Pages: [1]   Go Down

Author Topic: Proposed comparison between Leica M9 and Sony NEX7 (using existing images)  (Read 8892 times)

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto

Hi,

Michael Reichmann posted a comparison between the Leica M9 and the NEX7. I would suggest an alternative presentation of the images, that answers a couple of other questions.

There are two factors making these comparisons less straightforward. One is that the NEX7 is a cropped frame camera (at least compared with the Leica), the other is that the sensor on the NEX7 has significantly higher resolution. The comparison I would make would be:

Leave the NEX7 image as it is, but upscale the Leica image to 9180x6120 pixels. I'd suggest upscaling using either bicubic or bicubic smoother. Both images should be sharpened to taste. Actual pixel crops of both images could be shown side by side.

This evaluation of the test would illustrate a few things:

1) Does increased resolution compensate for the OLP-filtering?
2) Does the lens outresolve the M9 sensor?
3) What would the image quality of the Leica M9 be if it used a 54 MPixel CMOS sensor instead of a 18 MPixel CCD-sensor?

I have done a similar test using my Sony Alpha 55 SLT and Sony Alpha 900, which essentially predicts the performance of a 36 MP full frame camera. The answer in my case seems to be that a lot of difference can be seen on test chart, but less on the other parts of the test image. My test is here: http://echophoto.dnsalias.net/ekr/index.php/photoarticles/60-what-about-36mp-dslrs

I would also add that I am most thankful for Michael sharing his experience of equipment that most of us are only dreaming about.

Best regards
ERik
« Last Edit: December 15, 2011, 05:30:17 pm by ErikKaffehr »
Logged
Erik Kaffehr
 

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
Re: Proposed comparison between Leica M9 and Sony NEX7 (using existing images)
« Reply #1 on: December 15, 2011, 02:14:44 pm »

Hi,

Michael Reichmann posted a comparison between the Leica M9 and the NEX7. I would suggest an alternative presentation of the images, that answers a couple of other questions.

There are two factors making these comparisons less straightforward. One is that the NEX7 is a cropped frame camera (at least compared with the Leica), the other is that the sensor on the NEX7 has significantly higher resolution. The comparison I would make would be:

Leave the NEX7 image as it is, but upscale the Leica image to 9180x6120 pixels. I'd suggest upscaling using either bicubic or bicubic smoother. Both images should be sharpened to taste. Actual pixel crops of both images could be shown side by side.

This evaluation of the test would illustrate a few things:

1) Does increased resolution compensate for the OLP-filtering?
2) Does the lens outresolve the M9 sensor?
3) What would the image quality of the Leica M9 be if it used a 54 MPixel CMOS sensor instead of a 18 MPixel CCD-sensor?

How would you handle the different crop factors? It does not make sense to crop the 18MP Leica to 7.7MP so that the field of view is the same. The easiest way would to move the Leica back so that the field of view of the full frame image of the Leica is the same as the cropped frame Sony. Also what about depth of field and diffraction? One would have to use the same aperture diameter, not f/number, to equalize depth of field as outlined by Roger Clark.

Regards,

Bill
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Proposed comparison between Leica M9 and Sony NEX7 (using existing images)
« Reply #2 on: December 15, 2011, 02:25:18 pm »

Hi,

Read again...

I suggest that Leica image would be upscaled so it would match the APS-C image would it be full frame.

Best regards
Erik

How would you handle the different crop factors? It does not make sense to crop the 18MP Leica to 7.7MP so that the field of view is the same. The easiest way would to move the Leica back so that the field of view of the full frame image of the Leica is the same as the cropped frame Sony. Also what about depth of field and diffraction? One would have to use the same aperture diameter, not f/number, to equalize depth of field as outlined by Roger Clark.

Regards,

Bill
Logged
Erik Kaffehr
 

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/
Re: Proposed comparison between Leica M9 and Sony NEX7 (using existing images)
« Reply #3 on: December 15, 2011, 06:19:51 pm »

Using a 35mm lens of similar quality on the NEX-7 seems like a much more real world test, doesn't it?

In the end, photographers care about perspective and crop, so they will often use a 35mm on the NEX-7 to replicate the framing of a 50mm.

The very wide array of lenses available for the NEX platform is one of its best value, so it should be easy to find a good 35mm, shouldn't it?

