Why the DxO bashing?

[Loswr]

If that's not the case for you, I'd suggest the tools aren't the problem, but rather the tool user.

Nope, that's not the case. I think, perhaps, you've conflated some test methods with actual shooting, hopefully not merely for dramatic effect.

.. only the 7D and 5D2 gave me low ISO problems with FPN. Pity, I really liked the 7D otherwise, too.
So, those tools did not perform to my requirements, and off they went.

Can you get how that puts into perspective why I found my 7D, and especially the 5d2, "disappointing?" A hint if you're missing it, they're also the 2 highest priced bodies of the bunch.

I owned both the 7D and the 5DII, and in tens of thousands of shots I didn't have a single one ruined by FPN (note: none were shots with the lens cap on). Your cameras may have been defective, or I had cameras with magic FPN-proof sensors, or I just exposed and processed my images properly. The first is possible, the second pretty implausible, the third is most likely.

Specifically regarding the 7D, Roger Clark stated, "Thye 7D camera has lower fixed pattern noise at ISOs less than 800 than many other Canon cameras tested, including the Canon 1DX." Maybe Roger has a 7D with one of those magic sensors, too? Not that I have any issues with my 1D X, either...

But you can keep on blaming your tools, if it makes you feel better about your inability to use them properly.

 

[Marsu42]

t ignore this post, but I'm going to reply because you may give the impression to someone reading it that the 5D digic 4 cameras are in some way inferior to the others, and this is just not the case.

I know I should also ignore this post but reading this people might think it's about the digic processor, but afaik the pattern noise has more to do with the general layout of the camera (pcb), the number of readout channels and how fast the data is read. Canon seems to have concentrated on improving this problem though, at least from what I can tell using 60d->6d.

 

[privatebydesign]

If that's not the case for you, I'd suggest the tools aren't the problem, but rather the tool user.

Nope, that's not the case. I think, perhaps, you've conflated some test methods with actual shooting, hopefully not merely for dramatic effect.

I think, as always, you are the one aiming for dramatic effect, after all you have taken Neuro's quote completely out of context, which was, a badly exposed image, something we know you are prone to.

People who found the fpn from the 5D MkII and 7D to be particularly problematic are the ones who tended to underexpose, which is outdated advice originally intended to preserve highlights from clipping, and who then who didn't take any time at all in working out the best way to process those underexposed files.

Remember when you said "I'd like to see what you can do to try a stripey 7D file under the same conditions, and see how much work you'd have to put in, and what kind of results could be obtained."

Well my reply still stands "Send me some RAW files, 7D and/or 5D MkII, I don't care, I'll even do another video on what I did to them."

You didn't get the results you wanted because you didn't take the time to learn to use them. We could turn your comment "Can you get how that puts into perspective why I found my 7D, and especially the 5d2, "disappointing?" A hint if you're missing it, they're also the 2 highest priced bodies of the bunch." 180º, have you noticed the cameras you get better results with are the ones programmed to look after people who don't know what they are doing? The bodies where user input is far more important you can't get results from. Now what does that say?

 

[sdsr]

I'd like to see what you can do to try a stripey 7D file under the same conditions, and see how much work you'd have to put in, and what kind of results could be obtained.

[....]

And yes, I'm no Photoshop guru, but neither should I have to be. Far quicker and better for future-proofing to just choose better tools that don't require me to fix such things in post.

Since this is a thread about DxO (albeit a different branch of it) I though I would mention the improved "prime" (I think that's what they call it - I don't have it here in my office) noise reduction component of the latest version of their software. The other day I ended up with a badly underexposed photo of one of our cats, who was posing in a rather dramatic way - I inadvertently took the first shot, with bounce flash, before I was in the same room as he was in, so of course it bounced in the wrong place, missed him altogether and I ended up with a rather dark image. I rather liked the result nevertheless, but also thought I would see what would happen if I tried to brighten it. As I had taken it with my 5DIII rather than 6D, there was a little visible banding in the noise which I couldn't quite remove in LR (not that I tried terribly hard). I then tried it in DxO, selecting "prime" noise reduction mode and the banding vanished completely, and automatically, with no further tweaking on my part. Whether that would help with the banding you're complaining about I have no idea (I've never used a 7D, and have no idea how underexposed your problem photos are, let alone how much you want to brighten them), but what it did for me certainly didn't require guru status.

