My K-7 alpha test report and blog

melander · 06-20-2009, 07:26 AM

I think this link have been shown alot of times. And as they say, a picture...
D300-K20D-D50
Compare cameras

D90-K20D-D50
Compare cameras

casq · 06-20-2009, 07:52 AM

Originally posted by melander

Cloud21: Remember that D90 got the same flaw like every other brand. Iso 1600 is no way near iso 1600. D90 Iso 1600 is in fact iso 1125 while pentax Iso 1600 is 1383. So crank up the iso on the D90 to iso 2000 and compare there instead.
Same goes at iso 3200. D90 = 2300 K20D = 2700 D300 = 2100!!

Start comparing real iso instead of in camera Iso numbers. D300 needs somewhere around iso 4500 to compete with K20d iso preformance, who wins then?

I have heard this same theory before and i dont actually see what the problem is. To my understanding if the exposure looks like an 1600 then its one, despite what the cameras "real iso" is. Some manufacturers get the proper 1600 ISO with less "real iso" but the exposure is nevertheless the same and hence equal despite having different "real iso". You want to compare ISO 1600 shots you do it with the settings the camera gives you, you dont start playing around with real iso because the outcome will be a different exposure instead of the 1600.

filorp · 06-20-2009, 08:14 AM

What does it mean that canikon has lower real iso? is this mean that at the same iso for instance 1600, pentax makes you choose 1/25 at f4 and canikon 1/15 at f4?

casq · 06-20-2009, 08:35 AM

I think the exposures are more dependant on metering than real iso, best way to find out would be to go manual with the cameras. I think the end result would be equal exposures despite different real isos.

cloudswimmer · 06-20-2009, 11:08 AM

Originally posted by melander

I think this link have been shown alot of times. And as they say, a picture...
D300-K20D-D50
Compare cameras

D90-K20D-D50
Compare cameras

I've never looked at DXO's site before.So according to them the K10 is better than the K20 right?

**Rondec** · 06-20-2009, 11:34 AM

The whole point of ISO is that equal settings on different (brand) cameras should produce roughly equivalent results. That is to say, if you use ISO 1600, the same lens at f 4.0, and a shutter speed of 1/60 second, you should get the same photo. Metering has nothing to do with this. The exposure is determined completely by those settings and nothing else. The whole point is that if a company artificially inflates their ISO and states 1000 to be 1600, the noise will of course look better, but you will have to shoot with a slower shutter speed to get the same exposure.

filorp · 06-20-2009, 11:41 AM

Originally posted by Rondec

The whole point of ISO is that equal settings on different (brand) cameras should produce roughly equivalent results. That is to say, if you use ISO 1600, the same lens at f 4.0, and a shutter speed of 1/60 second, you should get the same photo. Metering has nothing to do with this. The exposure is determined completely by those settings and nothing else. The whole point is that if a company artificially inflates their ISO and states 1000 to be 1600, the noise will of course look better, but you will have to shoot with a slower shutter speed to get the same exposure.

this is what im talking about.... this is it? im gonna check with my canikon colleagues...

casq · 06-20-2009, 12:02 PM

There is evidence to show that a D90 needs a slower shutter?. I think you'd get roughly the same ISO 1600 shots if you line up different cameras and bypass the metering with manual controls, unless the meterings dont vary i believe they will vary but i dont know if its enough. Real iso is not really a factor in this, i think it simply shows that they achieve same for less.

This kind of reminds me of the native ISO issue, people used to think that lower is always the best and desirable but it wasnt that simple. Nowadays things can be designed so that optimal ISO is about 200 which gives the best DR. I think it might be sort of the same case with real iso and ISO.

Its also a good question what real iso exactly is because not only does high-iso values differ the lower ones seems to fluctuate too. I believe given the same target exposure, they all produce equivalent images despite having different real isos.

Its not an issue that different manufacturers have different real isos, there is no cheating or any kind of magic involved. D90 et al high-iso capabilities are widely documented and very solid, be it raw or jpg. You get your 1600 shot not ISO 1000 shot. Or does other camera brands produce ISO 1000 that are as fast or faster as D90's ISO 1600?.

edit.
Also if achieving good dxomark scores and great high-iso shots not only on dxomark but in practice, is this easy (just tone down real iso) how come other manufacturers havent figured it out yet?. This is as simple as engineering gets, so what are others waiting?. In this regard nikon's sensor is ahead of the pack, im not expecting K7 to surpass D90 but i expect it to get close enough. Never been a huge fan of high-iso so its not a big loss to me, i much rather take some extra resolution to play with it also helps a bit with noise if get some in my pictures.

