Ad
Ad
Ad
Pages: [1] 2 »   Bottom of Page
Print
Author Topic: NEX 7 downsize compared with M9  (Read 7990 times)
raining2011
Newbie
*
Offline Offline

Posts: 1


« on: December 14, 2011, 06:02:35 PM »
ReplyReply

In the latest NEX 7 review, the NEX 7 photo is downsized to match the image size of M9 and draw the conclusion NEX 7 is much sharper than M9. I doubt this method. In computer science, there are actually many different methods to downsize a image, but most of these methods actually sharpen the image. Though NEX 7 photo is not sharpened when converting RAW, but the downsize process is actually a "sharpening" and downsize process. It is not a fully reasonable comparison due to downsize method's built-in "sharpening". And the author also crop M9's image (I assume the M9 image size after crop is 18MP/1.5=12MP, so 24MP downsized compared with 12MP ?).  It can only show the advantage of higher pixels, the advantage of 24MP vs 18MP, may not show NEX 7 is much sharper than M9.
Logged
theguywitha645d
Sr. Member
****
Offline Offline

Posts: 970


« Reply #1 on: December 14, 2011, 08:34:30 PM »
ReplyReply

If you go to DPreview you can compare the Nex 7 to the M9, A900, and a host of other cameras.

While I don't agree with Michael's methodology, on any internet forum populated by experts, he is bound to please no one. Since he detailed his methodology, in that context, the results show an aspect of how these cameras work. You can also go to DPreview...
Logged
ErikKaffehr
Sr. Member
****
Online Online

Posts: 7252


WWW
« Reply #2 on: December 14, 2011, 11:06:44 PM »
ReplyReply

Hi,

According to EXIF data the M9 image is shot at f/22. The image seems to be quite soft, which is consistent with f/22. On the other hand the image shows some Moiré which would not be expected at f/22. I don't know if Leica M9 records shooting aperture.

Thanks to the pointer anyway!

The raw images can be downloaded, by the way.

Best regards
Erik




If you go to DPreview you can compare the Nex 7 to the M9, A900, and a host of other cameras.

While I don't agree with Michael's methodology, on any internet forum populated by experts, he is bound to please no one. Since he detailed his methodology, in that context, the results show an aspect of how these cameras work. You can also go to DPreview...
Logged

BernardLanguillier
Sr. Member
****
Offline Offline

Posts: 7770



WWW
« Reply #3 on: December 15, 2011, 04:35:01 AM »
ReplyReply

How does that related to past claims on this site than the M9 is significantly sharper than a D3x/1Ds3/A900?

http://www.luminous-landscape.com/essays/experiment.shtml

In that article, Michael uses the wording "faith" to describe the claims apparently made by some DSLR owners that their cameras might deliver images as sharp as those of the Leica M9.

Cheers,
Bernard
Logged

A few images online here!
georgl
Full Member
***
Offline Offline

Posts: 140


« Reply #4 on: December 15, 2011, 06:38:53 AM »
ReplyReply

He did not compare the actual resolution of the M9 and Nex-7 but the resolving power of the sensor pixel-pitch. Does a pixel-pitch of 4 micron ( +AA-filter) actually create more detail than 6.8 micron with a sharp lens? The downsampling just adds to the confusion.
He did not compensate sensor size (by changing position or using a 75mm on the M9), he shows what an hypothetical full-frame Nex-7 would achieve with a 54MP-sensor downsampled to 40MP image size.