 

[Loswr]

The other day I ended up with a badly underexposed photo of one of our cats...

Dammit, man, you can't just say that and then not post your cat photo on the Internet!!

 

[gruhl28]

Perhaps this thread should be put out of its misery - well, after the cat picture is posted. This has to have the highest ratio of arguing to useful, informative comment of any thread I've read on this forum.

 

[Orangutan]

This has to have the highest ratio of arguing to useful, informative comment of any thread I've read on this forum.

I see you're new around here.

Welcome to the forums!

 

[sanj]

This has to have the highest ratio of arguing to useful, informative comment of any thread I've read on this forum.

I see you're new around here.

Welcome to the forums!

hahahha

 

[JumboShrimp]

Original Poster (OP) is back and thank you all for the (sometimes) informative, but always entertaining, discussions. I know I'm bucking the trend, but I sort of like a single numerical equipment rating. Bottom line evaluation, so to speak.

There was little or no comparison to what the folks do at DPReview, though. Seems to me that DxO and DPReview basically have the same procedural rating system. DxO gives a number, whereas DPReview gives a percentage and an award, such as 89% Gold Award. My impression through the CR forum is that DPReview is the lesser of the two evils, but honestly don't know why when they are so similar to DxO.

Any opinions/comments comparing methodology and results between DxO and DPReview?

 

[Loswr]

My impression through the CR forum is that DPReview is the lesser of the two evils, but honestly don't know why when they are so similar to DxO.

Any opinions/comments comparing methodology and results between DxO and DPReview?

The big difference is that DPR's % score attempts to rank the camera (build, ergonomics, IQ, AF, etc.), whereas DxOMark's score is for the sensor and only the sensor. Pair a great sensor with poor autofocus, you get great DR and low noise...and a blurry image. DxO doesn't care, to them it's still great. DPR would mark down the overall score due to the poor AF.

Since people buy cameras, and not bare silicon sensors, DPR's single number score is a bit less useless than DxOMark's sensor score.

 

[jrista]

My impression through the CR forum is that DPReview is the lesser of the two evils, but honestly don't know why when they are so similar to DxO.

Any opinions/comments comparing methodology and results between DxO and DPReview?

The big difference is that DPR's % score attempts to rank the camera (build, ergonomics, IQ, AF, etc.), whereas DxOMark's score is for the sensor and only the sensor. Pair a great sensor with poor autofocus, you get great DR and low noise...and a blurry image. DxO doesn't care, to them it's still great. DPR would mark down the overall score due to the poor AF.

Since people buy cameras, and not bare silicon sensors, DPR's single number score is a bit less useless than DxOMark's sensor score.

It's also a lot clearer that DPRs ratings (which are conveniently percentages, something everyone fully understands) are subjective, based on the reviewers experiences as well as technical tests with the camera.

This is in contrast to some arbitrary number that requires you to go investigating HOW that number is derived, something the very vast majority of DXO viewers DO NOT do. That scalar number is a black box output that does not factor in enough information in order to be truly accurate, and yet it is boldly claimed to be "scientific". It may well indeed be produced via a scientific process, but the number is otherwise utterly meaningless, yet given all to much precedence, by the unwary general public.

That's the danger of DXO...their bold claim to science and yet black box effect.

 

[Aglet]

..you are basically referring to DSLR cameras that were released between February 2005 to August 2010.

Yes.
The 6D and 70D are noticeably improved for low ISO pattern noise so my gripes are confined to Digic 4 SLR bodies. OTH, Digic 4 PowerShot G11 (& G12?) behave pretty well.

