[Reltih]

It's hard to know what all of the various calculations really mean, but I understand what you're saying. I've be curious to know what the calculations are at 12,800 and if it is still at a 25% spread.

Of course it is very hard to make sure both cameras work exactly at the same actual settings. Most of us may not be able to do that.

As per DxO, at ISO 12,800, the measured values of K-3 and Df are 11,836 and 9,540 respectively, that means the same difference (24%).

---

Last edited by Reltih; 01-08-2014 at 09:28 PM.

[dosdan]

As per DxO, at ISO 12,800, the measured values of K-5 and Df are 11,836 and 9,540 respectively, that means the same difference (24%).

Since high ISO changes are performed with digital, rather than analogue gain, there should be no relative change between the two cameras in noise performance from about ISO3200 upwards, unless the amount of any NR being applied to the raw image data changes.

JPEG will be a different matter, but DxOMark concerns itself with raw data only.

Dan.

[SteveM]

There are different variations between the cameras (shake reduction and difference with ISO), but does it really matter? DXO rates the K3 for low light ISO at 1216 and the Df at 3279 in their tests.....but I don't know what that really means.

All I can speak to is how each camera works in my hands. I've spent some more time with both cameras and a few different lenses on each. The K3 holds it's own for ISO, I like the saturation from Pentax and the dynamic range seems a bit better especially when pushed in pp. The Nikon has slightly faster AF (but have noticed it hunts in low light but that's expected as there is no light assist), the high ISO is better which helps for indoor candids. Simple observations is all......

[dosdan]

DXO rates the K3 for low light ISO at 1216 and the Df at 3279 in their tests.....but I don't know what that really means.

The Sports/LL Score is explained here: https://www.pentaxforums.com/forums/172-pentax-k-3/246214-dxomark-k3-k5ii-k5.html#post2623307

Since the ratio of sensor sizes in the Df to K-2 is 2.34:1, but the ratio of their Sports scores is 2.69:1, that suggests that Df has a sensor with a higher fill-factor/higher quantum efficiency/better performing microlenses/less transmission loss in the CFA. I suspect the reason is the the Df probably has high-performance dual-layer microlenses and a higher fill-factor.

Dan.

---

Last edited by dosdan; 01-10-2014 at 12:01 AM.

[john5100]

Given that the sensor is from the d4 then that makes sense.

[tuco]

... DXO rates the K3 for low light ISO at 1216 and the Df at 3279 in their tests.....but I don't know what that really means.

It means you can just set your camera's auto ISO upper limit to ISO 6400 and shoot away enjoying photography and not worry about all this boring, pixel peeping tech talk.

[ClinchShots]

Pentax natural white balance capture >>>

[john5100]

It means you can just set your camera's auto ISO upper limit to ISO 6400 and shoot away enjoying photography and not worry about all this boring, pixel peeping tech talk.

LOL...good one

[Poit]

Great real world review / comparison, really interesting to see. Thanks very much for posting it.

Something about the side-by-side shot makes me very proud to be a K-3 owner...not sure why. It also creates a strong desire for Pentax/Ricoh to continue putting effort into developing great cameras, and bringing the Pentax name back to being one which is taken seriously in the photography world.

For me, what this review really said was what an awesome system Pentax would have if they could bring the benefits of FF to the K-3 package...

[Class A]

Half of what I shoot is in low light, so there is definitely an allure from having the high ISO performance that full frame offers.

I'd love to use a full frame camera myself, but better "high ISO performance" of FF is plain and simply a myth.

Have a look at this old but still valid thread that explains why FF has zero low-light / high-ISO advantage over APS-C.

Now, there are indeed scenarios where an FF camera can be better in low light, but whenever that's the case this is always due to faster lenses being available for FF (provided sensor technologies are comparable). A 50/1.4 on FF would require a 33/0.9 on APS-C to achieve the same low-light performance. However, there is no 33/0.9 and this is the only reason why you may -- in certain scenarios -- be better off with an FF compared to an APS-C camera.

As expected, there is noticeably more noise with the K3.

With a fair comparison, the expectation would have been to see pretty much the same noise (as the sensors aren't that different in their noise performance).

The way you conducted the comparison, yes, one expects to see more noise with the K-3.

