.card computer (only describe our main machine), profile calibrator for all our devises, and fully linearized 112cm 11 pigment ink plotter (also use tonner and chemical lasers) that prints at 2880dpi! (As I say in my site we can have more than 98% profile accuracy for the first -test- print, even if we print gold or silver.)
  Based on the above tests, the results are as I quoted earlier, both in DR and noise! I obviously think that there are all shorts of crap spread around the web and I am really sorry that people "bite" on those, it seems that crooks pretending to be photographers without pictures is a good business in the "corrupted" times we live on. Cheers, Theodoros. www.fotometria.gr
P.S. I'd rather not continue the above conversation unless you quote your own test or experience and not refer "dogmatically" all the time, to things that you've read and that have created false impressions and theories on you. In fact let theory creators to defend their theories by themselves(!), Don't let them use you.
  Ah.. and another thing... "I never stated anything else than that shrinking pixel size, within reasonable limits will not increase noise" that you state up there IS A CHANGE IN YOUR ORIGINAL POSITION because "within reasonable limits" is absent in what you was stating before, also... 24mpx on an APS-c sensor is "beyond reasonable limits" by definition because its todays... "upper limit"!!!!   

[fotometria gr]

Hi,

Reasonable limits are simply not stretching technology beyond its capacity. For instance I'd say that going for 2 micron pixels may not be reasonable. The pixels would have very low full well capacity, or fill factor can sink below reasonable. My feeling is that going down to around 4 microns may be quite optimal with the technology we have.

Actually, I'm posting my findings on my website, http://echophoto.dnsalias.net/ekr/index.php/photoarticles . Obviously I have much smaller resources than you and also narrower interest.

This posting was originally about the possibility of a full frame NEX system which would be able to handle legacy lenses with an up to date sensor. There is a significant interest using the NEX series as an imager for legacy lenses, Leica M-lenses but perhaps also Leica-R, NEX-7 is not ideal for that because of the crop factor. You essentially stated that the image quality from the NEX7 will be crap, because it is mainly intended for video. Is this finding based on your testing?

Contrary to your views, I see nothing wrong with relying on other persons tests, as long as the tests are well documented.

Regarding DxO, they have built a large data base of sensor data. It's easy to draw erroneous conclusions from the data, but it seems that the data is well respected. For instance you state that your D7000 is better then your D700 according to DxO-mark, but that only applies to DR at low ISO. Tonal range is better on the D700 according to DxO. I'd really suggest that you check the paper by Emil Martinec on sensor noise, it explains the issues very well.

By the way, if you would take the time to check out the DxO-mark figures for the Sony Alpha 77 and the Sony Alpha 55 you would see that noise characteristics are very close at "print" setting, so there seems to be little disadvantage with the increased pixel count.

Best regards
Erik

I never said that NEX-7 or a77 will be "crap", I clearly state it will prove to be up to most people standard but lesser than its 16.2mpx counterparts and also state that its video will prove to be superb. I did base my prejudgement on the design of the (same performance) a77 which clearly is a design with video priorities especially in its focusing system and I stated that I will be buying the camera because I plan to replace my videocameras with APS-c sensor DSLRs to get the "cinema feeling" that the equivalent to 35mm cinema DOF will bring in my videos! I also explained why a 24mpx sensor is ideal for video. I have nothing to add on my view of the DXO data and why I wouldn't trust them. I won't comment to the 4 microns hatch of yours, its just that my hatch is different and all it remains is for time to tell..., so I suggested to wait for a few weeks and let the "dust settle". Finally, those of us that want to use our amazing glass on a NEX-7 via an adapter, its nothing stopping us to do so on the NEX-5 as well. OTOH do you really think that the target group of the NEX system is people with FF priorities?
 A. The system is clearly a hybrid video/stills system aimed for the consumer that wants things compact and thus Sony would never put an FF sensor on it because this would kill video.
 B. They would never produce a second line of lenses (with larger circle), because this would bring confusion to the less experienced of the target group and would to some extend affect sales. A second line of lenses would also increase the final product cost and thus shrink sales even further. Regards, Theodoros www.fotometria.gr
 P.S. Technically it can be done..... (of course)