Forum

Pages: [1]   Go Down
  Print  
Author Topic: Surprisingly Poor Resolution?  (Read 2339 times)
borispmchan
Newbie
*
Offline Offline

Posts: 6


« on: December 15, 2013, 05:40:53 PM »

This a comment for this page on the website
I've checked and cross-referenced other sites' tests and it appears GX7 indeed has slightly poorer resolution on imatest, resolving only 23-2400lines whilst E-P5 can do around 25-2600lines. Looking deeper, it appears GX7's JPEG engine renders things more softly and less contrasty and in reality the difference is not as great as the MPix scores shown here. I would guestimate the MPix score of Olympus 75mm on GX7 is more like 23/26*12 = 10.6MPix rather than 7MPix.
Logged
Emilie_DxOMark
Administrator
Hero Member
*****
Offline Offline

Posts: 530


« Reply #1 on: December 17, 2013, 10:18:03 AM »

Hello,

Thanks for your interesting post. Please keep in mind that most of other analysis are provided for JPEG picture, a good sharpen algorithm could increase sharpness easily.

Best regards,

The DxOMark team
Logged
Anders W
Newbie
*
Offline Offline

Posts: 12


« Reply #2 on: December 17, 2013, 04:52:04 PM »

The resolution of the GX7 sensor has been tested on the basis of RAW files, not just on the basis of out-of-camera jpegs, and has been found to be roughly on a par with that of other micro four thirds cameras with 16 MP sensors. See the results from optycne.pl here, for example.

http://www.optyczne.pl/239.4-Test_aparatu-Panasonic_Lumix_DMC-GX7_Rozdzielczość.html

There is simply no way your results could be as far out of line as they actually are (worse than for the G1, the very first MFT camera with 12 MP sensor) unless a) you made some mistake in your testing or b) tested a faulty unit.

Note that some MFT cameras have problems with so-called "shutter shock" (blur induced by shutter action) in certain shutter-speed ranges and that this problem might potentially affect individual copies of a certain camera more than other copies of the same camera (e.g., because some component is not fixed in a perfectly rigid manner). Note also that some MFT cameras, including the GX7, have electronic shutters that should normally let you bypass any blur-related problems caused by the mechanical shutter.

When you tested the GX7: Did you use the mechanical shutter or the electronic one?
Logged
Adventsam
Newbie
*
Offline Offline

Posts: 2


« Reply #3 on: December 18, 2013, 09:38:29 PM »

Come on dxo, you really are off the pace with this set of results, they are almost a joke! It seriously puts your reputation on the line as it just looks stupid and does not match user experience. I would argue that my gx7 is sharper than the em-5,in your own optics pro9 raw, so how do these results stack up!
Logged
hoodlum
Newbie
*
Offline Offline

Posts: 28


« Reply #4 on: December 18, 2013, 11:02:16 PM »

This is not the first time a body scored much lower than other bodies with similar sensors.  The Sony A580 also had resolution scores almost half of other similar Sony sensors.  I agree that it is either a testing or sample issue. 

Personally I would like DXO to add more lenses to their existing tests rather than test every body that comes out.  The sensor AA filters don't change that often, especially once you have tested bodies with no AA filters. 
Logged
Emilie_DxOMark
Administrator
Hero Member
*****
Offline Offline

Posts: 530


« Reply #5 on: December 20, 2013, 11:23:33 AM »

The resolution of the GX7 sensor has been tested on the basis of RAW files, not just on the basis of out-of-camera jpegs, and has been found to be roughly on a par with that of other micro four thirds cameras with 16 MP sensors. See the results from optycne.pl here, for example.

http://www.optyczne.pl/239.4-Test_aparatu-Panasonic_Lumix_DMC-GX7_Rozdzielczość.html

There is simply no way your results could be as far out of line as they actually are (worse than for the G1, the very first MFT camera with 12 MP sensor) unless a) you made some mistake in your testing or b) tested a faulty unit.

Note that some MFT cameras have problems with so-called "shutter shock" (blur induced by shutter action) in certain shutter-speed ranges and that this problem might potentially affect individual copies of a certain camera more than other copies of the same camera (e.g., because some component is not fixed in a perfectly rigid manner). Note also that some MFT cameras, including the GX7, have electronic shutters that should normally let you bypass any blur-related problems caused by the mechanical shutter.

When you tested the GX7: Did you use the mechanical shutter or the electronic one?

Dear Anders,
 
Thanks for your feedback. The lens results using the Panasonic GX7 are indeed surprising.  Optyczne.pl is a very good website and their reviews are great.
 
