[normhead]

K-3 is the second image... if you look in the bands on the left hand side, you can clearly see are more distinct, but, I botched the test... believe it or not, I was puzzled as to why I had to back up the camera to get the same field of view on the K-3 shot. When putting the lenses away I realized I had the 1.4 converter still attached to the K-3 and didn't notice... so if the question is, does and A-50 ƒ2 resolve better on a K-3 with a 1.4 converter than it does on a K-5 with just the 50, I think the answer is yes. As for the question I intended to answer, I'll have to try again later.

I hate it when that happens.

ANd I took the target image down and everything...

[audiobomber]

It is important to note that DXO sharpness scores are weighted averages, not maximum sharpness.

Sharpness
The DxOMark resolution score shows sharpness performance of a lens-camera combination averaged over its entire focal length and aperture ranges.

The resolution score is computed as follows:
For each focal length and each f-number, we first compute sharpness and then weight it throughout the field, tolerating less sharpness in the corners than in the center. This gives one number for each focal and aperture combination.
Then, for each focal length, we select the maximal value of sharpness over the range of available apertures. We average this value over the whole range of focal length to obtain the DxOMark resolution score that we report (in P-MPix).

Note that for a wide-range zoom, there are huge differences between the resolutions for different focal lengths.

Sharpness is expressed in PMpix and is typically between 50% and 100% of the sensor pixel count, Differences below 1 P-MPix are usually not noticeable.

Best resolutions are usually attained for fixed focal lenses and moderate apertures (depending on the lens, between f/2.8 and f/8).
Metric Scores - DxOMark

[normhead]

I like what DxO has done, visiting today, there's a lot more information on their testing....and they've certainly clarified some issues for me....
However the hypothesis proposed by IR that poorly rated lenses in some cases got better increases than better lenses, can't really be evaluated, because DxO test all pretty high quality lenses. The lenses IR said improved the most were 70-300 or 18-250 type lenses .. IR suggested that high quality lenses would gain less than poor quality lenses, and I don't see anything in these number that wouldn't support that, because DxO tested no poor quality lenses.

That being said the IR results were pretty similar to the DxO results, with differences probably deriving from different test conditions and measuring procedures.

[audiobomber]

I like what DxO has done, visiting today, there's a lot more information on their testing....and they've certainly clarified some issues for me....

Yeah, I linked that graph twice previously in this thread. I guess you didn't notice.

How do embed the graph?

[normhead]

Do a screen capture and crop the part i want then post it like any picture is what I do. I always like to put as much info out there as possible in the thread, it saves people from going back and forth between pages and let's people interpret the data in the manner they choose rather than living with my take on it.

Unfortunately I can't find the IR D7000 vs D7100 comparisons to link to. I'm not sure if they took them down or I just can't find them, but given the age of those cameras now, and the number of times I've looked, I suspect they took them down.

[audiobomber]

Do a screen capture and crop the part i want then post it like any picture is what I do. I always like to put as much info out there as possible in the thread, it saves people from going back and forth between pages and let's people interpret the data in the manner they choose rather than living with my take on it.

Unfortunately I can't find the IR D7000 vs D7100 comparisons to link to. I'm not sure if they took them down or I just can't find them, but given the age of those cameras now, and the number of times I've looked, I suspect they took them down.

Seems like a lot of work.

Here's my lazy link to the D7100/D7000 comparison:
Lens performance depends on sensor - DxOMark

That's the equivalent of a jump from K-5 to K-3.

---------- Post added 2015-02-24 at 15:13 ----------

However the hypothesis proposed by IR that poorly rated lenses in some cases got better increases than better lenses, can't really be evaluated, because DxO test all pretty high quality lenses. The lenses IR said improved the most were 70-300 or 18-250 type lenses .. IR suggested that high quality lenses would gain less than poor quality lenses, and I don't see anything in these number that wouldn't support that, because DxO tested no poor quality lenses

The poorest lens DXO tested in K-mount is the Tamron 18-200mm. It scored 4mp on the K-5 II, and only 5mp on the K-3. The numbers are horrible, but that's a 25% increase in resolution.

The sharpest lens tested for K-mount is the Sigma 35mm f1.4. It scored 10mp on the K-5 II, 16mp on the K-3, a 60% increase in resolution. So IR's conclusion does not seem to hold water.

[geomez]

I like what DxO has done, visiting today, there's a lot more information on their testing....and they've certainly clarified some issues for me....
However the hypothesis proposed by IR that poorly rated lenses in some cases got better increases than better lenses, can't really be evaluated, because DxO test all pretty high quality lenses. The lenses IR said improved the most were 70-300 or 18-250 type lenses .. IR suggested that high quality lenses would gain less than poor quality lenses, and I don't see anything in these number that wouldn't support that, because DxO tested no poor quality lenses.

That being said the IR results were pretty similar to the DxO results, with differences probably deriving from different test conditions and measuring procedures.

Two of the three FF lenses on this list show no improvement from K-5 to K-3 but all the APS-C lenses do. Could this be in favor of the theory that FF lenses don't do as well as APS-C lenses on APS-C cameras?

