D7000 replacement in April...

[David Hull]

Interesting points about the banding noise, my older XXD Canon sure has tons of it above ISO 2500.  I need to try some software to fix some of my work, that's for sure.  Photoshop works well until the banding starts.

Notice in your link to Nikon's own data, that the DR falls to about 10 stops by ISO 800.  I mean, who really cares if you've got 13, or 14 stops below that?  What counts is what's above ISO 800. 

If you're shooting below ISO 800, you're still going to use highlight tone priority if you have elements in the image which are extremely bright and you don't want them to look too pale or blown out...or at least I do.  That is COMPRESSING the dynamic range of what gets stored on the file.  You certainly don't need that many stops down where the noise is lowest anyway...because a jpg is only 8 bits.  Does this sound wrong to you?  If so, what would you do with a "future" sensor that was 16 or 18 stops at ISO 100?  Shoot raw, then print from a 48 bit upsampled TIFF file (probably 20 GB in size) and display the resulting "print" on a transparent backlit, multi-stage screen using some exotic led's or xenon lamps?  Get real.  Our eyes see huge dynamic range, but print media can't display it.  They can only display a dynamically compressed approximation, as in HDR technique, etc.  It's called "painterly" for a reason.  The reason is dynamic range compression.

You might give the NR plug-ins from Topaz or Nik a try but I don't think they were effective on the two dimensional pattern noise that was present in the older cameras.  On the newer stuff like the 5DIII and 6D where the noise pattern is much less and only vertical, they appear to be quite effective (at the expense of some slight detail reduction).

To me the only real benefit of the additional DR is that you can pull the shadows a bit more on those cameras w/o worrying about pattern noise -- there are a few cases where this would be beneficial.  However, to get any benefit from this at all, you will need to run your camera with the ISO set at ISO 400 or less (preferably ISO=100).  I have always found it interesting that in the majority of the examples provided, the shot could be made with either camera if it were used correctly.  By that I mean that in almost every case, the shooting scenario does not preclude the use of a multi shot approach -- take a look at the examples provided in this thread.

[jrista]

yes, You can believe that if its suits you.
this is my 5dmk2 and my d800, exposed in the same way  to get the high lights in the sky intact
then the darker parts are lifted and adjusted the same. The results from the Canon is visible more pattern noise , banding, lower signal , less color information.

You should run the 5D II RAW image through Topaz DeNoise 5, and use the debanding feature. You might be surprised how much DR is recovered. The blacks may not be as rich, as Canon uses a bias offset which eats a chunk of shadow DR, but the noise would very likely disappear. DeNoise has shadow tone recovery features as well, which could restore or deepen some of those shadow tones. Once you eliminate the banding noise, Canon sensors fare a hell of a lot better against there competitors.

[CarlTN]

Interesting points from all of you.  I do need to try Topaz...however mine is currently an older XXD body, not a 5D2.

Mikael, that's a good illustration of your point.  However, in that situation, multi-shot HDR is really more called for, in my opinion.  My body has a bit worse noise than the 5D2...hahaha.  How would the D3s compare using similar under-exposed technique at low ISO, to the D800 (accounting for image dimensions, however you might want to do that)?

My cousin has the 5D3, and he mostly uses Lightroom 4 (although he has CS6).  I tried to reduce some of the "grain" noise on a couple of his RAW shots done at around ISO 4000...on his computer using LR4.  I must say, LR4 is far less effective at reducing that, than CS5 ACR is, on similar grain noise on my own files, done with my older body.  Of course, those have similar amounts of noise way down at ISO 1250...haha.  His files were exposed normally, not under-exposed.  This grain noise seems to have a hard texture...sort of like you're trying to sand it smooth with sandpaper that is too fine.  You can argue over which type of noise it is, I'll read it.

