Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
09-19-2009, 04:41 PM   #1
Veteran Member




Join Date: Jun 2008
Location: Borås, Sweden
Photos: Gallery
Posts: 3,165
Dynamic range measurements on Dxomark - K20D - this doesn't seem right

Well, I am a bit puzzled. I know Dxomark isn't the end all, but was trying to explain a few things to somebody this evening and pulled it up to illustrate. One of the cameras picked was the K20D. Now, I remember that I thought it has quite good dynamic range when I owned one, but the Dxomark measurements seem to indicate otherwise. Any clue what's up with this? I have a very hard time believing the numbers compared to what I've seen with my own eyes in images.

09-19-2009, 05:26 PM   #2
Pentaxian
Wheatfield's Avatar

Join Date: Apr 2008
Location: The wheatfields of Canada
Posts: 10,157
I suspect DXO measures jpeg output rather than raw output. Their testing, like pretty much all testing being done these days is suspect in regards to whether they are really presenting anything that is either germaine to the picture making process or even accurate with regards to what they are striving for, and it gets in the way of the photography by presenting information that has little, if any bearing on photography.
09-19-2009, 07:23 PM   #3
Veteran Member
nostatic's Avatar

Join Date: Feb 2008
Location: socal
Photos: Gallery
Posts: 1,576
QuoteOriginally posted by pingflood Quote
Any clue what's up with this? I have a very hard time believing the numbers compared to what I've seen with my own eyes in images.
Machine tests all have bias. And they do not reflect reality. Trust your eyes, not a bunch of plots on a web site.
09-19-2009, 09:55 PM   #4
Veteran Member
jct us101's Avatar

Join Date: Jun 2008
Location: Rohnert Park, CA
Posts: 3,695
Dxomark has always been ridiculous.

09-20-2009, 03:39 AM   #5
Pentaxian




Join Date: Jun 2009
Location: GMT +10
Photos: Albums
Posts: 10,834
QuoteOriginally posted by Wheatfield Quote
I suspect DXO measures jpeg output rather than raw output. Their testing, like pretty much all testing being done these days is suspect in regards to whether they are really presenting anything that is either germaine to the picture making process or even accurate with regards to what they are striving for, and it gets in the way of the photography by presenting information that has little, if any bearing on photography.
I suspect there would be some benefit for people here in visiting the dxomark site and actually reading about how they do their testing.

DXOLabs ALWAYS and only measure on the basis of RAW.

Their testing methodology is also quite rigorous and based on realistic scenarios of real world use. It is reputable and as far as I am qualified to judge very professionally credible.

You have to remember DXO Labs do their testing not just to compile a database of benchmark information about camera sensors. That information is only a by-product.

Their testing is designed to feed information into the development of their imaging software - specifically their very good RAW converter software - so that it can be used to best effect when working with the RAW output of all the cameras (and lenses, which they also test and supply image correction options for) out there.

As to the dynamic range of the K20D it may looks quite OK of course. No one would argue with you there.

But relative to other cameras there are clearly technical and even visually perceptible differences that DXO Labs results legitimately highlight.

Ugh. Cavemen attitudes: 'What are you using a telescope for? We can see the stars fine without any stinking telescope. And why are you bothering to plot the stars on a chart? A telescope or your stupid chart don't reflect reality'.
09-20-2009, 05:06 AM   #6
Veteran Member




Join Date: Jan 2008
Location: Ste-Anne des Plaines, Qc., Canada
Posts: 2,014
QuoteOriginally posted by rawr Quote

Ugh. Cavemen attitudes: 'What are you using a telescope for? We can see the stars fine without any stinking telescope. And why are you bothering to plot the stars on a chart? A telescope or your stupid chart don't reflect reality'.
That kind of comment is completely uncalled for. Why insult people posting what they think. I don't see anybody insulting you. Do you?
09-20-2009, 07:52 AM   #7
Pentaxian
Wheatfield's Avatar

Join Date: Apr 2008
Location: The wheatfields of Canada
Posts: 10,157
QuoteOriginally posted by flyer Quote
That kind of comment is completely uncalled for. Why insult people posting what they think. I don't see anybody insulting you. Do you?
I probably insulted him at some time in the past.
Bad behaviour comes home to roost, I suppose.
This doesn't, however, alter my opinion that DXO and it's ilk are doing nothing that is especially useful with their published "tests".

