Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
DxOMark For Pentax K-5.
#21
[quote name='Kunzite' timestamp='1289052925' post='4000']

Hard to accept, isn't it? <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/smile.gif' class='bbc_emoticon' alt=':lol:' />



jenbenn, you're correct - this is revolutionary. But what one should do is to double and triple check; not just rejecting the result he doesn't like. I already did it, and the K-5 is showing it's impressive DR.



Brightcolors, and how would a slight NR applied at ISO 3200+ affect the DR at ISO80?

[/quote]



This has nothing to do with acceptance. Rather with logic and critical interpretation. If you believe this test, because you want to beleive it, its fine. Its like religion, you can believe anything you want without reason. For my part, I dont like a religious attidude when it comes to scientific tests. If a test procedure has shwon in the past, that its results differ from the real life capapbilities of camera/lens, I have to conclude that the method used is inadequat or flawed.



As I have pointed out on the example of the Canon 16-35 lens Dxo results contradict photzone's test of the same lens,not by marginal difference, but by claiming the lens to have a complete differnt characteristic. (According to DXo sharpness of the lens DEcreases when stopping it down while photozone, slrgear, lenstip and all other testers including myself have found that the Sharpness INcrease when stopping it down) It makes me wonder that Dxo didnt bother to check their results after they have found such an extraorinary behaviour of the lens. I have yet not heard of any lens that looses sharpness when stopping down (the exception beeing that the camera/lens combination is difraction limited at the widest aperture, of course)



They also try to tell me that the canon 40D produces better raw files then the 50D. No it doesnt. If you open the files from each camera in the same raw converter, print them the same size and look at the prints, you#ll find the 50D to be slightly better or equal, depending on the subject).



This leads me to conclude that their test are either unreliable and flawed or has no significane for practical photography.



I would be very happy if the pentax had this high a dR. It actaully may well have. Its just that DXO is incompetent to tell. When imaging resource, dpreview and maybe photozone had a look at the cam we may know more. Until then noone can tell.
#22
[quote name='Kunzite' timestamp='1289061058' post='4004']

Nonsense, like your post about "doctored" RAW? You made no effort to understand how and to which extend the RAW is "doctored", and you're completely ignoring no NR was detected below ISO3200. If you really want to pointing out how flawed the DXOMark is, then point out to it's own shortcomings (e.g. completely ignoring image definition), don't blame the camera(s). This way, you're just trying to reject their findings without providing any argument - at least, that's how it looks.



Defensive reactions, like yours? <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/wink.gif' class='bbc_emoticon' alt=':lol:' /> I understand your skepticism; but you should also accept the K-5 might be that good.



Yes, I really mean that. You're mistakenly presuming I'm talking about carelessly using the full 14.1EV for every picture, but of course this isn't the case. I wouldn't want to get "contrastless bland images". I'm not a big fan of HDR, either <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/wink.gif' class='bbc_emoticon' alt='Smile' />

But let me tell you, being able to underexpose an image by 9.5 stops (@ISO80) and recover it via software, is certainly impressive. And a definite proof they don't use NR at such low sensitivities.

[/quote]

DXOmark almost never detects RAW NR. And again, I did NOT blame any camera. SIlly stuff.



There is a VERY clear pointer to some kind of NR, the strange/weird totally straight line in their DR graphs. That is not normal, and it shows something is being done to improve DR measurements.



I gave very clear arguments, read my posts again. Im not the one trying to defend the K5 (you are) when it is not even being attacked.



Anyway. Care to show any 9.5 stops underexposed images? And the recovered versions? Without any examples it is a bit odd to talk about it.
#23
[quote name='jenbenn' timestamp='1289061463' post='4005']

Has nothing to do with acceptance. It has to do with the credibility of the site. DXo tests are senseless as I have pointed out on the example of the Canon 16-35 lens (Dxo results contradict photzone's test of the same lens,not by marginal difference, but by claiming the lens to have a complete differnt characteristic. (sharpness DEcrease when stopping down vs. Sharpness INcrease when stopping down ) This leads me to conclude that their test are unrelaible and flawed.



I would be very happy if the pentax had this high a dR. It actaully may well have. Its just that DXO is incompetent to tell. When imaging resource, dpreview and maybe photozone had a look at the cam we may know more. Until then we cant tell.

[/quote]Let me put it another way, when DxOMark gave the 7D a better score than the K-7 everyone in the pentax forums were blabbing that they don't believe DxO and such and they just couldn't except the fact that the 7D is a much better camera, but now it is a different story they worship DxO, and even started saying it (the K-5) is a FF killer and so on. Myself I would much rather wait for real side by side comparisons.
#24
[quote name='jenbenn' timestamp='1289061463' post='4005']

Has nothing to do with acceptance. It has to do with the credibility of the site. DXo tests are senseless as I have pointed out on the example of the Canon 16-35 lens (Dxo results contradict photzone's test of the same lens,not by marginal difference, but by claiming the lens to have a complete differnt characteristic. (sharpness DEcrease when stopping down vs. Sharpness INcrease when stopping down ) This leads me to conclude that their test are unrelaible and flawed.



I would be very happy if the pentax had this high a dR. It actaully may well have. Its just that DXO is incompetent to tell. When imaging resource, dpreview and maybe photozone had a look at the cam we may know more. Until then we cant tell.

[/quote]

Imaging resource does not do a good job at all, regarding "testing" the IQ and IQ possibilities. The only thing that is a little bit useful from them is the studio RAW files.



DPReview is not all that much better, they do not even know how to test DR, and their noise assessments are failed too.



