I don't know really how to read this but what I do know is that it scored the same as the D3s (but the K-5 is better in DR and Color Depth) and that is unbelievable it pretty much squashes the 7D, but in the high ISO would the 5D II be like a half stop better? as the 5D II scored 1815 and the K-5 1162
http://dxomark.com/index.php/en/Camera-Sensor/Compare-sensors/%28appareil1%29/676|0/%28appareil2%29/483|0/%28appareil3%29/628|0/%28onglet%29/0/%28brand%29/Pentax/%28brand2%29/Canon/%28brand3%29/Nikon
[quote name='Christos' timestamp='1288977826' post='3974']
I don't know really how to read this but what I do know is that it scored the same as the D3s (but the K-5 is better in DR and Color Depth) and that is unbelievable it pretty much squashes the 7D, but in the high ISO would the 5D II be like a half stop better? as the 5D II scored 1815 and the K-5 1162
http://dxomark.com/index.php/en/Camera-Sensor/Compare-sensors/%28appareil1%29/676|0/%28appareil2%29/483|0/%28appareil3%29/628|0/%28onglet%29/0/%28brand%29/Pentax/%28brand2%29/Canon/%28brand3%29/Nikon
[/quote]
Forget it. Dxo tests a dont tell you anything about real-life camera performance in actual photography. They are contradictory to many other testing sites. The "scientific" method DXo uses and the zillions of graphs they employ only gives them the appaerance of being particularly reliable. (It may well be that the K5 is better than the 7D, I have no clue).
Just look at their test of the Canon 16-35mm lens. Look especially at the resolution graph for the 16mm end. Isnt that funny? According to dxo the border ond edges are a lot sharper at f/2.8 than they are at f/8. What a joke. Anybody who has ever shot with that lens can tell you that stopping down increases edge sharpness and not dereases it.
Regarding cameras: One thing that strikes me is that the measuremnts totally ignore the pixel count of cameras. A camera with more pixels can deliver higher image quality than a camera with less pixels even if its noise perfomance is slightly worse.
In addition, all the other measurments regarding dynamic range etc, quite often do not conform to what I am seeing from real life images.(e.g. that one camera is better than the other; quite often I would visually judge a camera equal or better while DXo considers it to be worse than another camera)
Whats also weird is that according to dxo the k5 has more than two stops more dynamic range at iso 100 than the 5D MArk II. Of course I cant totally deny that this is correct, because I havent seen any K5 images yet, but it seems highly unlikely. 14 stops DR would be absolutly revolutionary for a sensor with this high a pixel density.
My only explantion is that Dxo have adopted some sort of theoratical measuring method, which has absolutly nothing to do with the visual results that a camera or lens can produce. Really, stay away form their testing site. Their software is good, but they know shit about testing equipment in a practical relevant way.
If you want to know which camera is better for you, download raw files from each camera, once they are available and open them with your preferred raw converter. Now you can see which camera gives the best results for your workflow. www.imaging-resource.com is a good source for comparable raw files shot under the same lighting conditions.
The Pentax K5's RAW data is doctored, and even though DXOmark's graphs show that, they fail to mention it or take it into account in their "scores".
Look at the 18% SNR graph. The 5D mk II graph shows a textbook example of what a normal graph would look like, a smooth curve.
With the D3s we see a jump/angle at 12800, which shows us that there an EXTRA NR level is being applied. We can't tell what happened before 12800.
The Pentax shows an angle at ISO 1600.
With dynamic range it gets even more weird. The Canon 5D mk II shows a normal curve, the Nikon D3s a normal curve, but with its usual angle indicating extra NR. The Pentax K5 however shows a straight line. In short... doctored RAW data, fooling the "measurements" done by DXOmark.
[quote name='Brightcolours' timestamp='1288980006' post='3976']
The Pentax K5's RAW data is doctored, and even though DXOmark's graphs show that, they fail to mention it or take it into account in their "scores".
Look at the 18% SNR graph. The 5D mk II graph shows a textbook example of what a normal graph would look like, a smooth curve.
With the D3s we see a jump/angle at 12800, which shows us that there an EXTRA NR level is being applied. We can't tell what happened before 12800.
The Pentax shows an angle at ISO 1600.
With dynamic range it gets even more weird. The Canon 5D mk II shows a normal curve, the Nikon D3s a normal curve, but with its usual angle indicating extra NR. The Pentax K5 however shows a straight line. In short... doctored RAW data, fooling the "measurements" done by DXOmark.