I can see why the D800 gets such high marks from DXO, but still, I don't think it tells the whole story.  And ergonomically, the D800 is unusable for me.  Also, I don't care for Nikon's lens lineup...especially their 135 f/2 lens.  That alone would make me not ever switch to Nikon.  Talk about a lens for old farts, looks like it was designed in the 1950's.  There are plenty of other reasons too, of course.

[jrista]

Interesting points from all of you.  I do need to try Topaz...however mine is currently an older XXD body, not a 5D2.

Mikael, that's a good illustration of your point.  However, in that situation, multi-shot HDR is really more called for, in my opinion.  My body has a bit worse noise than the 5D2...hahaha.  How would the D3s compare using similar under-exposed technique at low ISO, to the D800 (accounting for image dimensions, however you might want to do that)?

My cousin has the 5D3, and he mostly uses Lightroom 4 (although he has CS6).  I tried to reduce some of the "grain" noise on a couple of his RAW shots done at around ISO 4000...on his computer using LR4.  I must say, LR4 is far less effective at reducing that, than CS5 ACR is, on similar grain noise on my own files, done with my older body.  Of course, those have similar amounts of noise way down at ISO 1250...haha.  His files were exposed normally, not under-exposed.  This grain noise seems to have a hard texture...sort of like you're trying to sand it smooth with sandpaper that is too fine.  You can argue over which type of noise it is, I'll read it.

I can see why the D800 gets such high marks from DXO, but still, I don't think it tells the whole story.  And ergonomically, the D800 is unusable for me.  Also, I don't care for Nikon's lens lineup...especially their 135 f/2 lens.  That alone would make me not ever switch to Nikon.  Talk about a lens for old farts, looks like it was designed in the 1950's.  There are plenty of other reasons too, of course.

HDR is really the right way to go for high DR scenes, even with the D800. One thing no one ever discusses is the loss of color fidelity in the shadow tones when doing extreme shadow lifting like that. It doesn't really matter if there is banding noise there or not...you are still losing the same amount of color fidelity with either a Canon or a Nikon camera. The only way to really preserve the color fidelity of the shadows in a high DR scenario is to take bracketed shots, and combine as HDR in post. You should end up with much greater (i.e. orders of magnitude greater) color fidelity in the shadows, as well as finer highlight tonality, when using HDR vs. under-exposing and lifting shadows by 3-5 stops.

[jrista]

yes, You can believe that if its suits you.
this is my 5dmk2 and my d800, exposed in the same way  to get the high lights in the sky intact
then the darker parts are lifted and adjusted the same. The results from the Canon is visible more pattern noise , banding, lower signal , less color information.

You should run the 5D II RAW image through Topaz DeNoise 5, and use the debanding feature. You might be surprised how much DR is recovered. The blacks may not be as rich, as Canon uses a bias offset which eats a chunk of shadow DR, but the noise would very likely disappear. DeNoise has shadow tone recovery features as well, which could restore or deepen some of those shadow tones. Once you eliminate the banding noise, Canon sensors fare a hell of a lot better against there competitors.

why would I do that? no software who eliminate banding etc are doing that with out that Im  loosing resolution/details
it is much better if Canon improves theirs read out noise

Well, as always, you are entirely missing the point.  However, there is no point in arguing with you, as your position is staunch... I really don't know why you don't just go out and get a D800 and be done with it. Why wait when you clearly prefer an alternative brand?

[jrista]

I have much  more money invested in  Canon L lenses than Nikon , and I have a d800 since april last year.
Theres is no problem for me  to have more than one system , and I do not understand  you answers as long we are discussing DR, APS, sensors etc

one statement before was there is no APS with 14 stops DR- yes there are, Nikon and Pentax
you told us there will be a problem to use Sony sensors, patents etc  - Nikon have  used Sony sensors with there own color filter and micro lenses since D3x
there are lot of  facts  that are false and  I think  that should be corrected regardless of whether you use only Canon or not

Nikon has only used Sony Exmor technology when the sensor was fabricated BY SONY. You missed the point I was trying to make before regarding the D5200 sensor. But, I'm not going to explain it again.