For the following reasons:

1. You are using the same f-stop for both cameras. This biases the comparison with 1.17 stops in favour of the Nikon. Actually, with the ISO 12,800 shot, the Nikon uses f/2.2 whereas the Pentax uses f/2.5. That's another 0.25 stop bias towards the Nikon.
2. You are looking at 100% crops. The higher magnification of the K-3 images also magnifies the noise. This biases the comparison with a 0.58 stop in favour of the Nikon.
3. You are using the same ISO sensitivities. The Nikon lies more about the true ISO value, e.g., at a nominal ISO 25600, the Nikon really only does ISO 17780 whereas the Pentax really does ISO 23529. This biases the comparison with 0.4 stop in favour of the Nikon.

So in total, the comparison is biased towards the Nikon with 2.4 stops (based on the image where you used f/2.2 for the Nikon and f/2.5 for the Pentax).

Of course this is not a 2.4 stop better performance by the Nikon. Rather, the comparison is stacked to disadvantage the Pentax by a huge factor of 2.15 stops. No wonder, the Nikon images look cleaner.

It shows a key difference between APS-C and FF: DoF.

That's incorrect. Different formats do not imply different DOF, as long as you scale everything along with the sensor.

I simply wanted to use the cameras as I normally would, using the same settings on each to see what the result would be.

I see where you are coming from, but what does "the same settings" mean?

Using the same numbers (e.g., for f-stop and ISO, 100% view for both) appears to establish a level playing field, but in reality it creates an unfair comparison.

Consider two runners, one with foot size 12 and the other with foot size 14. Would it be a fair race if you let them both use shoes of size 12? Using the same number for the shoe size (despite the fact that one runner has much larger feet) is analogous to using the same (as in "same numbers") exposure settings for two cameras despite the fact that one camera has a bigger sensor.

Again, I do see where you are coming from and I don't want to criticise your exercise, but it would be very wrong to use your "results" to conclude that the Df has much better noise performance than the K-3. If you compare the cameras at DxOMark, the Df looks quite a bit better, but that's because DxOMark uses the same f-ratio for all formats. Hence, the SNR advantage for the Df is mainly reflecting the surplus of total light let in by an FF lens set to the same f-stop as an APS-C lens.

[dgrPhotos]

The way you conducted the comparison, yes, one expects to see more noise with the K-3.

For the following reasons:

1. You are using the same f-stop for both cameras. This biases the comparison with 1.17 stops in favour of the Nikon.
2. You are looking at 100% crops. The higher magnification of the K-3 images also magnifies the noise. This biases the comparison with a 0.58 stop in favour of the Nikon.
3. You are using the same ISO sensitivities. The Nikon lies more about the true ISO value, e.g., at a nominal ISO 25600, the Nikon really only does ISO 17780 whereas the Pentax really does ISO 23529. This biases the comparison with 0.4 stop in favour of the Nikon.

So in total, the comparison is biased towards the Nikon with 2.15 stops.

In regards to 1 & 2, because the bigger sensor in the Nikon can utilize more of the light coming in and has bigger photosites makes it "cleaner" at higher iso values. #3 becomes moot because of #1 and #2. That's doesn't make it a myth, that's just physics.

[Fauxton]

Comparing a Nikon Df to a Pentax K3 is a little like comparing an apple to an orange, but that’s the point of this post; neither is a *better* camera, each being better at certain things. This isn’t a laboratory controlled pixel peeping review, but simply observations when using my cameras in real life.



I’ve been shooting Pentax for over 30 years, so I instinctively move along with Pentax as their product line evolves, so adding a Nikon body to my bag is a big change (and added expense!). I should mention that I have no intention of leaving Pentax. I still think it’s the best choice for everyday photography along with telephoto work. So why the Nikon? Half of what I shoot is in low light, so there is definitely an allure from having the high ISO performance that full frame offers. Why the Df? I’ve spent decades shooting film, so a film style body packed with the D4 sensor is a natural attraction. The Df currently holds the title of offering the best high ISO performance on the market.



There are also a couple of lenses I have wanted to own and either can’t find them, or they simply are not available for Pentax. I have started off with ordering the Df with its special edition 50/1.8, the Zeiss Apo Sonnar T* 135mm F/2.0 ZF.2 and the Zeiss Distagon 21mm F/2.8 ZF.2. I’ll need to take time to compare landscape work between the 21 and a similar focal length on Pentax, but based on what I am seeing so far, the 21/2.8 would be better matched with the K3.



I took both cameras to Butchart gardens at night with the intent of testing high ISO performance. A quick comment regarding the K5 vs. K3, I’ve taken a number of shots and the K3 seems as good to my eye. All Pentax shots are uing the FA 77 and the Nikon is using the 135. Anyways, I started at ISO 3200;

The K3 is first up. All shots are using the FA77.