[normhead]

Interesting observation... I would have though those two lenses would do well, but the are not only FF lenses, but lenses designed for film. It makes you wonder if there's a difference.

[Pål Jensen]

(feel free to label the following as an ingnorant rant): I have spent 35 years photographing with larger formats up to 8x10 and immediately saw that this guy's arguments are full of #$%!!. All you have to do is get away from the "magical" full-frame (ie: 36x24mm), which this guy seems to feel is the gold standard of photography and has no understanding of lenses/focal length/aperture and their relation to film format.

This is true for every proponent of "equivalence"
You can use any format as reference and misuse the law of reciprocity to suit you argument. .

---------- Post added 02-25-15 at 12:05 AM ----------

Where it's useful is when trying to plan another system based on a second format, where you may be trying to get something 'extra' from the format addition or change. You'll need to know about equivalence if you want to not waste your money in one way or another.

The problem is that there is no DOF equivalence among formats when comparing different focal lengths yielding the same angle of view on their respective formats. It is a misunderstanding based on mixing what the various terms actually mean, and an oversimplification of DOF.
The DOF of a lens is not defined by a single number calculated from the F.stop and the focal length in use as proposed by equivalence ; DOF is based on aperture, focal length, focus distance and subject magnification. DOF of a lens forms a spectrum of possibilities or a space if you like. If you compare "DOF space" of two lenses of different focal lengths but similar angle of view on their formats, they will overlap a lot, but some areas will not; they will never totally overlap. For some reasons, never explained, only a single point outside this overlap, and only a point not covered by the smaller format is of value for the equivalence proponents. All other DOF characteristics are ignored. We are never given a reason for this. Nor, for that matter, why DOF wide open is the only factor of importance among the factors that make up reciprocity. Nor are we told why everything must be "normalized" to FF to be of any value.
When you take into calculation that DOF is dependent on focus distance and magnification as well, it is obvious that there are images you can shoot on APS which cannot be done on FF even by cropping when using "equivalent" lenses. In fact, in cases I have calculated, the APS "equivalent" lens will have thinner minimum DOF than the FF (even if they have the same max aperture) simply because it has inherent larger magnification (this is why you can take fantastic macro images with the Q).

[Nicolas06]

This is true for every proponent of "equivalence"
You can use any format as reference and misuse the law of reciprocity to suit you argument. .

---------- Post added 02-25-15 at 12:05 AM ----------



The problem is that there is no DOF equivalence among formats when comparing different focal lengths yielding the same angle of view on their respective formats. It is a misunderstanding based on mixing what the various terms actually mean, and an oversimplification of DOF.
The DOF of a lens is not defined by a single number calculated from the F.stop and the focal length in use as proposed by equivalence ; DOF is based on aperture, focal length, focus distance and subject magnification. DOF of a lens forms a spectrum of possibilities or a space if you like. If you compare "DOF space" of two lenses of different focal lengths but similar angle of view on their formats, they will overlap a lot, but some areas will not; they will never totally overlap. For some reasons, never explained, only a single point outside this overlap, and only a point not covered by the smaller format is of value for the equivalence proponents. All other DOF characteristics are ignored. We are never given a reason for this. Nor, for that matter, why DOF wide open is the only factor of importance among the factors that make up reciprocity. Nor are we told why everything must be "normalized" to FF to be of any value.
When you take into calculation that DOF is dependent on focus distance and magnification as well, it is obvious that there are images you can shoot on APS which cannot be done on FF even by cropping when using "equivalent" lenses. In fact, in cases I have calculated, the APS "equivalent" lens will have thinner minimum DOF than the FF (even if they have the same max aperture) simply because it has inherent larger magnification (this is why you can take fantastic macro images with the Q).

There is obviously always an equivalence. The deph of field lenses provide is continuous. Focus a little far a away, you get more. Get a bit longer/ Go more near to the subject, you get less. Only apperture can't be continous but with 1/3 of stop you have a good enough approximation. 1/3 a stop is just arround 10% difference in deph of field.

For dof, as long as it is in the range of thing the lens can do at all, you be able to get the one you want. For framing, that true of course too.

Eventually the perspective would be different, but if you change the focal just the level needed as the same time as the sensor, the perspective could be kept the same. The bokeh/rendering would be different anyway.

But yes you can get same framing, same dof, same perspective compression by choose the right corresponding focal length, apperture and keeping the same subject distance by changing format... As long as the corresponding lense exist (or near enough for that purpose) you'd get it.

Difference in bokeh/rendering would stay. They would be subble in many case and has much can be set as being part of the precise model of lense you use than the format too.

So mathematically ou can do it, of course, this is even basic math.

Your idea is like because 2 vehicule have 2 different wheels diameter, they obviously can't run at the same speed. Ultimately, this might be true, but in many case even between a bicycle and an airplane, there a shared range of speed, here in 0-40km/h range. The one that look at appertures and say f/2.8 stay f/2.8 and look at focal length and say focal lens stay the same are the one that would conclude that obviously because bicycles and trucks have arround the same wheel diameter they would drive the same speed and that most cars because they come with smaller wheels obviously comes with more limitation and can't go has fast. Reality is that 3 kind of vehicule have a shared range of speed even if the formula in term of wheel diameter and turns per minute to know the vehicule speed might appear a bit complex. Hey there even a "Pi" irrationnal number in it!