**Rondec** · 06-20-2009, 12:54 PM

I think this is something that would be really hard to test. First of all, it is hard to find lenses that are exactly the same in different mounts. Secondly, we are talking about a small part of one stop difference between cameras. No camera company would try to sneak ISO 800 as ISO 1600, but by shaving a few points off of it, nobody really notices and they can get the lower noise they want. I wouldn't even know how to do such a test.

TourDeForce · 06-20-2009, 01:32 PM

Originally posted by Rondec

I think this is something that would be really hard to test. First of all, it is hard to find lenses that are exactly the same in different mounts. Secondly, we are talking about a small part of one stop difference between cameras. No camera company would try to sneak ISO 800 as ISO 1600, but by shaving a few points off of it, nobody really notices and they can get the lower noise they want. I wouldn't even know how to do such a test.

Screw mount adapters & a Takumar lens.

Class A · 06-20-2009, 06:33 PM

Originally posted by yipchunyu

how are the performance with low iso? I think the K20D is not very good compare to K10D

According to the DXOmark measurements (select the SNR 18% tab), the difference is rather small.

EDIT: Previously I wrote about my surprise how well the Nikon D40 sensor (same SONY CCD sensor as in the K100D) compares. It appeared to be better in almost all disciplines (noise, dynamic range, tonal range, colour sensitivity). I had a closer look and discovered that the default graphs from DXOmark show per-pixel-performance. If you normalize the figures to a common image size (see the "Print" tab in the DXOmark comparisons) then a practical comparison can be made. The 6MP CCD sensor shows a very good performance but is then often (but not always) beaten by the 14.6MP CMOS sensor.

Note that the K20D's noise is apparently better than that of the K10D when measured in terms of the same image size (rather than per pixel).

falconeye · 06-20-2009, 07:14 PM

New blog entry:
Comparative resolution study K-7 vs. K20D

[IMGWIDELEFT]http://falklumo.smugmug.com/photos/569142520_FFhhX-X3.jpg[/IMGWIDELEFT]

The K-7 may have a weaker AA filter than the K20D. Just a bit. Please, go to the page linked above for all the gory details

In practice though, the differences are rather marginal. The important point then is that I found no evidence for rumours that the K-7 is a bit soft. To the contrary

Class A · 06-20-2009, 10:12 PM

Originally posted by falconeye

[Please, go to the page linked above for all the glory details

Bring 'em on! (the catwalk models).

In the overview comparison it seems that the K20D seems to offer a tad more resolution and also a bit more contrast. You might be right about the K-7's AA filter being slightly weaker but I'm a bit puzzled in terms of overall resolution/contrast (which is not just a result of the AA filter). Overall the results don't appear to be conclusive, or are they?

The data as it currently presents itself doesn't quite figure for me. The more I looked at unaltered shots, the more the two cameras looked alike though. If I had to pick one based on my, admittedly not super systematic, examinations of your shots, I'd pick the K20D.

In any event, let's not forget that your hardware isn't the latest and greatest. The difference indicated by Pentax seems to relevant given the subtle differences we are seeing.

falconeye · 06-21-2009, 02:18 AM

Originally posted by Class A

Overall the results don't appear to be conclusive, or are they?

Your comment pretty much summarizes what I thought about it too. And why I spent so many words...

I am probably splitting hairs here and so, what I found is like for the noise test: "almost like K20D"

However, this is a result as well

Class A · 06-21-2009, 03:28 AM

Originally posted by falconeye

what I found is like for the noise test: "almost like K20D"

However, this is a result as well

Definitely.

Many thanks for doing these time consuming tests in this thorough manner! I had high expectations with you being an alpha tester but you are certainly delivering!

06-20-2009, 12:02 PM	#128
casq New Member Join Date: Jun 2009 Location: Finland Posts: 7	There is evidence to show that a D90 needs a slower shutter?. I think you'd get roughly the same ISO 1600 shots if you line up different cameras and bypass the metering with manual controls, unless the meterings dont vary i believe they will vary but i dont know if its enough. Real iso is not really a factor in this, i think it simply shows that they achieve same for less. This kind of reminds me of the native ISO issue, people used to think that lower is always the best and desirable but it wasnt that simple. Nowadays things can be designed so that optimal ISO is about 200 which gives the best DR. I think it might be sort of the same case with real iso and ISO. Its also a good question what real iso exactly is because not only does high-iso values differ the lower ones seems to fluctuate too. I believe given the same target exposure, they all produce equivalent images despite having different real isos. Its not an issue that different manufacturers have different real isos, there is no cheating or any kind of magic involved. D90 et al high-iso capabilities are widely documented and very solid, be it raw or jpg. You get your 1600 shot not ISO 1000 shot. Or does other camera brands produce ISO 1000 that are as fast or faster as D90's ISO 1600?. edit. Also if achieving good dxomark scores and great high-iso shots not only on dxomark but in practice, is this easy (just tone down real iso) how come other manufacturers havent figured it out yet?. This is as simple as engineering gets, so what are others waiting?. In this regard nikon's sensor is ahead of the pack, im not expecting K7 to surpass D90 but i expect it to get close enough. Never been a huge fan of high-iso so its not a big loss to me, i much rather take some extra resolution to play with it also helps a bit with noise if get some in my pictures. Last edited by casq; 06-20-2009 at 12:34 PM.