[LetTheRightLensIn]

I also encourage anyone who disputes that DXO is truly neutral to download DXO Optics Pro and test it out on Canon raw files. Of all the available CR2 RAW editors, DXO Optics Pro seems to fare the worst, by far...especially in the noise and NR department. (Lightroom performs significantly better with CR2 files and NR, as does DPP, obviously.) The poor performance of DXO Optics Pro with CR2 files makes me wonder whether DXO really knows how to process them correctly, and whether that may somehow factor into the growing rift between anything Nikon and anything Canon in their tests.

What does RAW processed images from DxO have to do with their sensor tests which test the RAW file, untouched, directly?

[Aglet]

As for the Canon 70-200 f4/L having "terrible bokeh"...I've had mine for 3 years, and I love the bokeh.  It is very smooth, at least on my crop camera.  I don't notice much bokeh at the wider end, but near to 200mm, it seems very smooth to me.  (Closeups of wildflowers, etc...really "pop"). ..

I don't know about the f/4 Ls
I DO know the f/2.8 L II can have horrid bokeh, I posted an example in the lens gallery.  It's far from the only one I've got.

[Aglet]

yes, You can believe that if its suits you.
this is my 5dmk2 and my d800, exposed in the same way  to get the high lights in the sky intact
then the darker parts are lifted and adjusted the same. The results from the Canon is visible more pattern noise , banding, lower signal , less color information.

You should run the 5D II RAW image through Topaz DeNoise 5, and use the debanding feature. You might be surprised how much DR is recovered. The blacks may not be as rich, as Canon uses a bias offset which eats a chunk of shadow DR, but the noise would very likely disappear. DeNoise has shadow tone recovery features as well, which could restore or deepen some of those shadow tones. Once you eliminate the banding noise, Canon sensors fare a hell of a lot better against there competitors.

Skuze me for bein' cheeky but..

You otta try a camera from Nikon, Pentax, Fuji or Olympus then you won't have to try smearing the banding away with software plugins.

[jrista]

yes, You can believe that if its suits you.
this is my 5dmk2 and my d800, exposed in the same way  to get the high lights in the sky intact
then the darker parts are lifted and adjusted the same. The results from the Canon is visible more pattern noise , banding, lower signal , less color information.

You should run the 5D II RAW image through Topaz DeNoise 5, and use the debanding feature. You might be surprised how much DR is recovered. The blacks may not be as rich, as Canon uses a bias offset which eats a chunk of shadow DR, but the noise would very likely disappear. DeNoise has shadow tone recovery features as well, which could restore or deepen some of those shadow tones. Once you eliminate the banding noise, Canon sensors fare a hell of a lot better against there competitors.

Skuze me for bein' cheeky but..

You otta try a camera from Nikon, Pentax, Fuji or Olympus then you won't have to try smearing the banding away with software plugins.

I've tried Nikon, Pentax, and Olympus. Can't stand the Pentax and Olympus cameras. They have good low ISO IQ, but that is where it all stops. Nikon cameras also have good low ISO IQ, and they have a hell of a lot better selection of lenses...but I can't stand their ergonomics. I'd also take Canon glass over any other glass, including Nikon's, any day! There is a hell of a lot more to IQ than simply sensor. Canon definitely lacks in that department, but they excel at everything else, and its the whole package that really leads to consistent IQ.

And if we are trying out each others ideas, I still say actually try DeNoise 5. When applied to the RAW, the debanding doesn't smear, and the results are pretty amazing. They might as well call it DR Recovery rather than Debanding, as that is basically what it does.

[jrista]

yes, You can believe that if its suits you.
this is my 5dmk2 and my d800, exposed in the same way  to get the high lights in the sky intact
then the darker parts are lifted and adjusted the same. The results from the Canon is visible more pattern noise , banding, lower signal , less color information.