Of course, there is no reason to think that the NEX-7 has sharper pixels than, say, an A900 so you might as well do the comparison between 2 FF cameras which will enable to use the exact same focal lenght, if not the same lens. There are many amazing 50mm lenses for all the FF cameras, even if the A platform is probably not the one with the widest offering.

It is very easy to convert a Leica R 50mm lens to F mount for instance.

Cheers,
Bernard

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Proposed comparison between Leica M9 and Sony NEX7 (using existing images)
« Reply #4 on: December 15, 2011, 06:34:04 pm »

Hi,

The test I suggest shows what the sensor is capable with a truly excellent lens, also what benefits are possible with a very high resolution sensor.

Actually, I'd think the test would be more interesting with the NEX5 as 16 MPixels on APS-C corresponds to 36 MP on full frame, and it seems full frame with 36 MP will be around next year.

Leica is rumored to have an AF camera with interchangable lenses in development, and I'd certainly hope it would be full frame and able to use existing M-lenses. And I would expect it to have a 36 MP CMOS sensor. That is of course only speculation, and some folks here at LuLa may know more under an NDA.

But I absolutely agree that for picture taking ability testing with 35 mm vs. 50 mm would be optimal.

Best regards
Erik


Using a 35mm lens of similar quality on the NEX-7 seems like a much more real world test, doesn't it?

In the end, photographers care about perspective and crop, so they will often use a 35mm on the NEX-7 to replicate the framing of a 50mm.

The very wide array of lenses available for the NEX platform is one of its best value, so it should be easy to find a good 35mm, shouldn't it?

Of course, there is no reason to think that the NEX-7 has sharper pixels than, say, an A900 so you might as well do the comparison between 2 FF cameras which will enable to use the exact same focal lenght, if not the same lens. There are many amazing 50mm lenses for all the FF cameras, even if the A platform is probably not the one with the widest offering.

It is very easy to convert a Leica R 50mm lens to F mount for instance.

Cheers,
Bernard

Logged
Erik Kaffehr
 

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
Re: Proposed comparison between Leica M9 and Sony NEX7 (using existing images)
« Reply #5 on: December 15, 2011, 06:40:05 pm »

Hi,

Read again...

I suggest that Leica image would be upscaled so it would match the APS-C image would it be full frame.

I did read again, but no matter what you do with the Leica image, it will be at a severe disadvantage if only a portion of the pixels (those corresponding to the area of the Nex 7 are used.

Regards,

Bill
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
AQuick comparison using DPReview images
« Reply #6 on: December 15, 2011, 07:34:15 pm »

Hi!

I have downloaded RAW samples from DPReview, loaded into ACR at default settings, and resized the M9 image to same horizontal pixels as NEX7, using bicubic.

The screen dumps are at actual pixels, NEX7 to the left and Leica to the right. The screen dump shows the images as imported. On the second one I applied some sharpening on the Leica to compensate for sharpness lost in upsizing.

According to EXIF the Leica image was shot at f/22. I don't know if EXIF information is to be trusted. Does the Leica M9 record shooting aperture?! Closing down to F/22 would definitively affect sharpness negatively. If the comparison image really was shot at f/22 DPReview should not have publish it!

Best regards
Erik
Logged
Erik Kaffehr
 

Johnny_Johnson

  • Guest
Re: Proposed comparison between Leica M9 and Sony NEX7 (using existing images)
« Reply #7 on: December 15, 2011, 10:32:53 pm »

Hi Erik, the M9 doesn't record the shooting aperture. It's a guesstimate that's made by the Adobe software that is sometimes reasonably close but at other time very inaccurate.

Cya,
Johnny
Logged

torger

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3267
Re: AQuick comparison using DPReview images
« Reply #8 on: December 16, 2011, 03:23:36 am »

Hi!

I have downloaded RAW samples from DPReview, loaded into ACR at default settings, and resized the M9 image to same horizontal pixels as NEX7, using bicubic.

The screen dumps are at actual pixels, NEX7 to the left and Leica to the right. The screen dump shows the images as imported. On the second one I applied some sharpening on the Leica to compensate for sharpness lost in upsizing.

According to EXIF the Leica image was shot at f/22. I don't know if EXIF information is to be trusted. Does the Leica M9 record shooting aperture?! Closing down to F/22 would definitively affect sharpness negatively. If the comparison image really was shot at f/22 DPReview should not have publish it!

One problem with the DPReview studio shot is that it is not particularly made for sharpness test, meaning that focusing is quite often not very good, and one don't really know where they have put the focus. At least so it seems after viewing multiple of these.

Anyway, I would be surprised if the result is not that NEX7 has less sharp pixels but higher total resolution anyway, which your tests indicate.
Logged

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/
Re: AQuick comparison using DPReview images
« Reply #9 on: December 16, 2011, 09:25:11 am »

Anyway, I would be surprised if the result is not that NEX7 has less sharp pixels but higher total resolution anyway, which your tests indicate.

Again, if that it the case... what were all these past claims about the superiority of the M9 compared to any other DSLR?

18mp vs 24mp is a tiny difference. We have been told again and again on LL that the lack of AA filter makes a huge difference in resolution... and now we are saying that it would be no more than 10%...

Cheers,
Bernard

torger

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3267
Re: AQuick comparison using DPReview images
« Reply #10 on: December 16, 2011, 09:44:19 am »

Again, if that it the case... what were all these past claims about the superiority of the M9 compared to any other DSLR?

18mp vs 24mp is a tiny difference. We have been told again and again on LL that the lack of AA filter makes a huge difference in resolution... and now we are saying that it would be no more than 10%...

I may be on deep water here, but I guess that the harder you press lens resolving power and diffraction, the less reduction of resolution due to the AA filter. The AA filter discussion started when cameras were quite low res, and then you lost more resolution than you do on a modern high res sensor, but people still think about how it was back in the days. I haven't really verified if this is true though.

I read a bit too quick I thought the Leica was 14 megapixels, not 18... anyway, bicubic scaling does smear things though, so any picture that is scaled gets a disadvantage. If one intends to scale up, I think it is better with more megapixels and AA filter so you get a fairly soft scaling-friendly file. Low res files without AA filter (=aliased) are the worst for scaling I think.

I did not know that people considered Leica to be superior in terms of sensor... I thought Leica was more about range-finder romance and some interesting lenses, like the 50mm f/0.95 :-). I've also heard that it has high quality lenses in terms of resolution etc, but I have not seen any formal tests so I don't know how they compare to general DSLRs. In theory the wide angle lenses could be better thanks to the lack of mirror.
« Last Edit: December 16, 2011, 09:52:20 am by torger »
Logged

indusphoto

  • Full Member
  • ***
  • Offline Offline
  • Posts: 132
    • 500px
Re: AQuick comparison using DPReview images
« Reply #11 on: December 16, 2011, 02:07:00 pm »

I may be on deep water here, but I guess that the harder you press lens resolving power and diffraction, the less reduction of resolution due to the AA filter. The AA filter discussion started when cameras were quite low res, and then you lost more resolution than you do on a modern high res sensor, but people still think about how it was back in the days. I haven't really verified if this is true though.

True. The AA filter is there to reduce artifacts when the pattern density is more than half the pixel density. (Digitization of a signal should be at least twice the frequency of signal, when signal frequency exceeds that, accurate reconstruction of signal is not possible). So as the pixel density (horizontal lines and vertical lines) have increased over the years, AA filter have also kept pace meaning they allow higher and higher frequency to get in (in other words they have become weaker). So the AA filter for NEX-7 24MP sensor is fairly weak and not really limiting real life patterns. The M-9 does not really have any advantage over NEX-7 in this regard, at least nothing significant.

The other thing is that it is well known that down sampled images have much better acuity and thus appear sharper. It is really very simple. If you take four fuzzy dots going from light to dark, and sample them down to two dots with the same DR, you will have two dots left with much higher contrast. So while up-sampling of M-9 images would degrade sharpness, down-sampling Sony image actually increased apparent sharpness.

So no matter how you look at it M-9 sensor is at a disadvantage and NEX-7 will look sharper. And it should! What is the point of having more pixels if you can not resolve more details! Sure NEX-7 is APS-C but it is not pinhead sized cell phone sensor. So it should resolve more given a good lens, and it did, otherwise Sony had failed.

