[JohnBee]

UPDATE...

Imaging Resource has since updated(reshot) their K-5 Still Studio scenes to correct the DOF issue that was present in these shots. Therefore... we can consider the case closed and the K-5 seems right where it belongs in terms of resolution performance.

So 'Thank-You' to Imaging Resources and DPReview for taking the time to make things right.

JohnBee

...this issue has since been resolve and the follow-up/conclusion can be seen here: https://www.pentaxforums.com/forums/pentax-k-5-forum/123138-k-5-vs-d7000-soft...-part-2-a.html


- original thread...

Hello,

I'm writing this in response to a series of tests that have been coming about the web lately. And I've found(even in RAW) where the K-5 seems to produce softer images than the competition. Or.. specifically, in contrast to other systems using the same sensor.

The first of the issues were uncovered by Imaging Resources. Which incidentally, have admitted to shooting the K-5 with improper settings(f/5.6 vs f/8). Which they claimed will be re-shot with the proper settings.

However, even in this scene(where focus was proper), there are clear detail distinctions between the K-5 and the D7K and Sony using the same sensor.

I'm tempted to say AA filter. However, I'm not at all certain if the AA filter can or would cause finer detail distortions such as the ones observed in the standard studio scene around the resolution dial(or scale).

I've also found this phenomenon(softness) to occur in other head to head testing as well including the K-7, which appears to be much sharper pixel per pixel than the K-5.

All I can say is, I do hope this is not an issue with the camera or, worst yet... a permanent condition that we will just have to live with. Though I for one would love to get to the bottom of it.

Taken from Imaging Resource:

The K-5 Still Life images were shot at f/5.6 vs f/8. (The lens is probably ever so slightly sharper overall at f/5.6 due to diffraction, but we feel that the depth of field is more important for the still life subject. The flat targets are shot at f/5.6.). The flat target show that that the K-5's AF was working very accurately with those subjects with the 70mm macro.

EXHIBIT A: D7000, ISO100, NR_0
[IMGWIDELEFT]http://www.bertin.ca/tmp/K-5/D7000_CRP1.jpg[/IMGWIDELEFT]
EXHIBIT B: K-5, ISO100, NR_0
[IMGWIDELEFT]http://www.bertin.ca/tmp/K-5/K-5_CRP1.jpg[/IMGWIDELEFT]
EXHIBIT C: D7000, ISO100, NR_0
[IMGWIDELEFT]http://www.bertin.ca/tmp/K-5/D7000_CRP2.jpg[/IMGWIDELEFT]
EXHIBIT D: K-5, ISO100, NR_0
[IMGWIDELEFT]http://www.bertin.ca/tmp/K-5/K-5_CRP2.jpg[/IMGWIDELEFT]
EXHIBIT E: D7000, ISO100, NR_0
[IMGWIDELEFT]http://www.bertin.ca/tmp/K-5/D7000_CRP3.jpg[/IMGWIDELEFT]
EXHIBIT F: K-7, ISO100, NR_0
[IMGWIDELEFT]http://www.bertin.ca/tmp/K-5/K-7_CRP3.jpg[/IMGWIDELEFT]
EXHIBIT G: K-5, ISO100, NR_0
[IMGWIDELEFT]http://www.bertin.ca/tmp/K-5/K-5_CRP3.jpg[/IMGWIDELEFT]

-

ORIGINAL FILES(RAW/DNG):

D7000, ISO100 RAW/NEF Sample: http://216.18.212.226/PRODS/D7000/FULLRES/D7000hSLI00100_NR0.NEF
K-5, ISO100 RAW/DNG Sample: http://216.18.212.226/PRODS/K5/FULLRES/K5hSLI00100_NR_0.DNG

I have more RAW/DNG samples coming in from DPReview as well....

---

Last edited by JohnBee; 11-20-2010 at 04:04 AM.

[Eric Seavey]

When I first purchased my K-7, I found the images looking soft too, even in raw. I found that when I adjusted the sharpness , that viewing the photos in the software that came with the camera, that the photos seemed sharper.
What software (raw editor) are you using to compare the images?
Are you seeing this softness using the same lens when comparing the k-5 to the k-7?

[ozlizard]

Hey if you're going to pixel peep and measurebate you'll see all sorts of things. Show me a 24"x36" print because that is only true measurement. After all isn't that what these tools are supposed to do?

[cbaytan]

Hello,

I'm writing this in response to a series of tests that have been coming about the web lately. And I've found(even in RAW) where the K-5 seems to produce softer images than the competition. Or.. specifically, in contrast to other systems using the same sensor.

Same worries, this is why I hold my K-5 order and ordered K-r first, imaging resource's corrected shots will affect my prior decision, then I will watch for more K-5 samples to see what is the real story behind..

There is also a shade on K-5 shots looks like a thick AA filter or exposed to excessive UV light on the mountain at the samples. Those I've seen are no better than my K-7 up to know. I hope that I am wrong.

[Wheatfield]

What lens are they using for the K5 shots? What lens are they using for the D7000 images?
Are they both shot at the same aperture or is one lens at the sweet spot, the other just shot at a randomly chosen aperture?
Are both cameras optimized or are we seeing the traditional oversharpening by Nikon vs. Pentax?
Without knowing this sort of stuff, tests like this are worse than useless.
To me, it's just another case of people putting themselves out there as testers without the ability to run a test.