For the Panasonic GX7, the tests were performed on one camera body, loaned by Panasonic. It is very unusual to have sample variation between cameras these days but we will perform more tests on another camera sample in early January. (The lab will be closed between Christmas and New Year’s Day).
 
We are aware of the possible vibration problem produced by a mechanical shutter.
 
The following describes our procedure for shooting Through Focus MTF Targets. For each measured focal length and aperture setting:
-       Distance set to 35x the focal length
-       Camera is positioned in the middle of a focusing rack; autofocus is initiated to test accuracy (in live view mode if possible)
-       Autofocus switched OFF
-       We then perform more than 40 shots on a long focusing rack, bracketing around the focus distance to find the optimal focusing position (without using autofocus). With this step, called “through focus”, we are 100% certain to locate and record the optimum or “best focus” position for each lens, when mounted on the camera.
 
There are many sources of vibration that could impact sharpness / resolution measurement (i.e. mechanical shutter, mirror for vibration, tripod-type used etc …)
 
-          To avoid problems arising from vibration we use a very specific protocol:
·         Lab is in perfect darkness
·         Heavy (72 kg) studio stand
·         Remote control is used
·         Step 1: mirror lock where possible
·         Step 2: Timer:  2s
·         Step 3:  Time exposure 1s in total dark in order to get rid of the shutter vibrations
·         Flash between 1/4000s and  1/1000s depending on the aperture, synchronized on the second curtain
·         Electronic shutter where possible

Best regards,

The DxOMark team
Logged
Emilie_DxOMark
Administrator
Hero Member
*****
Offline Offline

Posts: 530


« Reply #6 on: December 20, 2013, 11:26:01 AM »

This is not the first time a body scored much lower than other bodies with similar sensors.  The Sony A580 also had resolution scores almost half of other similar Sony sensors.  I agree that it is either a testing or sample issue. 

Personally I would like DXO to add more lenses to their existing tests rather than test every body that comes out.  The sensor AA filters don't change that often, especially once you have tested bodies with no AA filters. 

Good remark, even if the A580 results were checked on 2 or 3 cameras the results lead to the same conclusion.
The A580 had an pretty strong AA filter.
Maybe it is the same explanation for GX7, that is what we will check in January.
About lens recommendation, we think it is important for reader to have their own lens result on their specific camera.
In 2014 we will try to perform more lens test. Stay tuned Smiley

Best regards,

The DxOMark team
Logged
Emilie_DxOMark
Administrator
Hero Member
*****
Offline Offline

Posts: 530


« Reply #7 on: December 20, 2013, 11:27:02 AM »

Come on dxo, you really are off the pace with this set of results, they are almost a joke! It seriously puts your reputation on the line as it just looks stupid and does not match user experience. I would argue that my gx7 is sharper than the em-5,in your own optics pro9 raw, so how do these results stack up!

Thanks for your feedback. We always take a particular attention to our test and perform many of them to validate the final measurements.
What we can say is that the camera we got had such results. We will perform some new tests on a new camera early January.
If you want more details about our testing conditions please check the previous post Smiley

Best regards,

The DxOMark team
Logged
Anders W
Newbie
*
Offline Offline

Posts: 12


« Reply #8 on: December 22, 2013, 04:19:53 PM »

Hi Emilie,

Thanks for your feedback and the description of your testing procedures. I am glad to hear that you will be testing another GX7 unit in January and hope that the results it provides will be less perplexing than those you got with the first unit. Since your procedures seem to rule out blur due to camera shake as well as focus error, and since I am sure you are right that sample variation between cameras of a kind that would matter here is very rare, it is difficult to say what might have gone wrong. But it seems that something nevertheless did.

One further question about your test procedures: When you test a new camera, I suppose you actually test it with each and every lens that has already been tested with other bodies belonging to the same system (in this case micro four thirds) although I realize that is a lot of work. A short-cut would of course be to test the new camera with just a single lens or a few lenses and then extrapolate the results to the remaining lenses. But I would guess you do not employ such a short-cut. Have I understood things correctly?
« Last Edit: December 22, 2013, 04:22:53 PM by Anders W » Logged
Emilie_DxOMark
Administrator
Hero Member
*****
Offline Offline

Posts: 530


« Reply #9 on: January 07, 2014, 08:59:59 AM »

Hello,

Yes you did Smiley, and that represent the biggest part of the job. We have 3 labs dedicated to these tests. We always perform the maximum of test we can with every cameras we have. Every camera-lenses combination are checked carefully.

Best regards,

The DxOMark team
Logged
Pages: [1]   Go Up
  Print  
 
Jump to: