[kh1234567890]

I'm talking about 8-bit JPEG (which the majority of people use) where you don't have any of those advantages. You are talking about RAW & post processing.
Even in RAW and post you are still very limited by an 8-bit display.

The JPEG is gamma mapped as opposed to the RAW which is linear.

[audiobomber]

Going by the "screen" data sets provided by DXO, even without the ISO80 K-5 would still win.

OK, you win. The K-5 has up to .6EV better DR performance. And lower resolution and wonky auto-focus.

[normhead]

Going by the "screen" data sets provided by DXO, even without the ISO80 K-5 would still win. I gave you the link in my earlier post.



You were talking about data, I just followed suit. I did say the DXO final rating maybe subjective so ignore it and just see the DR data they gave us in "screen", which shows there are differences between The K-3 and the others and how it is lost compared to the D7100 and K-5. I'm just trying to convey that we should not discount their data.



Agree. A bit of dramatization on my part. I should just say the K-3 consistently records less DR than the D7100 and K-5 in ISO100-ISO800.

My observations using both cameras would confirm that at least for the K-5 and K-3. For landscape, I may continue to carry the K-5, keeping the K-3 on a long lens for wildlife. MInd you, the K-3 is still quite a bit better than my K-x or K20D. Let's keep it in perspective. But you aren't going to pull detail out of the depths of the shadows like you could on the K-5. Proper exposure is more critical.

[falconeye]

I didn't read all of this thread, but I am offering my short advice:

1. DxO tests are not w/o flaws but as of today, are still the most accurate source for the kind of properties they measure (sensor efficiency).

2. Do not use their overall score. Use their 3 independent sub scores as they are and resist to combine them. And add #MP as a forth independent score.

3. sensorgen.info converts the 3 subscores into quantum efficiency, read noise and full well capacity which may be more easily to understand and compare.

4. Differences of 15 in overall score are 1 stop. A difference of 5 is imperceptible.

5. One thing DxO misses to measure is the true dynamic range at low spatial frequencies (they are just using a scaling formula). This implies they are sometimes missing banding and uneven illumination. Something which should be taken into account as a fifth score taken from elsewhere. Other things missing are heat or readout speed.

6. Lens tests are almost impossible to interpret in a meaningful way. The lens scores suck. Use DPReview who republish DxO lens test results in a more digestible and meaningful way.

[wahid_satu]

OK, you win. The K-5 has up to .6EV better DR performance. And lower resolution and wonky auto-focus.

Resolution, noise and DR are related, if again we're going with the DXO data. If we're viewing prints, then it doesn't matter. If we're cropping it matters in that although you get more pixel in the same crop area between a 24MP image and 16MP image, if there are more noise on that 24MP crop, it would not be more usable. However, in decent light where there wouldn't be more pixel noise on the 24MP images then you are right you will get that extra resolution.

For the record, I still would buy the K-3 had I the means. It is one hell of a camera and overall arguably better than any other APS-C out there.

[audiobomber]

4. Differences of 15 in overall score are 1 stop. A difference of 5 is imperceptible.

And a difference of 2 is meaningless.

[normhead]

I didn't read all of this thread, but I am offering my short advice:

1. DxO tests are not w/o flaws but as of today, are still the most accurate source for the kind of properties they measure (sensor efficiency).

2. Do not use their overall score. Use their 3 independent sub scores as they are and resist to combine them. And add #MP as a forth independent score.

3. sensorgen.info converts the 3 subscores into quantum efficiency, read noise and full well capacity which may be more easily to understand and compare.

4. Differences of 15 in overall score are 1 stop. A difference of 5 is imperceptible.

5. One thing DxO misses to measure is the true dynamic range at low spatial frequencies (they are just using a scaling formula). This implies they are sometimes missing banding and uneven illumination. Something which should be taken into account as a fifth score taken from elsewhere. Other things missing are heat or readout speed.

6. Lens tests are almost impossible to interpret in a meaningful way. The lens scores suck. Use DPReview who republish DxO lens test results in a more digestible and meaningful way.

That sounds authoritative enough for me.

[bxf]

Not that I believe the small difference in DR is significant, but just to put a bit more weight on the absence of ISO 80 on the K-3, let's not forget that the D7100 achieved the 13.72Ev DR at ISO 69 (measured), not ISO 100.

[Khukri]

Hello Rbefly!

What you say may have some truth in it. Equally, look at the facts, objectively. When it comes to:
1. Action photos - Pentax AF speed and accuracy does not match Canon/Nikon.
2. Landscapes - FF bodies of Canon and Nikon offer better image quality overall compared to Pentax’s crop bodies.
3. In camera image stabilisation for tele (where it matters most) is also not as effective as in-lens variant.
4. Add to above, limited lens’ choice compared to Nikon/Canon, not to mention pricing.

So, although Pentax glass is beautiful and the bodies are feature rich, in practical usage the Pentax system falls short in these very important parameters. If Pentax improved on above, a lot of Canon/Nikon guys would bite.

As for the original subject of this thread, DXO findings are not at great variance to those of many other test labs. To that extent their findings are at par.

[Rondec]

Hello Rbefly!

What you say may have some truth in it. Equally, look at the facts, objectively. When it comes to:
1. Action photos - Pentax AF speed and accuracy does not match Canon/Nikon.
2. Landscapes - FF bodies of Canon and Nikon offer better image quality overall compared to Pentax’s crop bodies.
3. In camera image stabilisation for tele (where it matters most) is also not as effective as in-lens variant.
4. Add to above, limited lens’ choice compared to Nikon/Canon, not to mention pricing.