06-20-2009, 06:33 PM	#131
Class A Pentaxian Join Date: Aug 2008 Location: Wellington, New Zealand Posts: 11,251	Originally posted by yipchunyu how are the performance with low iso? I think the K20D is not very good compare to K10D According to the DXOmark measurements (select the SNR 18% tab), the difference is rather small. EDIT: Previously I wrote about my surprise how well the Nikon D40 sensor (same SONY CCD sensor as in the K100D) compares. It appeared to be better in almost all disciplines (noise, dynamic range, tonal range, colour sensitivity). I had a closer look and discovered that the default graphs from DXOmark show per-pixel-performance. If you normalize the figures to a common image size (see the "Print" tab in the DXOmark comparisons) then a practical comparison can be made. The 6MP CCD sensor shows a very good performance but is then often (but not always) beaten by the 14.6MP CMOS sensor. Note that the K20D's noise is apparently better than that of the K10D when measured in terms of the same image size (rather than per pixel). Last edited by Class A; 06-20-2009 at 10:26 PM.

06-20-2009, 07:14 PM	#132
falconeye Veteran Member Join Date: Jan 2008 Location: Munich, Alps, Germany Photos: Gallery Posts: 6,871 Original Poster	Comparative resolution study K-7 vs. K20D New blog entry: Comparative resolution study K-7 vs. K20D [IMGWIDELEFT]http://falklumo.smugmug.com/photos/569142520_FFhhX-X3.jpg[/IMGWIDELEFT] The K-7 may have a weaker AA filter than the K20D. Just a bit. Please, go to the page linked above for all the gory details In practice though, the differences are rather marginal. The important point then is that I found no evidence for rumours that the K-7 is a bit soft. To the contrary Last edited by falconeye; 07-31-2009 at 01:43 AM.

06-20-2009, 10:12 PM	#133
Class A Pentaxian Join Date: Aug 2008 Location: Wellington, New Zealand Posts: 11,251	Originally posted by falconeye [Please, go to the page linked above for all the glory details Bring 'em on! (the catwalk models). In the overview comparison it seems that the K20D seems to offer a tad more resolution and also a bit more contrast. You might be right about the K-7's AA filter being slightly weaker but I'm a bit puzzled in terms of overall resolution/contrast (which is not just a result of the AA filter). Overall the results don't appear to be conclusive, or are they? The data as it currently presents itself doesn't quite figure for me. The more I looked at unaltered shots, the more the two cameras looked alike though. If I had to pick one based on my, admittedly not super systematic, examinations of your shots, I'd pick the K20D. In any event, let's not forget that your hardware isn't the latest and greatest. The difference indicated by Pentax seems to relevant given the subtle differences we are seeing. Last edited by Class A; 06-21-2009 at 12:17 AM.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Sony Alpha 900	dylansalt	Canon, Nikon, Sony, and Other Camera Brands	18	10-27-2010 11:01 AM
Any field test report for the kr?	opiedog	Pentax DSLR Discussion	1	09-15-2010 11:47 AM
rawtherapee v3.0 alpha 1 is out	deejjjaaaa	Digital Processing, Software, and Printing	14	01-07-2010 09:41 PM
sigma 35-135mm alpha IV	pdxbmw	Pentax SLR Lens Discussion	3	12-29-2009 01:47 PM
Sony Alpha dSLR-A200	deejjjaaaa	Pentax News and Rumors	52	01-09-2008 03:30 PM

06-20-2009, 07:26 AM	#121
melander Veteran Member Join Date: Jul 2008 Location: Sweden, Umea Posts: 876	I think this link have been shown alot of times. And as they say, a picture... D300-K20D-D50 Compare cameras D90-K20D-D50 Compare cameras

06-20-2009, 07:52 AM	#122
casq New Member Join Date: Jun 2009 Location: Finland Posts: 7	Originally posted by melander Cloud21: Remember that D90 got the same flaw like every other brand. Iso 1600 is no way near iso 1600. D90 Iso 1600 is in fact iso 1125 while pentax Iso 1600 is 1383. So crank up the iso on the D90 to iso 2000 and compare there instead. Same goes at iso 3200. D90 = 2300 K20D = 2700 D300 = 2100!! Start comparing real iso instead of in camera Iso numbers. D300 needs somewhere around iso 4500 to compete with K20d iso preformance, who wins then? I have heard this same theory before and i dont actually see what the problem is. To my understanding if the exposure looks like an 1600 then its one, despite what the cameras "real iso" is. Some manufacturers get the proper 1600 ISO with less "real iso" but the exposure is nevertheless the same and hence equal despite having different "real iso". You want to compare ISO 1600 shots you do it with the settings the camera gives you, you dont start playing around with real iso because the outcome will be a different exposure instead of the 1600.