Anybody noticed that there was a Leica lens in front of the sensor! Try putting a Sony lens in front!

P.S. I am not a Leical fanboy, never had one, never will. I think Canon rules!
« Last Edit: December 16, 2011, 02:53:58 pm by sunnycal »
Logged

madmanchan

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2115
    • Web
Re: Proposed comparison between Leica M9 and Sony NEX7 (using existing images)
« Reply #12 on: December 17, 2011, 11:51:03 am »

Hi Erik, the M9 doesn't record the shooting aperture. It's a guesstimate that's made by the Adobe software that is sometimes reasonably close but at other time very inaccurate.

Minor correction:  the guesstimate is made by the camera, not the Adobe software.  More specifically, the guesstimate is written into the image metadata by the camera, and the Adobe software simply reports the value stored in that metadata.
Logged
Eric Chan

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: AQuick comparison using DPReview images
« Reply #13 on: December 17, 2011, 04:32:37 pm »

Hi,

The DPReview shots with the Leica M9 are soft (not very sharp). Either they have *** up badly or it is for real. Erwin Puts'a samples are similar to the DPReview shots, but he describes his methods and he did use focus bracketing.

I would like to have a Leica M9 to test, so I look at the few published samples that are usable.

BTW, the difference between Nikon D3X and Sony Alpha 900 in Erwin's tests is remarkable. It may bee that Sony has much thicker OLP filter than Nikon? Or Nikon simply has better lenses?

Best regards
Erik

Again, if that it the case... what were all these past claims about the superiority of the M9 compared to any other DSLR?

18mp vs 24mp is a tiny difference. We have been told again and again on LL that the lack of AA filter makes a huge difference in resolution... and now we are saying that it would be no more than 10%...

Cheers,
Bernard

Logged
Erik Kaffehr
 

uaiomex

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1211
    • http://www.eduardocervantes.com
Re: Proposed comparison between Leica M9 and Sony NEX7 (using existing images)
« Reply #14 on: December 18, 2011, 03:33:00 pm »

Historically, the fairest encountering for two parties to meet has always been at some point near the middle between the two. Both depart the same distant from their comfort zones. Shouldn't we apply this same principle comparing sensors of different pixel number when doing it on a monitor?. Since these kind of comparisons usually end here? I think that equaling the number of pixels in this fashion should be the way to start. I'm aware that there are many kinds of comparisons and many intentions to do them.

Anyway, as I understand things, the only just way to compare Xcamera IQ to Ycamera IQ should be done by printing. Maybe 3 sizes should suffice. Those that are the most printed by photographers. Say, 8X10, 16X20 and 24X30. I mean full size prints, not crops (crops are in a way like pixel peeping. You need the full monty to perceive the real quality obtained from a determined camera). Better if inspected by different people to draw conclusions. Since print scanning doesn't require an expensive drum scanner, the reviewer could provide those scans to illustrate and perhaps strength his findings, perceptions and points of view.

The other decisive comparison would be to take this to offset printing but that's not viable.

I appreciate dearly anybody's effort to do these comparisons. They are helpful and enlightening always. They help a lot to construct our OWN final opinions, which I think it's the real intention. Please never stop doing them. But in my most humble opinion, these efforts are at their best as pixel peeping entertainment if there are no real life prints involved in the procedure.
Eduardo
 
« Last Edit: December 18, 2011, 03:58:44 pm by uaiomex »
Logged

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/
Re: AQuick comparison using DPReview images
« Reply #15 on: December 18, 2011, 06:09:30 pm »

The DPReview shots with the Leica M9 are soft (not very sharp). Either they have *** up badly or it is for real. Erwin Puts'a samples are similar to the DPReview shots, but he describes his methods and he did use focus bracketing.

I would like to have a Leica M9 to test, so I look at the few published samples that are usable.

BTW, the difference between Nikon D3X and Sony Alpha 900 in Erwin's tests is remarkable. It may bee that Sony has much thicker OLP filter than Nikon? Or Nikon simply has better lenses?

Any chance that the accuracy of focus, or lack thereof, might explain part of the difference?

We have known for a long time that the difference between a perfect live view tuned focus and an approximate one obtained by the eye can rob a sensor from a large chunck of its resolution potential.