Yes, resolving power could be increased by using oversampling from a smaller pixel-pitch -but I'm not sure about the relevance within this test!?
« Last Edit: December 15, 2011, 06:41:47 AM by georgl » Logged
SteveGJ
Newbie
*
Offline Offline

Posts: 1


« Reply #5 on: December 15, 2011, 07:10:03 AM »
ReplyReply

The LL test is not a fair reflection of what the Leica can do at full-frame. By producing a common crop size for the two from the same viewpoint, it’s essentially comparing a cropped Leica with the NEX, not an FF with APS-C. In effect, a 24MP APS-C camera (albeit with downsizing to 18MP) and an 8MP APS-C crop of a full frame. (Even if it's only common FoV sub-crops which are compared, that ratio of 3:1 on "native pixels" and 2.25:1 on "downsized pixels" still holds). It is hardly surprising that the NEX then out-resolves the Leica as it's effectively just comparing the resolution at the sensor which (if the lens is good enough) is bound to favour the NEX. The Leica is simply using less than 45% of the total pixels in the frame (or about 67% of the line-pairs across the whole frame).

You cannot perform a proper test of two different lens format sizes without doing one of two things. If you must use the same prime lens, you have to move the small sensor camera further away to match the FoV of the larger sensor. This can’t be perfect, as perspective will change, but for two-dimensional subjects (where the subject plan is parallel to the camera’s focal plane), or for subjects far enough away that the depth isn’t an issue, it allows for comparing resolution. Note that it is theoretically necessary to alter the aperture as well to equalise the DoF, but this introduces other lens issues – however, if resolution is compared at the common point of focus on both, then this is not an issue. To avoid any diffraction limiting on the NEX-7, an aperture of F8 or wider must be used.

However, if you want a full comparison of the two different frames, you have to take two photos from the same place with two different lenses with focal lengths chosen to give matching FoVs. Also, the aperture on the smaller sensor has to be wider to equalise the DoF. Of course, using two different lenses introduces another, major, variable, but it’s the only way of truly comparing two images taken with cameras of different sensor sizes with the same perspective.

Ideally the test would need to be repeated using a 50mm lens on the Leica with a 35mm on the NEX-7 of similar quality (with a little bit of shuffling of the tripod position to equalise the two FoVs). The NEX image should be shot at (say) f5.6 with the Leica at around f8. I would then compare an NEX-7 image downsized to 18MP with the native Leica. However, if you want to compare with the NEX at its maximum resolution, you'll need to up-scale the M9 image to 24MP.

Note, I don't own either of these cameras (albeit I have a Sony A700), so I've no axe to grind in favour of the red-spotted one, but I do think the result doesn't reflect the natural advantage of the M9's larger sensor size.
Logged
theguywitha645d
Sr. Member
****
Offline Offline

Posts: 970


« Reply #6 on: December 15, 2011, 08:26:21 AM »
ReplyReply

I say we push Michael into a pool filled with Jello and won't let him out until he gives us what we want. Or at least dock his pay.
Logged
theguywitha645d
Sr. Member
****
Offline Offline

Posts: 970


« Reply #7 on: December 15, 2011, 08:44:09 AM »
ReplyReply

Hi,

According to EXIF data the M9 image is shot at f/22. The image seems to be quite soft, which is consistent with f/22. On the other hand the image shows some Moiré which would not be expected at f/22. I don't know if Leica M9 records shooting aperture.

Thanks to the pointer anyway!

The raw images can be downloaded, by the way.

Best regards
Erik





No test is going to please everyone. There is never going to be a level playing field in that an absolute reference is going to be established--we either need a 18MP Nex 7 or a 24MP M9 (or do you equalize the image plane or object plane), at least to please all the possible detractors--just look at the nit picking with the MFD/8x10 review. But, I think if you have been around images for a long time, like your self, you can look at images taken with slightly different conditions and extrapolate an answer. I think Michael's test shows both cameras are very good. I like the DPreview tests as you can see the affects of noise and look at how different targets appear with each system. As as general test, I prefer the DPreview methodology.

If you want to calculate system MTF or such, then another test will be needed. I doubt the conclusion that both the M9 and Nex 7 are good cameras will change.