[..keep on blaming your tools, if it makes you feel better about your inability to use them properly.

If 20+ other Canon bodies (let's not even consider the Exmor sensored bodies), often used the same way, did not produce objectionable FPN when pushed then how can you conclude that's a user fault? The 7D is KNOWN to have stripey shadows with only a small push that you can even do in DPP. Too bad you don't have yours yet so you could provide a lens cap shot so we could see if it had stripes or not.

e.g.
7D non-pushed
www.canonrumors.com/forum/index.php?topic=9299.msg169599

and crop from same slightly pushed file in DPP
www.canonrumors.com/forum/index.php?topic=9770.msg176368#msg176368

..pattern noise has more to do with the general layout of the camera (pcb), the number of readout channels..

very true. I've identified image noise under some circumstances with my old 40D that seems to be directly related to electronic system noise, most likely an onboard voltage regulator. If I could clean that power supply up i'd have more useful 1600 and 3200 iso on that one.
Digic 4 is only fingered as being the processor in the most egregious DSLR FPN culprits, PowerShots G11 & 12 are Digic 4 and cleaner than the SLRs at base ISO.

People who found the fpn from the 5D MkII and 7D to be particularly problematic are the ones who tended to underexpose, which is outdated advice originally intended to preserve highlights from clipping..

I don't understand how that is outdated advice.
Keeping highlites just short of clipping is how to retain highlite detail; they're not being exposed as a midtone.
Push the rest up as desired, or even further, as in this example from my 60D which survived just fine with a manually exposed shot to retain cloud detail while pushing the rest up in post.

www.canonrumors.com/forum/index.php?topic=8065.msg154889#msg154889

Well my reply still stands "Send me some RAW files, 7D and/or 5D MkII, I don't care, I'll even do another video on what I did to them."

I haven't forgotten. when I can make the time I'll prep a file for you to work on. I'd like to see if you can process the stripes out without losing detail. Likely will be one of the 7D sunset shots from link above.

.. have you noticed the cameras you get better results with are the ones programmed to look after people who don't know what they are doing? The bodies where user input is far more important you can't get results from. Now what does that say?

it says nothing conclusive. I generally shoot difficult scenes in manual exposure to retain highlite detail levels where I want them so how can an unused comsumer camera's AE features possibly matter?

tried it in DxO, selecting "prime" noise reduction mode and the banding vanished completely

I do have DxO 9 but haven't run any of the old stripey files thru it. I have seen improved results on some of my older hi ISO files that it did a nice job on. Thanks for letting me know it worked for you; I have some 5d2 files and 7d files I hope it can fix without excess time spent doing so.

 

[Rienzphotoz]

Only in CR or similar forums did I ever come across a few people talking about what kind of noise is there in some dark/unimportant area of the image or some posterization in the OOF area etc ... unfortunately, now I've begun to look for those issues, instead of concentrating on the most important aspect of the image i.e. subject matter, the message the images conveys, composition etc :-[
I need to get back to what's important to me in an image.

 

[jrista]

[..keep on blaming your tools, if it makes you feel better about your inability to use them properly.

If 20+ other Canon bodies (let's not even consider the Exmor sensored bodies), often used the same way, did not produce objectionable FPN when pushed then how can you conclude that's a user fault? The 7D is KNOWN to have stripey shadows with only a small push that you can even do in DPP. Too bad you don't have yours yet so you could provide a lens cap shot so we could see if it had stripes or not.

e.g.
7D non-pushed
www.canonrumors.com/forum/index.php?topic=9299.msg169599

and crop from same slightly pushed file in DPP
www.canonrumors.com/forum/index.php?topic=9770.msg176368#msg176368

The 7D is by far at it's weakest at ISO 100. I think Neuro is more like myself in that more often than not, he's shooting at a higher ISO. Past ISO 400, banding is pretty much non-existent, meaning all of the ISO settings between 400 and 3200 are pretty usable. By ISO 3200 itself and again the camera isn't all that usable.