Cheers,
Bernard

jameskatt

  • Newbie
  • *
  • Offline Offline
  • Posts: 5
Real world comparison between Full-frame and APS-c Sensors
« Reply #16 on: December 18, 2011, 08:10:58 pm »

If Michael was going to do a "real world" comparison between an APS-c sensor camera and a Full-frame sensor camera, then he should have used the appropriate lens on the APS-c camera that gives an equivalent point of view to the lens on the full-frame camera.

Then the comparison would be useful.

For example, if you use the Leica M9 with a 50 mm lens, then you have to use the Sony NEX7 with a 35 mm lens.  This would make it an Apples vs. Apples comparison.

Any complaints about how 35 mm lenses are poor compared to 50 mm lenses are mooted because this is a "real world" comparison.  Small sensor cameras have to use a wider lens to get the same field of view as a full-frame sensor camera.  Full-frame sensor cameras have to use larger lenses to get the same field of view as APS-c sensor cameras.  Each sensor-size comes with various compromises and strengths.

Logged

torger

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3267
Re: Real world comparison between Full-frame and APS-c Sensors
« Reply #17 on: December 19, 2011, 03:28:00 am »

Any complaints about how 35 mm lenses are poor compared to 50 mm lenses are mooted because this is a "real world" comparison.  Small sensor cameras have to use a wider lens to get the same field of view as a full-frame sensor camera.  Full-frame sensor cameras have to use larger lenses to get the same field of view as APS-c sensor cameras.  Each sensor-size comes with various compromises and strengths.

I agree. The reason for choosing one system above an other in terms of image quality I think is as much related to available lenses as to the sensor properties. If the lenses suck for the fields of view and apertures you need, a superior sensor is not much worth. With the mirrorless cameras the adapter options are usually quite good though, but I guess a basic comparison would use Leica lenses for the Leica system and Sony lenses for the Sony, and if testing for landscape be sure to test some wide angles, a 24mm for the Leica and thus 24/1.5 = 16mm for the Sony.

I pretty much chose Canon fullframe as landscape platform just because of TS-E 24mm II.
« Last Edit: December 19, 2011, 03:40:17 am by torger »
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: AQuick comparison using DPReview images
« Reply #18 on: December 19, 2011, 04:37:34 am »

Hi,

Erwins test was very carefully done, with full focus bracket, so he must have nailed focus.

Erik
Any chance that the accuracy of focus, or lack thereof, might explain part of the difference?

We have known for a long time that the difference between a perfect live view tuned focus and an approximate one obtained by the eye can rob a sensor from a large chunck of its resolution potential.

Cheers,
Bernard

Logged
Erik Kaffehr
 

jameskatt

  • Newbie
  • *
  • Offline Offline
  • Posts: 5
Re: Proposed comparison between Leica M9 and Sony NEX7 (using existing images)
« Reply #19 on: December 20, 2011, 12:02:28 am »

Michael did the "real world" comparison as I have requested.  GREAT!

What I see is that at the BASE ISO, the NEX7 does have slightly better resolution than the M9.

This is similar to Michael's comparison of the 15 megapixel Canon G10 compact camera at ISO 50 nearly equaling the performance of 30 megapixel medium format cameras years ago.

Of course, any camera at the base ISO will perform excellently with very little noise and with full resolution capture. This is what we are seeing with the NEX7.

It certainly argues that MORE PIXELS are better when you get to use low ISOs like in landscape and slow people photography.  It would be like the old days of film when ISO 100 had the best resolution for practical use.

The full-frame sensor with its larger pixels and greater light gathering potential offers the possibility of higher resolution at higher ISOs - where the larger pixels are capable of maintaining resolution despite the higher noise. Small sensor cameras always are at a disadvantage.

Thus a comparison at high ISO would be nice between the NEX7 and M9.

In any case, I hope Canon sees the data and realize that high resolution small frame cameras are also very very useful.

I would love to see Canon also do 24-30 megapixel sensors in an APS sensor camera.  For landscape work, they would make ideal travel cameras.

I would love to see Canon do 60 megapixel full-frame sensors.  There would be less clamor for medium format if this was done.  Certainly for studio work and well-lighted photos, these cameras would absolutely give gorgeous photos since high ISO wouldn't be needed.
Logged
Pages: [1]   Go Up