I find that very hard to believe, can you post the untouched RAW files please?

It is a contrived scenario. The D800 image has clearly been exposed better. The 5D II image has much deeper blacks, indicating it was underexposed relative to the D800, thus putting at a relative disadvantage (above and beyond any actual disadvantage it may actually have). The whole point of those shots is to purposely make the majority of the image "black", requiring multiple stops of shadow recovery. The bulk of the pixel area of Mikael's 5D bashing images are lifted 4, 5, 6 stops, which is a ludicrous comparison in any but the most extreme of the extreme circumstances. I think that is a bit of a DPReview Nikonian forums "fad"...they do that a lot over there, just for sh*ts and giggles.

[ishdakuteb]

yes, You can believe that if its suits you.
this is my 5dmk2 and my d800, exposed in the same way  to get the high lights in the sky intact
then the darker parts are lifted and adjusted the same. The results from the Canon is visible more pattern noise , banding, lower signal , less color information.

well... number of people and i, in this forum, have been asking for your raw files but we have not seen anything yet.  however, i do not know how good you are in photography? but if i have a chance to meet you, i must see how good you are in photography as well as operating dslr(s).  this message is written by a person who uses dslr for just almost a year, feb 10, 2013 will be a year to be exact.

note:  as canon already states, it offers only two stops recovery.  and if you are a photographer who shoot under or over expose for more two stops, you should change your career.  one thing that i should let you know that you should not think that i know nothing about nikon.  in fact, i play around with nikon cameras every single time being at costco...

[CarlTN]

Jrista, I don't think I've given you the credit you deserve for your posts here.  Very good.  I especially like them on page 1 of this thread...I hadn't read it close enough.

You all seem to be doing well, but there's too much arguing.  What is this, a pissing contest? 

Regarding Pentax...um, I do know for a fact, that their sensor is the same as Nikon's crop sensor from the D7000.  Is that the same as the D5200?  I don't know or really care.  Whether it's made by Sony, Toshiba...Minolta...Contax...Voigtlander...Ongaku...Koetsu...Feastrex...Mitsubishi...Tata Motors...Adrian Newey...Danicka Patrick kissing a nerd for charity...I don't know. 

[neuroanatomist]

As always, if you want to 'prove' your point, it helps to design your 'test' with a bias toward your desired outcome (such as ETTL vs. ETTR).

[David Hull]

yes, You can believe that if its suits you.
this is my 5dmk2 and my d800, exposed in the same way  to get the high lights in the sky intact
then the darker parts are lifted and adjusted the same. The results from the Canon is visible more pattern noise , banding, lower signal , less color information.

You should run the 5D II RAW image through Topaz DeNoise 5, and use the debanding feature. You might be surprised how much DR is recovered. The blacks may not be as rich, as Canon uses a bias offset which eats a chunk of shadow DR, but the noise would very likely disappear. DeNoise has shadow tone recovery features as well, which could restore or deepen some of those shadow tones. Once you eliminate the banding noise, Canon sensors fare a hell of a lot better against there competitors.

why would I do that? no software who eliminate banding etc are doing that with out that Im  loosing resolution/details
it is much better if Canon improves theirs read out noise

True, but until they do there is an option for those using the later cameras that tend to produce a more unidirectional form of banding (and less of it at that).  And… since this tends to be more evident in the shadow areas, any smearing is probably less obvious.  The point is, there are several viable work-arounds for any low ISO deficiencies that the Canon sensors might have.  At this point, I consider this to be a feature that Nikon has and Canon does not.  Like any other "feature" one has to determine how important it is to what they want/need to do.  For those who prefer Canon (for whatever reason that might be), there are options for the fairly limited number of real world cases where the read noise poses a significant problem.  To me, this is a rational way to look at it which bypasses most of the hype.

Yes, I am the same guy from DPR.  I have to sau that I like their UI a lot better it is harder to keep track of things here.