Last edited by Wheatfield; 09-20-2009 at 08:31 AM. Reason: addition for clarity of thought
09-20-2009, 08:38 AM   #8
Pentaxian




Join Date: Mar 2009
Location: Perth Australia
Photos: Albums
Posts: 1,511
rawr: I heard 'science' isn't all that its cracked up to be!

You need faith in pentax to take good photos! hellauhja!

09-20-2009, 10:12 AM   #9
DAZ
Veteran Member
DAZ's Avatar

Join Date: Sep 2006
Location: Everett, WA USA
Photos: Gallery
Posts: 710
DXO is probably more then acceptably accurate but the real issue is what is it that you see that you think it not lining up with what you see in your K20D. It maybe a misunderstanding of what DXO is saying and how that relates to the real world.

Take the DXO DR for the K10D (11.6) and for the K20D (11.1). There are several things to consider just in looking at these 2 numbers. First to consider is that they are in Ev. This means that it is not linear. Consider going from a 10 to an 11 (one stop) takes 2 times as much light. So from 11.6 to 11.1 is .5 or half a stop. Is a half a stop something you are likely to see? Probably not. Although one whole stop is probably something that could be perceived with a discerning eye it still may not in the real world be all that significant. To get that 1 stop you need to expose all the way to the right. Then you probably need to do a little PP to make it look like you want it to. If you are off by just that one stop in your exposure you gave up that 1 Ev of DR and with the K10D to K20D it is only .5 Ev. Now this kind of thing can be extremely important to some but for a lot of photographers they don’t even look to see if when they take a photo that they are even 1 Ev let alone .5 Ev from the right. Now a DR of some thing like Ev of 11 is good. If you ware going from a DR of lets say Ev 6 to EV 11 then 11 would be fantastic. But to go from Ev of 11 to 12, well you can measure it but for a lot of people (those without the discerning eye) they will not see it. There are probably a lot of other things to think about when buying a camera then just that half stop. As an DR Ev of 11 is good to really see some thing look at a DR of an Ev of 15–16. At that point things will start to look unreal.

Just because there is a difference in a number doesn’t mean that there is a significances to you and I don’t think that DXO is trying to tell you that it is. They are only giving you the information. It is up to us to detriment how to weigh the information relevant to what we need.

This is true of a lot that is happening it the world now. We have gotten so good at measuring things that just because we can measure something doesn’t necessarily mean anything.

DAZ
09-20-2009, 10:12 AM   #10
Senior Member




Join Date: May 2009
Location: Delaware
Posts: 177
DXOmark measures raw sensor output to the extent that they have downgraded their noise results for several Pentax models, including the K20D because they claim that the higher ISOs are smoothed a bit, even for the RAW images. So what? What counts is how the camrea delivers images and how much headroom the RAW files have. Look at the Imatest results from the Imaging Resources K20D review - the K20D (RAW processed throught ACR) comes out in the top tier of cameras tested, although just making it. It is just behind the Nikon D3 while beating the Nikon D300. They also mention the retention of fine detail that is smoothed away by the competion. In their printing tests the camera does quite well even at poster size prints. I don't know what else one could want from a camera.
09-20-2009, 10:58 AM   #11
Pentaxian
Wheatfield's Avatar

Join Date: Apr 2008
Location: The wheatfields of Canada
Posts: 10,157
QuoteOriginally posted by DAZ Quote


.... Consider going from a 10 to an 11 (one stop) takes 2 times as much light. So from 11.6 to 11.1 is .5 or half a stop. Is a half a stop something you are likely to see? Probably not. Although one whole stop is probably something that could be perceived with a discerning eye it still may not in the real world be all that significant.
Consider also that the EV range of the "average" landscape generally won't exceed six stops( if this was not the case, slide film would not work), and the scenes that do are usually used by measurbators to show how badly their lens handles chromatic aberration, not to show how badly their sensor handles a long range scene.
So, for most of us, it is really immaterial if a sensor (and it's cohabiting firmware) can provide 10 stops, 11 stops or more of exposure range. It just doesn't matter.
To paraphrase a previous post, what is the point of charting the stars if you are doing nothing beyond plotting them onto a chart?
09-20-2009, 11:37 AM   #12
Veteran Member
nostatic's Avatar

Join Date: Feb 2008
Location: socal
Photos: Gallery
Posts: 1,576
QuoteOriginally posted by rawr Quote
Ugh. Cavemen attitudes: 'What are you using a telescope for? We can see the stars fine without any stinking telescope. And why are you bothering to plot the stars on a chart? A telescope or your stupid chart don't reflect reality'.
And those that focus on machine test data as opposed to the actual output that will be viewed by a human often understand numbers but are flummoxed by art.