Now the Df….using the Zeiss 135.



As expected, there is noticeably more noise with the K3. Both do well in detail (the scratch shown in the star on the centre left is blocked by a twig on the Pentax).

I did have some trouble nailing some shots with the Nikon, either through use error, or what I am assuming is the benefit of shake reduction (obviously not available with the Zeiss on the Nikon). The red tree trunk is a reflection off of the pond. This shot is with the Pentax @ISO3200. I was not successful getting a shot with the Nikon…



Both of these shots are at ISO 3200. The first shot is from the K3 with the second being the Nikon.





At ISO 12,800, as expected, the Nikon shines but the K3 holds its own;
Here is the K3….



And the Df….



Here is a 100% crop of the K3…..



and the Df……



Finally for ISO, here are two more shots at 12,800;

First the K3….



and the Df…..



So what about shooting in good light? First the K3 at ISO 100;



Then the Df at ISO 100



Another set with the first being the K3…



and then the Df….



100% crop from the K3…



100% crop from the Df….



Summary;

Personal preference and shooting style plays a big factor in choosing a camera. For example, I haven’t used both cameras in a studio setting, so if that’s where you spend most of your time, then you might (or might not) rate the cameras differently.

From first impressions with the Df, if I was to choose just one camera body, the K3 would be it. I’m looking forward to many years of use of the Df and can see it helping with half of what I shoot. I would be curious to compare it against the D800e, but already chose the Df over size and the way I like to use the camera. The Df has amazingly clean high ISO shots.

The K3 will continue be my go to camera for day to day shooting. Besides the cost factor, the build quality and form factor is better. The shell construction is much more solid, being all metal vs. the metal/plastic shell of the Df. The K3 offers in body stabilization and includes video which are not available on the Df. When I made the decision to pick up the Df, I had made the decision to stick with APS-C Pentax for Telephoto work, but I was a bit surprised at the performance of the K3 at lower ISO. As such, the K3 will see more use than I had originally thought.

K3 Pros

• Cost
• Build quality – metal vs. metal/plastic shell construction
• IQ at low ISO
• APS-C Crop for telephoto work
• In Body Image Stabilization
• Video

Df Pros

• IQ at high ISO
• Availability of aftermarket lenses (Zeiss / Voigtlander)

Thanks for the info Steve. The Df is a very very good camera. The K3 is as well. The Df has about a stop better noise performance. However as it has more pixels the 100% crop looks worse relative than it is. If you resize the crop to match the Df crop the apparent noise will reduce a bit. It will still be visible - more than the Df but not as bad as what you show there. Simply put you should get similar results in noise out of the K3 at roughly one ISO (corrected - as in 6400 agrees with the same 6400 measure in the other camera) below the Df if you keep the parameters somewhat equal.

The K3 won't match the low light capability of the Df - few if any do. On the other hand as you noted the K3 excels in other areas and it certainly is a pleasure to use. I'm glad you like it.

[jsherman999]

I'd love to use a full frame camera myself, but better "high ISO performance" of FF is plain and simply a myth.

Have a look at this old but still valid thread that explains why FF has zero low-light / high-ISO advantage over APS-C.

I have a problem with the term 'myth' here.

There is no larger-sensor advantage if you always must keep DOF the same for the same FOV. But no-one ever mandates that, and in fact 'accepting less DOF for the same FOV' often turns into 'welcoming less DOF for the same FOV' in a lot of cases, once you shoot for a while.

That thread is very good in that the OP lays out the equivalence involved. He does state very clearly that there's no 'noise' advantage if you intend to keep DOF the same (among the other things, like FOV and F-stop.) (Maybe at that time (2009) that was still a revelation?)

Saying it's a 'myth' confuses the issue, and IMO is misleading - if it can be stated like that, then it's also a 'myth' that aps-c sensors will give you better noise performance than 1/2.3'' P&S sensors can give you. While technically true, it's only true if we attach the odd mandate that we need to keep DOF the same in all cases (assuming for a moment the available lens disparity was less.) No-one ever says, "well, your aps-c DSLR is no better than my P&S because you have to stop down your lens to f/32 to match my P&S DOF, this negating the noise advantage!" You would always respond, "no, I don't need to stop down to f/32, I'm happy with my DOF at f/4, thank you." Same argument applies to FF vs. aps-c.