[/quote]Thanks though I was not even thinking of selling my 7D or 5D II to buy the K-5, but also thought that these tests were kind of weird, @jenbenn I also have the 16-35 2.8 II and yes it is sharper when stopped down in the corners, a very good landscape lens on the 5D II 
Methinks there will be a lot of Canon and Nikon fans questioning DXOmark now that there is an interloper within their ranks! <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/laugh.gif' class='bbc_emoticon' alt=':lol:' />
Do you not believe that Hoya / Pentax with their medical image and software expertise cannot leverage this within the K-5s algorithms? Fact is, according to DXO, Pentax rules in the APS-C world until the next big thing comes along.
11-05-2010, 09:27 PM
(This post was last modified: 11-05-2010, 09:35 PM by Brightcolours.)
[quote name='robbiec' timestamp='1288987093' post='3978']
Methinks there will be a lot of Canon and Nikon fans questioning DXOmark now that there is an interloper within their ranks! <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/laugh.gif' class='bbc_emoticon' alt=':lol:' />
Do you not believe that Hoya / Pentax with their medical image and software expertise cannot leverage this within the K-5s algorithms? Fact is, according to DXO, Pentax rules in the APS-C world until the next big thing comes along.
[/quote]
What about RAW don't you understand though... RAW is supposed to be RAW, not processed by whatever.
DXO mark is supposed to "measure" the sensor performance, and it does not do that when it can't take into account processing done to make a sensor look better in these tests.
Klaus has indicated interest in this body and will be the platform for future reviews - is he on to something? We will have to wait for the in-depth reviews - until then any speculation or comments are just speculations.
[quote name='Brightcolours' timestamp='1288992423' post='3979']
What about RAW don't you understand though... RAW is supposed to be RAW, not processed by whatever.
DXO mark is supposed to "measure" the sensor performance, and it does not do that when it can't take into account processing done to make a sensor look better in these tests.
[/quote]
Bollocks! Nikon has been manipulating their RAW output for years, as have Pentax, Canon seems to be the standout here in that they do not. (I rephrased the last sentence due to factual inaccuracy)
You should have a look out for images by the K-5 made by user JohnBee on Dpreview or Pentaxforums, the sensor in the K-5 is an excellent starting point for image creation, Pentax are just extracting the absolute maximum possible within set variables which in turn are set by the industry, i.e. ISO, EV, etc, etc
If DXO mark is supposed just to measure sensor performance then it would have to bypass all manufacturer software and communicate with the sensor (chip) directly. Do you think Canon would hand over the keys to their sensor technology or Sony or Samsung for that matter just so a company can post some stats?
RAW is a file format generated by a combination of available light, lens, sensor, i/o, a/d convertors manipulated by an OS created by a manufacturer to achieve maximum detail at a given industry approved setting, which in turn then is read by various 'Raw Convertors' to display an image.
How they achieve this is up to them. How come you don't get upset about the various ISO settings banded about by manufacturers? Take a Canon 7D for example, its stated ISO1600 is actually measured to be ISO1223. Or Nikon's D300s, its ISO1600 is measured at ISO1114. Now look at the Pentax K-5, its ISO1600 is measured to be ISO1417. Are you saying that Pentax have managed to circumvent the industry standard of ISO, a known variable for example?
Fact is, Pentax have taken a Sony sensor, current generation or next generation (your choice) and used their experience and inhouse capabilities to deliver an excellent image producing tool.
I am seriously looking forward to getting my own K-5 in the next few weeks and combining it with my FA43f1.9, K50f1.2 and FA77f1.8 to create images that I am happy with and proud to share with others.
S!
Robbie
[quote name='robbiec' timestamp='1288987093' post='3978']
Methinks there will be a lot of Canon and Nikon fans questioning DXOmark now that there is an interloper within their ranks! <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/laugh.gif' class='bbc_emoticon' alt=':lol:' />
Do you not believe that Hoya / Pentax with their medical image and software expertise cannot leverage this within the K-5s algorithms? Fact is, according to DXO, Pentax rules in the APS-C world until the next big thing comes along.
[/quote]
Whatever, no one is denying that the K5 is a fine camera. It may well be equal or better than any current canon or Nikon product. But by how much? And will the differnce matter to practical photography? I'd say very likly it will not matter. Just as the iq differences between the aps-c cameras released in the last two years are unnoticable unless you prefer your images as 100% crops and not as full photos. Anyway, i find pentax offers very compelling. All that kept me from buying into their system was the lack of full frame.
I'm waiting for mine, it will be very interesting to see how it compares to the K-x when it comes to ISO performance.
|
