August 19, 2009

Pentax K-7 in German magazine test

I don't normally copy/cite content from other sources. But I am going to make an exception now. I cite numeric test results which were recently published by German photo magazine ColorFoto (ColorFoto 9/2009). I do so because there are so few sources around which offer a comparison of the K-7 image quality when brought to lab tests.

You may read more about the K-7 test done by ColorFoto by fetching a printed copy. You may probably be able to download the full text after the printed copy went out of stores, by going here:

So, besides citing their numeric image quality test results, I am only going to say this much: They liked the K-7 a lot!

Cited test results:

Test (ISO)Pentax K-7Pentax K20DNikon D300Canon D50Olympus E-3
Resolution 100 [LP/PH]13271310125313271146
Resolution 400 [LP/PH]13251294124813261113
Resolution 800 [LP/PH]13181302118912861052
Resolution 1600 [LP/PH]12951287114411911053
Texture loss 1000.
Texture loss 4000.
Texture loss 8000.
Texture loss 16000.
Noise 100 [VN]
Noise 400 [VN]
Noise 800 [VN]
Noise 1600 [VN]
Dynamic range 100 [EV]8.58.510.59.58.5
Dynamic range 400 [EV]
Dynamic range 800 [EV]
Dynamic range 1600 [EV]
AF + shutter latency at ~10 EV [ms]390280270280360
AF + shutter latency at ~3.5 EV [ms]5801060470260730

Lenses used (image quality test / AF test):
  • Pentax DFA 100 f2.8 Macro / DA 18-55 f3.5-5.6 AL II
  • Nikon AF 60 f2.8 Macro / AF-S 27-70 f2.8 G ED
  • Canon EF 50 f2.5 Macro / EF 24-70 f2.8 L USM
  • Olympus ED Digital 50 f2.0 Macro / ED 12-60 f2.8-4.0 SWD

Light values (EV) in AF tests are computed from quoted 3000/30 lx figures by myself.

The image tests have been run from the JPGs produced out of camera. The default setting of the K-7 (as used in the test) applies noise reduction starting at over ISO 800 only which explains the relatively good values at ISO 1600. Most other brands obviously already apply noise reduction at ISO 800 or below.

The ColorFoto lab tests may be unique in that they provide a measurement of the loss of texture (typically caused by aggressive noise reduction algorithms).

Therefore, one may have a careful look at the "texture loss" measurements. Pentax has none, Nikon has it somewhat under control and Canon is more like a mess. So, depending on how you weight noise vs. texture, either Nikon or Pentax would win in the noise/texture department. Dynamic range when measured from JPGs does always include the tone curve applied. As there are plenty of tone curve parameters to play with, esp. on the K-7, I don't know how useful the corresponding measure is. I included it for the sake of completeness of citation.

More interestingly though, the K-7 sensor seems to have no significant difference to the K20D sensor (to be confirmed by raw file lab tests though) and the K-7 JPG engine seems to offer subtle improvement in resolution and noise/texture. This is in line with what I found out myself (cf. my other blog articles). And, this test sharply contradicts another blog article ricing high concerns...

Another figure being noteworthy of course is the AF measurement. In low light, the K20D AF.S figures were last in the field. The K-7 now seems to be more comparable to Nikon or Olympus with Canon still winning the competition. But the results are to be taken with a grain of salt: The difference of figures in full light between K-7 and K20D showcase errors in the measurements.


The first lab test of the K-7 image quality I am aware of confirms my early impression: image quality is on par with K20D or better. The JPG engine still produces a lot of noise but preserves a lot of texture detail too, compared to its direct competition.


  1. "ricing high concerns"... priceless :)

    BTW, what I find most disappointing in the images my K20D produces is color rendition and loss of dynamic range at high ISOs (1600 and above - which many people don't need at all, but I do).

  2. it's all about light, just get the subject properly iluminated, then metered and your k20d will provide excellent results, even at iso 1600-2500


Please if posting anonymously, choose a nickname for your post. Thanks.