[Ash]

If you stick with full frame lenses, good luck finding wider than 24mm without being a fisheye

Pentax has some rectilinear UWA lenses - FA 20, FA 20-35, M 20 to name a few...

[wildman]

The poorest lens DXO tested in K-mount is the Tamron 18-200mm. It scored 4mp on the K-5 II, and only 5mp on the K-3. The numbers are horrible,

If I have any issue with the DXO scores it's that they often, for me at least, don't translate into real world use very well.
If the Tammy 18-200 had a score of 8 instead of 4, on any given camera body, would that really translate into a lens anywhere near twice as "good" as a lens with a score of 4?

For instance Tammy 18-200 on a K20 = a "horrible" DXO score of 6.

In other words does a "horrible" DXO score necessarily give you, at least technically, a horrible image?

Real world shot taken with the 180 buck Tammy 18-200 mounted on a K20....

[normhead]

There is obviously always an equivalence.

"Obviously" is not a scientific term. After such a statement the most obvious response is "prove it".

I'd like to see a proof that included the relative size of circles of confusional various subject distances of subject and back ground proving that you can match both DoF and have the same size circles of confusion using different formats. I can imagine it being both possible and impossible. So I need to see a proof.

The poorest lens DXO tested in K-mount is the Tamron 18-200mm. It scored 4mp on the K-5 II, and only 5mp on the K-3. The numbers are horrible,

But a 20% increase in resolution for the lens, better than any of the DA and DA*s and Ltd. tested by DxO.

In other words does a "horrible" DXO score necessarily give you, at least technically, a horrible image?

Definitely the weakest part of the DxO logic is the inference that their standards mean something in terms of image production. They have never established that people prefer 4 Mp images to 8 Mp images, of a given image at a given size. SO what they do is pretty much done in a DxO vacuum, where they spit out numbers of unknown human value.

[Rondec]

"Obviously" is not a scientific term. After such a statement the most obvious response is "prove it".



But a 20% increase in resolution for the lens, better than any of the DA and DA*s and Ltd. tested by DxO.



Definitely the weakest part of the DxO logic is the inference that their standards mean something in terms of image production. They have never established that people prefer 4 Mp images to 8 Mp images, of a given image at a given size. SO what they do is pretty much done in a DxO vacuum, where they spit out numbers of unknown human value.

I have a hard time with a score like 4 megapixels with a zoom. Surely at some point in the zoom range it is better and at certain apertures. I couldn't give the DA 55-300 a single score. It is pretty decent at 200mm and f5.6, but at 300mm needs to be at f8 minimum for good results. And when DXO Mark is giving numbers that are well below the predicted numbers, what does that mean?

It doesn't really make sense that a lens will score 4 megapixels on a K5 (16 megapixel sensor) and then score 5 megapixels on a K3 (24 megapixel sensor). Why should the lens get any better moving to a sensor with higher pixel density?

My experience with the K5 II versus K3 is that there is some improvement in resolution, as long as you shoot in your lens's sweet spot, shoot low iso, and have good technique. But it certainly isn't earth shattering. As to whether or not a "20 percent" resolution increase would translate to an increase in printing size, I have no idea, but I don't think it would be a huge difference.

[normhead]

It doesn't really make sense that a lens will score 4 megapixels on a K5 (16 megapixel sensor) and then score 5 megapixels on a K3 (24 megapixel sensor). Why should the lens get any better moving to a sensor with higher pixel density?

It does if you don't support the notion that lenses are being out resolved by sensors. As long as the lens resolves more than the sensor captures, every increase in sensor density will lead to an increase in resolution.

As to whether or not a "20 percent" resolution increase would translate to an increase in printing size, I have no idea, but I don't think it would be a huge difference.

Photography tends to be defined by stops, 100% increases or 50% decreases tend to be perhaps not important, but noticeable. 10-20% increases, not so much.

Or to look at it. I took some great images, that produced really good prints... with my *IST DS. 6 MP. The MP thing is getting to be absolutely crazy and completely locked into the law of diminishing returns. We saw an 8 MP difference from the K-5IIs to the K-3. My *IST produced great pictures with 6MP. The last 8 MP I bought brought me a heck of a lot less satisfaction than the first 6 MP. There is a thing happening where as you get higher and higher in the MP count, each MP has less and less photographic technical value.

Consider a K-3 landscape at 2700 lw/ph on a 30x 20 picture. The vertical resolution will be 20/2700 or .0074 inches.
On a D810 it will be 20/3700 or .0054.

How close to the image do you have to be to be able to detect a difference of .002, (two thousdanths) of an inch? And that is on a 30x20 print.

Can a printer even make use of that difference. What output device can be calibrated so that a difference .002 inches can be detected?

I just do the math, now the BS starts.