So, although Pentax glass is beautiful and the bodies are feature rich, in practical usage the Pentax system falls short in these very important parameters. If Pentax improved on above, a lot of Canon/Nikon guys would bite.

As for the original subject of this thread, DXO findings are not at great variance to those of many other test labs. To that extent their findings are at par.

Interesting that you lump Canon and Nikon together in your assessment. From my point of view, Canon suffers significantly, even in the full frame market, due to the lesser quality of their sensors. This is born out by DXO Mark's testing (which as I said before I do find reliable). Both Sony and Toshiba are currently making better sensors than Canon. This makes your number 2 point half invalid. I don't truly believe that you are going to get significantly better images with a Canon 5D Mk III and whatever lens versus a Pentax and a DA 12-24 or Sigma 8-16mm.

As to point 3, I find in camera image stabilizing to be adequate for my purposes (I don't shoot beyond 300mm) and I get between two and three stops of benefit from it. As to whether in lens stabilization is truly better, it is hard to say. The only thing I think you can say for sure is that it stabilizes your viewfinder, at certain cost, both financial and with respect to size.

As to number 4, I hope that people do review Pentax's options before purchasing. If you need a 600mm f4 lens, it just isn't going to happen with Pentax. But, for most standard focal lengths, I believe that there is pretty good coverage. And I have no trouble including Sigma lenses as possible purchase items for whom there is a blank spot or a lens that is too expensive from Pentax.

[DDoram]

The K3 should not be compared to Canon/Nikon full frame bodies but to the 70D and the D7100. There is no proof that either camera has faster and more accurate AF than the K3.

Dale

[normhead]

2. Landscapes - FF bodies of Canon and Nikon offer better image quality overall compared to Pentax’s crop bodies.

A D800 offers better landscape IQ and I'm assuming the Sony A7r. I've yet to see comparisons between a K-3 and whatever Canon has to offer, but at low ISO I'm willing to bet a K-3 will stand up to any Canon FF models, so in answer to that I'd say, it depends. I would say the same about the D600, and 6D. I'd fully expect a K-3 landscape from 100-800 ISO to be better than a 6D and very close to the equivalent of a D600. Looking at the images over at Imaging Resource, I was finding my K-5 images comparable to the best Canon FFs.. in the ISO range indicated above. The K-3 has stepped it up a notch, Canon hasn't kept up.

The K3 should not be compared to Canon/Nikon full frame bodies but to the 70D and the D7100. There is no proof that either camera has faster and more accurate AF than the K3.

Dale

Back in the early K-5 days I saw, actual tests of focusing speed, and at that time in good light, a K-5 was almost as fast as a Canon or Nikon. within .015 of a second. However the K-5 fell behind as the light fell off, the Nikons and Canons often focusing a half second better in low light. I have seen no such comparisons since the K-30 and K-5 IIs came out, as Pentax has bumped their low light performance, there have been no comparisons that I know of, to suggest there is any Canikon advantage in any light. That being with the K-5 IIs. The K-3 being a whole new system even more advanced than the K-5 II is even more untested. Yet even in the K-3 review on the forum it is clearly stated that the K-3 doesn't focus as fast as a Nikon or Canon, but no one is showing us any numbers.

To me it's just repeating the results of tests done years ago and applying it to new equipment, so it may or may not be true. But I haven't seen one review anywhere that suggests Pentax is slower than a Canon or Nikon in good light that has given anything but the .015 number. So I view that statistic with more suspicion, with the release of every new camera body.

So maybe someone could point out where they are getting this stuff from. I'm not sure where I originally saw it. And it would be nice to look over some current stats, with current bodies. Those K-5 stats are getting real old right now. And those stats gave Pentax the edge in the % of keepers, achieving focus lock more accurately than either Canon or Nikon. So even back in the K-5 days, the Canon Nikon focus system didn't have an advantage. It appeared that they had sacrificed focus accuracy to achieve focus speed.

Whenever I had real numbers to look at, the trade offs and strengths and weaknesses were apparent. Right now, I don't own any Nikon of Canon equipment so I don't know. But it was never that the Pentax system was all bad, compared to the other two. That was a myth of the times.

A myth that was repeated despite evidence presented to the contrary.

[David]

We take a lot on trust, in terms of the objectivity and rigour of their testing procedures, and sample size is one thing people don't often think about, but which bothers me. You'd hope they would undertake testing of samples over a period of time to avoid batch problems and to comprehend minor software changes etc, but that's probably asking too much for a commercial enterprise to do. If our friend Falk Lumo were to do the same things, I reckon I'd be better assured of at least being told about the scope of testing and any associated caveats, if such were the case. As it is, I have no evidence to not trust DxO for what they produce but, as others have said, it isn't the whole story.

We'll said Rob, but we buy our cameras for lots of 'good' reasons than the third decimal point in pixel peeping.
David

Ps, K3 show and tell next week?

[RobA_Oz]

We'll said Rob, but we buy our cameras for lots of 'good' reasons than the third decimal point in pixel peeping.
David

Ps, K3 show and tell next week?

Too right! There's been an influx of K-3s to this small island in the last few weeks, so that's something to be happy about.

As we agree, there's more to a camera than mere benchmarking on measurables, as good as they may be.

[pTom]

+1
For this matter, I trust DXO . That's why I bought K-x , then K-5