Photozone does not do a much better job either, if you really want to compare IQ from different camera bodies...



So even after they review, we still can not really tell. <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/huh.gif' class='bbc_emoticon' alt=':lol:' />
#25
[quote name='Brightcolours' timestamp='1289062638' post='4008']

Imaging resource does not do a good job at all, regarding "testing" the IQ and IQ possibilities. The only thing that is a little bit useful from them is the studio RAW files.



DPReview is not all that much better, they do not even know how to test DR, and their noise assessments are failed too.



Photozone does not do a much better job either, if you really want to compare IQ from different camera bodies...



So even after they review, we still can not really tell. <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/huh.gif' class='bbc_emoticon' alt=':lol:' />

[/quote]

you are right in a way. The best way to evaluate a camera is of course to shoot the same subject, under the same light with all cameras, open the files in one's preferred raw converter, adjust for MP differences and print the pictures. Then you can compare and find the best camera for YOUR workflow.
#26
[quote name='Brightcolours' timestamp='1289062293' post='4006']

DXOmark almost never detects RAW NR. And again, I did NOT blame any camera. SIlly stuff.

[color="#0000FF"]"The Pentax K5's RAW data is doctored". Silly stuff indeed.

By the way, I only asked how a NR applied at ISO3200+ would affect the data measured for ISO80.[/color]



There is a VERY clear pointer to some kind of NR, the strange/weird totally straight line in their DR graphs. That is not normal, and it shows something is being done to improve DR measurements.

[color="#0000FF"]Hold your horses. Are you telling me that, just by looking at that line, you can say the K-5 is applying NR at every ISO?[/color]



I gave very clear arguments, read my posts again. Im not the one trying to defend the K5 (you are) when it is not even being attacked.

[color="#0000FF"]No, you haven't. Just claiming the RAW is "doctored" is not an argument, and I won't simply take your word on it. I want to know what "doctored" means; the word itself is meaningless for me.

Also, I never said you're defending the K-5.[/color]



Anyway. Care to show any 9.5 stops underexposed images? And the recovered versions? Without any examples it is a bit odd to talk about it.

[color="#0000FF"]Sure, it was done on dpreview: http://forums.dpreview.com/forums/read.a...e=36806297

For me, such shadow detail means they're not applying any NR @ISO80 - at least not on shadows (but would they need it, on mid-tones/highlights?)[/color]

[/quote]



[quote name='Brightcolours' timestamp='1289062638' post='4008']

Imaging resource does not do a good job at all, regarding "testing" the IQ and IQ possibilities. The only thing that is a little bit useful from them is the studio RAW files.

[color="#0000FF"]I find their RAW files to be quite useful.[/color]



DPReview is not all that much better, they do not even know how to test DR, and their noise assessments are failed too.

[color="#0000FF"]But they're good at showing the camera's interface; I'd rather look at their reviews, than search in the manual <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/wink.gif' class='bbc_emoticon' alt='Smile' />[/color]



Photozone does not do a much better job either, if you really want to compare IQ from different camera bodies...

[color="#0000FF"]I'm more interested in lens tests, from them.[/color]



So even after they review, we still can not really tell. <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/huh.gif' class='bbc_emoticon' alt=':lol:' />

[color="#0000FF"]All we can do is to take a hint from here, one from there...[/color]

[/quote]
#27
I explained exactly why we can see the RAW is doctored (from the Pentax K5), and that as such (again) the sensor score system of DXOmark is as nonsensical as ever. That says NOTHING about the Pentax K5 itself, even though you seem to think it does.



Oh well.



Thanks for the example, do you have any comparable examples from other cameras?
#28
I can't but agree the DXOMark score is nonsensical, as is any other system that tries to describe a camera by a single number. Yet one can find useful information, by looking carefully at what and how they're measuring.

The DR rating was already confirmed (for me), so it's not like I trust them (more than any other synthetic benchmark).



"Doctored" as in "To falsify or change in such a way as to make favorable to oneself"? If you said something like that about me, I would be offended; is it just my bad command of English, or a bad choice of words and broad generalization from your part?

I see no evidence of anything like this below ISO3200, and just a hint of NR above (details are retained up to ISO51200, which IMO is amazing for a 16MP non-FF camera). For all ISO range the noise is easy to work with, there's no "bad" NR algorithm applied that would transform it into color blotches. I'm saying this after studying images and/or talking with people who did the same. Then, what is "doctored"?

Maybe you're just not used with sensors with such a low read-out noise? But it doesn't mean they're "doctoring" RAWs.



To try anything like this, never crossed my mind. Not with my unforgiving K20D, anyway.
#29
[quote name='Sylvain' timestamp='1289041552' post='3994']

Thw, Gabor died?? when did it happen? I used to read his posts -not understanding him fully most of the time I must admit- with such a huge interest...

Sad news.[/quote]



[quote name='Christos' timestamp='1289041750' post='3995']

Gees just read that so sad, I hope he wasn't young.[/quote]



Gabor wasn't young. But it's all quite sudden for most of us. I only read it in DPReview couple of months ago. Sigh...
#30
Dpreview has published raw and jpg test shots of d7000, K5 and 60D. What I found intersting is that the d7000 and k5 seem indeed to hold more detail in the shawdos than the 60d . This is however only apprent in the dark box in the photos which contains these wool threads. Otherwise all cameras (except the 50D) look more or less identical, with the D7000 being abit less noisey but also a bit less sharp than the 60D at high iso.
  


Forum Jump:


Users browsing this thread:
1 Guest(s)