So... 'myth' is misleading, for these two reasons:

1) You are never required to keep DOF the same. If you can accept or welcome 1.3 stops less DOF for the FF shot for the same FOV and F-stop, you gain the noise advantage.
2) There simply are not always lenses fast enough available for aps-c that can match what's available for FF (linear aperture for the same FOV.) (& The greater the difference between sensor sizes, the greater this disparity is - see aps-c vs 1/2.3'')

IMO, 'often misunderstood' is 15 characters longer than 'myth', but worth the effort in accuracy payback.

Now, there are indeed scenarios where an FF camera can be better in low light, but whenever that's the case this is always due to faster lenses being available for FF (provided sensor technologies are comparable). A 50/1.4 on FF would require a 33/0.9 on APS-C to achieve the same low-light performance. However, there is no 33/0.9 and this is the only reason why you may -- in certain scenarios -- be better off with an FF compared to an APS-C camera.

It comes into play a lot also with the f/2.8 constant zooms. Until Sigma made the 18-35 1.8, there was no comparison anywhere in that range, if you needed to use a zoom for whatever you were shooting.

...If you compare the cameras at DxOMark, the Df looks quite a bit better, but that's because DxOMark uses the same f-ratio for all formats. Hence, the SNR advantage for the Df is mainly reflecting the surplus of total light let in by an FF lens set to the same f-stop as an APS-C lens.

Which is what you do as a photographer, most of the time just naturally - and 100% of the time when you're specifically trying to take advantage of the larger sensor to reduce noise.

.

[Rondec]

I have a problem with the term 'myth' here.

There is no larger-sensor advantage if you always must keep DOF the same for the same FOV. But no-one ever mandates that, and in fact 'accepting less DOF for the same FOV' often turns into 'welcoming less DOF for the same FOV' in a lot of cases, once you shoot for a while.

That thread is very good in that the OP lays out the equivalence involved. He does state very clearly that there's no 'noise' advantage if you intend to keep DOF the same (among the other things, like FOV and F-stop.) (Maybe at that time (2009) that was still a revelation?)

Saying it's a 'myth' confuses the issue, and IMO is misleading - if it can be stated like that, then it's also a 'myth' that aps-c sensors will give you better noise performance than 1/2.3'' P&S sensors can give you. While technically true, it's only true if we attach the odd mandate that we need to keep DOF the same in all cases (assuming for a moment the available lens disparity was less.) No-one ever says, "well, your aps-c DSLR is no better than my P&S because you have to stop down your lens to f/32 to match my P&S DOF, this negating the noise advantage!" You would always respond, "no, I don't need to stop down to f/32, I'm happy with my DOF at f/4, thank you." Same argument applies to FF vs. aps-c.

So... 'myth' is misleading, for these two reasons:

1) You are never required to keep DOF the same. If you can accept or welcome 1.3 stops less DOF for the FF shot for the same FOV and F-stop, you gain the noise advantage.
2) There simply are not always lenses fast enough available for aps-c that can match what's available for FF (linear aperture for the same FOV.) (& The greater the difference between sensor sizes, the greater this disparity is - see aps-c vs 1/2.3'')

IMO, 'often misunderstood' is 15 characters longer than 'myth', but worth the effort in accuracy payback.




It comes into play a lot also with the f/2.8 constant zooms. Until Sigma made the 18-35 1.8, there was no comparison anywhere in that range, if you needed to use a zoom for whatever you were shooting.




Which is what you do as a photographer, most of the time just naturally - and 100% of the time when you're specifically trying to take advantage of the larger sensor to reduce noise.

.

I think Class A makes it clear that the issue is one of what glass is available for APS-C versus full frame. If you had a group of f1 primes available for APS-C and f1.8 zooms, there would be no noise/iso advantage in going full frame. Of course, there aren't equivalent lenses out there for APS-C and if there were, they would probably be quite a bit more expensive than full frame (although the Sigma 18-35 is surprisingly economical), but still, full frame advantage is in the lenses available for it and not intrinsic to having a bigger sensor, or bigger megapixels or, fewer megapixels, which are all myths that are commonly stated.

[normhead]

Thanks for the great examples. A few examples showing exactly what you get and lose is worth a bazillion words on the internet in technical gobbledygook. And when you look at the images, its pretty easy to understand, despite having the smaller sensor, for many, the K-3 is the best choice. That is something you can argue the numbers all day and not show. NO one creates numbers with cameras, people with cameras create pictures.