I don't think anyone denies that the 7D has a banding problem at low ISO. That's well known. At ISO 400 sometimes you don't even need to push anything at all, and banding can be a slight problem in the midtones.

The 7D isn't really a landscape or studio camera, though. It's an action camera. It's an ok one, but lacking the very high ISO capabilities of a FF camera, it's limited in it's usable scope in that arena. The 70D has demonstrated some clear improvements in the area that the 7D used to dominate. It definitely has less noise, it's sharper, more usable at ISO 400 and 3200 (even though it actually has slightly more noise, it's less revolting noise). Not by a huge margin, but by enough of a margin.

I think in the long run, between the improvements made in the 70D and even more so the improvements made in the 6D, the next DSLRs from canon should be pretty good on the noise front. If there was ever a "biggest complaint" against the 7D, it would be it's poor handling of noise, in general. Second to that would be the perceived softness due to the AA filter. (Ironically, I personally love the 7D's AA filter, as it's a godsend for bird photography...no moire at all, especially with a big white...but most people are limited to smaller/cheaper lenses, so I understand the outcry for a weaker AA filter.)

 

[Rienzphotoz]

...
I think in the long run, between the improvements made in the 70D and even more so the improvements made in the 6D, the next DSLRs from canon should be pretty good on the noise front.
...

Well for that to happen, Canon would have to not recycle the sensor from both of those cameras into others. Hands up all those that think the sensor in the 70D won't appear elsewhere without modification?

If customers are willing to buy their "recycled" 70D sensor, what is the problem?

 

[Loswr]

If customers are willing to buy their "recycled" 70D sensor, what is the problem?

Canon has sold tens of millions of cameras with 'recycled' 18 MP sensors.

 

[Chosenbydestiny]

I think people overthink this stuff. If your client or your audience worked for DXO or are have a serious mental disorder concerning pixels, sure, you may have an issue. But since content still rules all.... As long as you can deliver results to your audience's standards, everything else doesn't matter. If you are your own audience, you are welcome to argue with yourself. ;D

 

[Rienzphotoz]

If customers are willing to buy their "recycled" 70D sensor, what is the problem?

Canon has sold tens of millions of cameras with 'recycled' 18 MP sensors.

Are you suggesting that DxO worshippers are jealous that Canon has "sold tens of millions of cameras with 'recycled' 18 MP sensors"? if so, I agree ;D ... I think they are just jealous that their "oh so superior DR capable" senors aren't flying off store shelves as much as Canon's "recycled" sensors ;D

 

[Loswr]

If customers are willing to buy their "recycled" 70D sensor, what is the problem?

Canon has sold tens of millions of cameras with 'recycled' 18 MP sensors.

Are you suggesting that DxO worshippers are jealous that Canon has "sold tens of millions of cameras with 'recycled' 18 MP sensors"? if so, I agree ;D ... I think they are just jealous that their "oh so superior DR capable" senors aren't flying off store shelves as much as Canon's "recycled" sensors ;D

Mostly I think they're frustrated that tens of millions of people seem to ignore what they perceive as the only important feature of a camera, namely an extra two stops of low ISO DR.

 

[Loswr]

Well if Canon can do that then they'll have a lens that has closer to linear performance with respect to DR along the ISO scale (a linear degradation of DR with ISO is concurrent with theory on the matter.)

Touché.

I'd almost be willing to bet that IQ won't be specifically addressed in the next round of sensors for FF either because Canon will have been focusing R&D efforts on getting DPAF working on FF sensors instead.

If Canon felt that sensor IQ needed to be addressed, they'd have done so. They've had a low ISO DR gap for years, and for most of those years they gained market share at the expense of their competition with more low ISO DR. Those who incessantly beat the low ISO DRum can't seem to grasp that what matters most (or exclusively) to them is far less important (or even irrelevant) to the vast majority of camera buyers.