06-20-2009, 08:14 AM	#123
filorp Veteran Member Join Date: Jun 2008 Location: Aberdeen Scotland Photos: Gallery \| Albums Posts: 398	What does it mean that canikon has lower real iso? is this mean that at the same iso for instance 1600, pentax makes you choose 1/25 at f4 and canikon 1/15 at f4?

06-20-2009, 08:35 AM	#124
casq New Member Join Date: Jun 2009 Location: Finland Posts: 7	I think the exposures are more dependant on metering than real iso, best way to find out would be to go manual with the cameras. I think the end result would be equal exposures despite different real isos.

06-20-2009, 11:08 AM	#125
cloudswimmer Forum Member Join Date: Mar 2008 Location: So.Cal. Photos: Gallery Posts: 89	Originally posted by melander I think this link have been shown alot of times. And as they say, a picture... D300-K20D-D50 Compare cameras D90-K20D-D50 Compare cameras I've never looked at DXO's site before.So according to them the K10 is better than the K20 right?

06-20-2009, 11:34 AM	#126
Rondec Loyal Site Supporter Join Date: Mar 2009 Location: Gladys, Virginia Photos: Gallery Posts: 27,652	The whole point of ISO is that equal settings on different (brand) cameras should produce roughly equivalent results. That is to say, if you use ISO 1600, the same lens at f 4.0, and a shutter speed of 1/60 second, you should get the same photo. Metering has nothing to do with this. The exposure is determined completely by those settings and nothing else. The whole point is that if a company artificially inflates their ISO and states 1000 to be 1600, the noise will of course look better, but you will have to shoot with a slower shutter speed to get the same exposure.

06-20-2009, 11:41 AM	#127
filorp Veteran Member Join Date: Jun 2008 Location: Aberdeen Scotland Photos: Gallery \| Albums Posts: 398	Originally posted by Rondec The whole point of ISO is that equal settings on different (brand) cameras should produce roughly equivalent results. That is to say, if you use ISO 1600, the same lens at f 4.0, and a shutter speed of 1/60 second, you should get the same photo. Metering has nothing to do with this. The exposure is determined completely by those settings and nothing else. The whole point is that if a company artificially inflates their ISO and states 1000 to be 1600, the noise will of course look better, but you will have to shoot with a slower shutter speed to get the same exposure. this is what im talking about.... this is it? im gonna check with my canikon colleagues...

06-20-2009, 12:54 PM	#129
Rondec Loyal Site Supporter Join Date: Mar 2009 Location: Gladys, Virginia Photos: Gallery Posts: 27,652	I think this is something that would be really hard to test. First of all, it is hard to find lenses that are exactly the same in different mounts. Secondly, we are talking about a small part of one stop difference between cameras. No camera company would try to sneak ISO 800 as ISO 1600, but by shaving a few points off of it, nobody really notices and they can get the lower noise they want. I wouldn't even know how to do such a test.

06-20-2009, 01:32 PM	#130
TourDeForce Veteran Member Join Date: Jan 2008 Location: Florida Photos: Gallery Posts: 514	Originally posted by Rondec I think this is something that would be really hard to test. First of all, it is hard to find lenses that are exactly the same in different mounts. Secondly, we are talking about a small part of one stop difference between cameras. No camera company would try to sneak ISO 800 as ISO 1600, but by shaving a few points off of it, nobody really notices and they can get the lower noise they want. I wouldn't even know how to do such a test. Screw mount adapters & a Takumar lens.

06-21-2009, 02:18 AM	#134
falconeye Veteran Member Join Date: Jan 2008 Location: Munich, Alps, Germany Photos: Gallery Posts: 6,871 Original Poster	Originally posted by Class A Overall the results don't appear to be conclusive, or are they? Your comment pretty much summarizes what I thought about it too. And why I spent so many words... I am probably splitting hairs here and so, what I found is like for the noise test: "almost like K20D" However, this is a result as well

Thread Tools	Search this Thread
	Search this Thread: Advanced Search

06-21-2009, 03:28 AM	#135
Class A Pentaxian Join Date: Aug 2008 Location: Wellington, New Zealand Posts: 11,251	Originally posted by falconeye what I found is like for the noise test: "almost like K20D" However, this is a result as well Definitely. Many thanks for doing these time consuming tests in this thorough manner! I had high expectations with you being an alpha tester but you are certainly delivering!