DxO admits that, for instance, they don't take resolution into account. So if you focus on their tests, the D700 has the lowest noise. But humans don't look at plots, we look at pictures. When you take a "higher noise" but also higher res camera and create equivalent outputs, all the sudden the numbers don't really tell the story.

Use technology where appropriate. But it is not the final word. Until Skynet becomes aware at least...
09-20-2009, 08:01 PM   #13
Veteran Member




Join Date: Jun 2008
Location: Borås, Sweden
Photos: Gallery
Posts: 3,165
Original Poster
QuoteOriginally posted by DAZ Quote
DXO is probably more then acceptably accurate but the real issue is what is it that you see that you think it not lining up with what you see in your K20D. It maybe a misunderstanding of what DXO is saying and how that relates to the real world.
I do understand the numbers and how they reflect upon "real world" and that's why I am so surprised to see them and what they imply. If there was a .5 stop difference in DR, yes, I would not expect much visible difference. But at ISO 400 it's a 1.5 stop difference which in high dynamic range scenes SHOULD be visible. I have never run side by side comparisons in person since I sold the K20D soon after changing brands, but still thought it performed admirably in high DR situations.
09-20-2009, 08:59 PM   #14
DAZ
Veteran Member
DAZ's Avatar

Join Date: Sep 2006
Location: Everett, WA USA
Photos: Gallery
Posts: 710
QuoteOriginally posted by pingflood Quote
I do understand the numbers and how they reflect upon "real world" and that's why I am so surprised to see them and what they imply. If there was a .5 stop difference in DR, yes, I would not expect much visible difference. But at ISO 400 it's a 1.5 stop difference which in high dynamic range scenes SHOULD be visible. I have never run side by side comparisons in person since I sold the K20D soon after changing brands, but still thought it performed admirably in high DR situations.
All I can say is that a scene may not be a high DR scene or it may have been a high DR but you did not care so you did not notice. The eye is not a good judge of DR as it is always making adjustments so unless you used your camera to measure it you may only think it is a high DR. Next if you do not blow the highlights in a high DR photo what you get is a little more shadows and at higher ISOs you get more noise in the shadows. So unless you are trying to pull stuff out of the shadows a high DR and a lower DR photo can look the same. DXO doesn’t care that you like the shadows for artistic reasons only how much signal to noise there is. I am not saying you personally can’t see these things only you may not have been looking. So if you got the top 9 of a scene and lost the bottom 4 is it some thing you would always miss?

Here is a photo I took the other day in a bar. The bar was not bight so it did not look to my eye like a high DR scene but it is IF I had cared about the shadows. If I had tried to pull up the shadows I would probably have seen the limit of the cameras DR. As it is I am probably only using 1/2 to ¾ of the DR of the camera. If I had used ISO 400 instead of ISO of 100 about the only thing that would look different is more noise in the shadows but the dark to light would look about the same. The thing about DR is in the shadows so are you looking at how much you can pull the shadows up?


DAZ
Attached Images
View Picture EXIF
PENTAX K20D  Photo 
09-21-2009, 11:53 AM   #15
Senior Member




Join Date: Jun 2008
Location: Atlanta, Georgia
Photos: Gallery
Posts: 182
According to DxOMark my K10D has as good dynamic range as the top of the line Hasselblad - please don't tell me that is not true
Attached Images
 
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
camera, dslr, dxomark, k20d, measurements, photography, range
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Dynamic Range of K20D and K200D cbaytan Pentax DSLR Discussion 8 05-22-2009 08:04 AM
Dynamic range 200% K20D knumbnutz Pentax DSLR Discussion 13 03-29-2009 03:53 AM
dynamic range K20D dafiryde Pentax DSLR Discussion 4 01-08-2009 10:51 AM
k20d or k10d - which has better dynamic range Michael Barker Pentax DSLR Discussion 10 12-18-2008 10:47 AM
K20d Dynamic Range? Don Boyd Pentax DSLR Discussion 2 04-07-2008 04:07 AM



All times are GMT -7. The time now is 08:20 AM. | See also: NikonForums.com, CanonForums.com part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top