[PiotrKrochmal]

Exactly, to make corresponding photos from both cameras they should use identical lenses for example sigma, tamron with different mount only. Both lenses have to be check and calibrated to be sure that all differences comes only from body. If that will be made proper you could make pixel peep.

In my opinion, to have professional test you should make special lens with interchangeable mount for all type of body.
Only that type of test will be real comparable tests for me.

Best regards and go make photos

[Wheatfield]

My 7D user friend is quite impressed so far with what I'm getting off my K7. But, I use lenses that are every bit as good as his L lenses....

[falconeye]

I'm tempted to say AA filter. However, I'm not at all certain if the AA filter can or would cause finer detail distortions such as the ones observed in the standard studio scene around the resolution dial(or scale).

I've not yet started my formal testing, so I don't know yet what exactly is going on here.

But for the record:

I notice that Exhibit C/D clearly shows that the AA filter on the D7000 (exhibit C) is too weak. The horizontal lines in the "Pure Brewed" label have all rainbow colors you imagine, just not the correct one: mid gray. The K-5 (exhibit D) renders just fine, be it due to AA filtering or another source of blur (defocus, lens, whatever): perfect color.

It is a known fact that Bayer filtered sensors need an AA filter and a properly adjusted sharpening operator to deliver optimum results: too weak an AA filter and you get ugly Color Moiré artifacts, too weak a sharpening operator and you get soft details. Side by side comparisons using AA filters of different strengths must use different sharpening operators too.

Maybe, there is a trend to deploy too weak AA filters to impress naive measurebators and which Pentax didn't follow. Maybe it is something else, I don't know.

DISCLAIMER: I do not claim that the K-5 AA filter is stronger than D7000's, nor do I claim they are equal.


But the D7000 AA filter is rather weak. Isn't it enough that video is full of color Moiré and edge flickr?


Attachment:

The D7000 (top) and K-5 (bottom) crops, the K-5 crop was sharpened with FocusMagic 1px radius.





UPDATE:

The samples submitted by JohnBee have been 200% crops. That confused my sharpening tools. I fixed it now. One should really state when posting higher than 100% crops!

[NaClH2O]

John, maybe it's my eyes, or just my less than ideal work monitor, but in the second set there seems to be a big difference between the contrast of the Nikon 7K image and the Pentax K-5 image. Since contrast is a main component of image sharpness, I'm wondering if there is a difference between the contrast values of the two shots

NaCl(it all comes back to the lens used)H2O

[JohnBee]

What lens are they using for the K5 shots? What lens are they using for the D7000 images?
Are they both shot at the same aperture or is one lens at the sweet spot, the other just shot at a randomly chosen aperture?
Are both cameras optimized or are we seeing the traditional oversharpening by Nikon vs. Pentax?
Without knowing this sort of stuff, tests like this are worse than useless.
To me, it's just another case of people putting themselves out there as testers without the ability to run a test.

I've updated the OP with an excerpt from IR regarding the testing method.
IMO. There should be no reason to doubt the results. However, they did agree to re-take the entire still life studio scene just to be sure.

Also, I have more samples coming in from DPReview which lean in the same direction. So it's really starting to look as though the K-5 has something of a soft side to it

[rawr]

...To me, it's just another case of people putting themselves out there as testers without the ability to run a test.

Agreed. Really JohnBee you should think about this a bit more before putting stuff like this out there. Look at all the variables that might influence what you are presenting here. I mean you are getting nowhere near to seeing raw sensor performance here.

[JohnBee]

I notice that Exhibit C/D clearly shows that the AA filter on the D7000 (exhibit C) is too weak. The horizontal lines in the "Pure Brewed" label have all rainbow colors you imagine, just not the correct one: mid gray. The K-5 (exhibit D) renders just fine, be it due to AA filtering or another source of blur (defocus, lens, whatever): perfect color.

HAHA! - And here I thought that's how it was suppose to look

And here's an A900 crop as well(which I trust):


Okay, so weak or 'no' AA filter at all is the issue here.
That is most interesting and its looking more like a give and take scenario to me.
However, what's also interesting is that both the Sony and the Nikon have the same type of IQ rendering as far as that goes. Guess they chose to take the same path.

Having said that, in other areas, in this particular case, it really looks as though the K-5 is loosing detail.
I don't know if the AA filtering is either beneficial or detrminental to IQ on the grand scheme of things, but it certainly does come across this way when interpreting results such as these and others.

[telfish]

John,

I think you are way off base here with this post. Do you have a K5? Well I have one, my wife has one and we are both very happy with the image sharpness and the color we got out of these cameras.

I disregard these sort of torture tests because they are not really comparable. They use unlike lenses and settings so they are worthless.

Please don't start a false meme on this. It's irresponsible. I am a real world shooter and my real world results with decent glass are superb.

[artobest]

For what it's worth, the wall behind the Samuel Smith bottle looks a great deal noisier in the D7000 shot than the K5.

[ManuH]

But for the record:

I notice that Exhibit C/D clearly shows that the AA filter on the D7000 (exhibit C) is too weak. The horizontal lines in the "Pure Brewed" label have all rainbow colors you imagine, just not the correct one: mid gray. The K-5 (exhibit D) renders just fine, be it due to AA filtering or another source of blur (defocus, lens, whatever): perfect color.

I notice a kind of purple halo and a lower contrast on the K-5 shot. Lens or focusing differences?

Some people also noted that the K-5 exhibit more purple fringing than the K-7/K20D. Could it be related to the AA filter as well?