I feel a certain population of photographers look at these tests to see if the new kid "beat" their favorite and they would lose some sort of stature by using an "inferior" product. To a point, we are just trying to count how many angles can dance on the head of a pin.
Logged
LGeb
Newbie
*
Offline Offline

Posts: 17


« Reply #8 on: December 15, 2011, 09:13:13 AM »
ReplyReply

This test set out to compare resolution of the same Leica lens on two different cameras. We can see the true resolution of the Leica sensor since it's image was only cropped. By downsampling the Sony we can see it is sharper than the Leica, but we can't see how much sharper. It is good to know the lens can out resolve the Leica sensor.

But I fail to see how this test gives any real world guidance. I would neither set out to crop the Leica shot, not downsample the Sony when planning an image. Choosing a longer lens on the Leica, or a wider lens on the Sony and comparing them would tell you more about the images you could get out of either system, which is what I think is important.
Logged
Dennishh
Full Member
***
Offline Offline

Posts: 151


« Reply #9 on: December 15, 2011, 05:43:54 PM »
ReplyReply

I'll bet Michael knows now what Galileo felt like when he proposed that the Earth revolves around the Sun. Maybe the Vatican can figure which camera has more resolution. I'm sure that this debate will live on for years and I for one am enjoying it very much. For some reason I feel that this can only make camera manufacturers more aware of what us as photographers want.
Logged
ErikKaffehr
Sr. Member
****
Online Online

Posts: 7252


WWW
« Reply #10 on: December 15, 2011, 06:36:33 PM »
ReplyReply

http://www.luminous-landscape.com/forum/index.php?topic=60329.msg486308#msg486308

No test is going to please everyone. There is never going to be a level playing field in that an absolute reference is going to be established--we either need a 18MP Nex 7 or a 24MP M9 (or do you equalize the image plane or object plane), at least to please all the possible detractors--just look at the nit picking with the MFD/8x10 review. But, I think if you have been around images for a long time, like your self, you can look at images taken with slightly different conditions and extrapolate an answer. I think Michael's test shows both cameras are very good. I like the DPreview tests as you can see the affects of noise and look at how different targets appear with each system. As as general test, I prefer the DPreview methodology.

If you want to calculate system MTF or such, then another test will be needed. I doubt the conclusion that both the M9 and Nex 7 are good cameras will change.

I feel a certain population of photographers look at these tests to see if the new kid "beat" their favorite and they would lose some sort of stature by using an "inferior" product. To a point, we are just trying to count how many angles can dance on the head of a pin.
Logged

Fine_Art
Sr. Member
****
Offline Offline

Posts: 1063


« Reply #11 on: December 17, 2011, 12:02:22 AM »
ReplyReply

As a Sony user I think this test puts the NEX 7 at an unnecessary advantage. Of course a 24MP sensor will out resolve an 18MP sensor given the lens can handle it.

I good way to normalize the test would be to take shots of an equal frame such as a large oil painting. How will the detail compare? I expect the nex to have more detail. I expect the M9 CCD to have better color accuracy.

I think Leica may be better served by increasing lens capacity again rather than trying to compete on cameras.
Logged
ErikKaffehr
Sr. Member
****
Online Online

Posts: 7252


WWW
« Reply #12 on: December 17, 2011, 12:19:18 AM »
ReplyReply

Hi,

CCD and CMOS are color blind. There is a filter array in front of the sensor that creates color. The color arises in post processing. The Color Grid Array can have different characteristics. So it may be possible that Leica may have better color, but it has absolutely nothing to do with semiconductor technology chosen.

http://en.wikipedia.org/wiki/Bayer_filter

If you would shoot a repro of an oil painting you would need a much different rendering than in ordinary photography. You would use scene referred rendering, which means that the aim is to reproduce color. Color rendition is normally tweaked to give more pleasant color. That tweaking involves adding an "S-curve", increasing saturation and so on.

Regarding what Leica should do is something that Leica needs to decide. But keep in mind that Leica is a small firm. At present there is only one camera that can utilize their lenses fully and that is the M9. Leica could of course develop lenses for APS-C size sensors.


Best regards
Erik

As a Sony user I think this test puts the NEX 7 at an unnecessary advantage. Of course a 24MP sensor will out resolve an 18MP sensor given the lens can handle it.

I good way to normalize the test would be to take shots of an equal frame such as a large oil painting. How will the detail compare? I expect the nex to have more detail. I expect the M9 CCD to have better color accuracy.

I think Leica may be better served by increasing lens capacity again rather than trying to compete on cameras.
« Last Edit: December 17, 2011, 03:48:48 AM by ErikKaffehr » Logged

BernardLanguillier
Sr. Member
****
Offline Offline

Posts: 7770



WWW
« Reply #13 on: December 17, 2011, 01:30:57 AM »
ReplyReply

I'll bet Michael knows now what Galileo felt like when he proposed that the Earth revolves around the Sun.

Are you saying that Galileo did lay claims 2 years before his condemnation that the sun revolves around the earth? :-)

Cheers,
Bernard
Logged

A few images online here!
mas55101
Jr. Member
**
Offline Offline

Posts: 77


« Reply #14 on: December 19, 2011, 06:49:43 PM »
ReplyReply

Now that the second and, I assume, final test has been done, and we all know how fine the sensor/processing engine are on the Sony, I would truly like to see a comparison article with the Sony and Panny Gx1.  My main concerns are finder quality, low light/high asa image quality, and manual focusing.  I think at least some of us will make a choice between the NeX7 and Gx1.

Thanks much.
Logged
jameskatt
Newbie
*
Offline Offline

Posts: 5


« Reply #15 on: December 19, 2011, 11:09:41 PM »
ReplyReply

I am fully satisfied by Michael's new methodology.  It is REAL WORLD.

And it doesn't show anything new.

The NEX7 at BASE ISO has higher resolution than the M9.  Period.

At BASE ISO, any camera - even a compact camera like the Canon G10 - will have essentially noise-free photos. Thus, these will show the full resolution capability of the camera.

Recall that years ago, Michael showed experts photos from the 15 megapixel Canon G10 at 50 ISO and photos from a medium format  30 megapixel digital camera.  They couldn't tell the two apart.

The comparison at the lowest ISO of either camera will give you their full resolution.  Of course, a priori, you already know that the NEX7 will have higher resolution than the M9 in such a comparison.  Michael did not even have to do the test.

It would have been interesting to see which had higher resolution at ISO 3200.

Then, the larger frame sensor of the M9 would have less noise than the NEX7.  And thus, the tables may be turned in favor of the M9.

For photos at higher ISO, the M9 would win over the NEX7 since sensor noise becomes much more prominent an issue.
For low ISO photos, where one has time and image stabilization, then the NEX7 Wins since noise is not as much of an issue.

As Michael said:  Horses for courses.
Logged
ErikKaffehr
Sr. Member
****
Online Online

Posts: 7252


WWW
« Reply #16 on: December 19, 2011, 11:43:36 PM »
ReplyReply

Hi,

One of the issues with M9 is that the sensor is noisy. It doesn't even have ISO 3200.

By the way, it is not as simple as you think. Resolution is a combination of resolution of lens and sensor. The NEX-7 has small pixels and puts much higher demands on the lens.

Best regards
Erik


It would have been interesting to see which had higher resolution at ISO 3200.

Then, the larger frame sensor of the M9 would have less noise than the NEX7.  And thus, the tables may be turned in favor of the M9.

For photos at higher ISO, the M9 would win over the NEX7 since sensor noise becomes much more prominent an issue.
For low ISO photos, where one has time and image stabilization, then the NEX7 Wins since noise is not as much of an issue.

As Michael said:  Horses for courses.
Logged

jameskatt
Newbie
*
Offline Offline

Posts: 5


« Reply #17 on: December 20, 2011, 12:01:35 AM »
ReplyReply

Of course, it isn't as simple a comparison.

But then, I only want to see "real world", "practical" comparison.

Michael did the right thing in re-doing the test.

Sure, the NEX7 puts higher demand on lenses since it has higher resolution.  But then it uses the part of the lens which has the best characteristic - the center - just like the Canon 7D's 18 megapixels uses the best part of the lens compared to the 5D.  This is why the 7D has "higher" resolution than the 5D per given lens.

Again the point it moot.  What Michael did on the re-do is real-world.

Too bad the M9 has a noisy sensor.  Leica fans should be throwing their hands up in disgust.  But then it is what it is - a noisy camera. So if it loses to the NEX7 when it comes to taking photos, so be it. It only has itself to blame - i.e. Leica.

With the second test, Michael shows the NEX7 is better than the Leica for low ISO photos.  Given how practical it is because of its size, that is a good thing.  Perhaps Leica should then get a better sensor to go with their $9000 camera body.  Sony makes great sensors for other companies, like Nikon.  Perhaps Leica can get a sensor from Sony.

Hi,

One of the issues with M9 is that the sensor is noisy. It doesn't even have ISO 3200.

By the way, it is not as simple as you think. Resolution is a combination of resolution of lens and sensor. The NEX-7 has small pixels and puts much higher demands on the lens.

Best regards
Erik

Logged
ErikKaffehr
Sr. Member
****
Online Online

Posts: 7252


WWW
« Reply #18 on: December 20, 2011, 12:08:06 AM »
ReplyReply

Hi,

Here is a comparison made at DPReview. Leica only goes to 2500 ISO, so I made comparison at 1600. I'd say that Nikon D3X is champ in this arena but it was missing in DPReview the comparison tool. The Leica test image from DPReview does not really impress at any ISO.

Erik


Of course, it isn't as simple a comparison.

But then, I only want to see "real world", "practical" comparison.

Michael did the right thing in re-doing the test.

Sure, the NEX7 puts higher demand on lenses since it has higher resolution.  But then it uses the part of the lens which has the best characteristic - the center - just like the Canon 7D's 18 megapixels uses the best part of the lens compared to the 5D.  This is why the 7D has "higher" resolution than the 5D per given lens.

Again the point it moot.  What Michael did on the re-do is real-world.

Too bad the M9 has a noisy sensor.  Leica fans should be throwing their hands up in disgust.  But then it is what it is - a noisy camera. So if it loses to the NEX7 when it comes to taking photos, so be it. It only has itself to blame - i.e. Leica.

With the second test, Michael shows the NEX7 is better than the Leica for low ISO photos.  Given how practical it is because of its size, that is a good thing.  Perhaps Leica should then get a better sensor to go with their $9000 camera body.  Sony makes great sensors for other companies, like Nikon.  Perhaps Leica can get a sensor from Sony.

« Last Edit: December 20, 2011, 12:14:28 AM by ErikKaffehr » Logged

jameskatt
Newbie
*
Offline Offline

Posts: 5


« Reply #19 on: December 20, 2011, 08:41:48 AM »
ReplyReply

Looks like NEX7 wins over M9 again at 1600.

From my point of view, I am interested in taking candid, non-flash photos of real people, who aren't posing for a photo, who are moving normally (e.g. children), in non-noon light such as indoors. I want to see natural appearing, low noise photos - something that matches what my eyes can see.  This is where digital photography has a difficult time making good photos.  The field is disappointing in this regard.

One needs a camera that easily goes up to at least ISO 25,000 to 50,000 so that the shutter speed can be at least 1/250th sec at f/8.  This is where the new Canon 1Dx becomes very interesting - albeit at a lower resolution than the NEX7.  However, I'd like to see a smaller, lighter camera capable of this.

To me, the NEX7 is just a more capable Canon G10.  The camera can take high resolution photos at low ISO speeds.  But it hasn't moved the state of the art far enough. It still cannot see close to what the human eye can see. I've been waiting years for that.
Logged
Pages: [1] 2 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad