Canon 5Ds/R New review with file samples

[Telecaster]

Multiple exposures don't work when there are moving elements in the scene (e.g. wind causing leaves to flutter). ND grads don't work when the horizon/transition zone isn't straight (e.g. trees or buildings in the way, or in the mountains, or when shooting a strongly-backlit subject with no option to use fill flash), are susceptible to flare and don't work well in the corners when using several stops and a UWA. Neither works well when stitching a rotational panorama - multiple exposures often won't stitch exactly the same way, while it's no good having a GND which moves with the lens as it rotates. Sometimes you just really need single-exposure DR.

I get all that: "exceptions that prove the rule." What I'm objecting to is taking such cases and treating them as though they're the norm within photography overall. They're not. The leap in dynamic range from transparency film to late 2000s sensors was significant, game-changing even. Now we're seeing more incremental gains…nice but IMO not deserving of the degree of angst we've been devoting to the subject.

-Dave-

[shadowblade]

I don't remember to have seen your pictures so maybe your style is totally different from mine and therefore your blending percentage is very different. My galleries are in my signature so easy to check.

Another reason which I often see is that people do not bracket and do exposure compensation (one way or the other) based on the histogram on the camera. They will very often underexpose by 1-2 stops (perhaps more sometimes) and therefore need to blend much more often. I always get an optimal exposure so therefore much less need for blending. even using Canon.

I've had a look through your gallery - the difference is pretty obvious. Most of your landscapes and not many are strongly backlit. Also, most of them were taken at relatively high latitudes - 40 degrees from the equator or more - where sunlight doesn't tend to be as strong.

I probably blended around 90% of 5D2 shots, and only 10% of A7r shots.

[shadowblade]

I get all that: "exceptions that prove the rule." What I'm objecting to is taking such cases and treating them as though they're the norm within photography overall. They're not. The leap in dynamic range from transparency film to late 2000s sensors was significant, game-changing even. Now we're seeing more incremental gains…nice but IMO not deserving of the degree of angst we've been devoting to the subject.

-Dave-

Shooting at ISO 3200-6400-12800 or higher also isn't the norm. Nor is shooting at 14fps. Yet every extra stop of high-ISO performance and every extra frame per second is greeted as if it were the biggest breakthrough in photographic technology ever, despite the fact that ISO 51200 is functionally identical to ISO 12800 pushed two stops in the same camera.

2-3 stops of DR is a huge difference. It's the equivalent of a 2-3 stop ND filter. Many otherwise-impossible shots on a 5D2 or 5D3 were saved by the ability to use such a filter, due to the transition zone being linear enough to allow it. But many others were missed, or taken suboptimally, because the horizon didn't allow it. With the extra DR of the A7r or D810, these shots are not a problem.

[Hans Kruse]

I've had a look through your gallery - the difference is pretty obvious. Most of your landscapes and not many are strongly backlit. Also, most of them were taken at relatively high latitudes - 40 degrees from the equator or more - where sunlight doesn't tend to be as strong.

I probably blended around 90% of 5D2 shots, and only 10% of A7r shots.

I'm curious to see some of your galleries.

[Jack Hogan]

2-3 stops of DR is a huge difference. It's the equivalent of a 2-3 stop ND filter. Many otherwise-impossible shots on a 5D2 or 5D3 were saved by the ability to use such a filter, due to the transition zone being linear enough to allow it. But many others were missed, or taken suboptimally, because the horizon didn't allow it. With the extra DR of the A7r or D810, these shots are not a problem.

I agree, the DR performance made available by Exmor sensors in full frame cameras - first seen in the D800 a mere three years ago - suddenly freed FF landscapers from a burdensome constraint.  I used to bracket often, now I almost never do.  I used to spend much time trying to blend bracketed files together so that they would not look artificial, almost never succeeding to my satisfaction.  With a single, linear raw file it's just a matter of playing with a couple of sliders.

YMMV because not everyone has the same shooting style or looks for the same types of scene.  So to each their own, as long as they understand what they are buying and why.

Jack

[Hans Kruse]

I agree, the DR performance made available by Exmor sensors in full frame cameras - first seen in the D800 a mere three years ago - suddenly freed FF landscapers from a burdensome constraint.  I used to bracket often, now I almost never do.  I used to spend much time trying to blend bracketed files together so that they would not look artificial, almost never succeeding to my satisfaction.  With a single, linear raw file it's just a matter of playing with a couple of sliders.

YMMV because not everyone has the same shooting style or looks for the same types of scene.  So to each their own, as long as they understand what they are buying and why.

Jack

I'm a full time professional photographer organizing photo workshops and shoot more than 100 days per year and landscapes almost exclusively. Although I do agree about the benefits of the Exmor sensors and I have this in my Nikon D810 I still shoot Canon and only a few percent of the shots need blending. In my view many landscape shooters do not realize that they will underexpose and not get the optimal exposure unless they bracket and choose the optimal exposure from looking at the exposure in post processing (Lightroom, Capture One and Raw Digger). If we had RAW histograms one could determine this while shooting although at the expense of time and attention to this while shooting.

I recommend on my workshops to bracket for two reasons 1) to select the optimal exposure without essential highlights clipped in post processing 2) to not constantly chimping to determine the exposure based on a JPG histogram. By shooting this way one can optimize the shooting for best composition and use the sometimes narrow window of best opportunity optimally in terms of making the most of it. Bracketing in the digital world is free. One can turn bracketing on and off based on the lighting conditions, but there is a high chance that when the amazing light comes that the photographer will forget to do this. The more "mechanical" the shooting approach is the less errors are made. This is my experience. Others may disagree

Ideally there should not only be RAW histograms but also a setting in the camera to get the optimal exposure and with margins for clipping. Sometimes there can be 1-2 stops difference in exposure and in terms of highlights there is essential no difference in the final edited picture, but there is a visible difference in the shadows in terms of noise and details. This is also true for Exmor sensors, but they are more tolerant. Even with such settings I might still bracket slightly to narrow the number of cases where a blend would be ideal (although perhaps not entirely needed).

I do not use ND grad filters because in many cases they do not fit the landscape I'm shooting and they are time consuming to use. As mentioned my stats is around 98% on Canon do not need blending and on Nikon around 99,5%. I shoot both systems to known them in detail so that I can help my workshop students with any detailed setting needed. The Exmor sensor has for me also made less blending needed. The problem is that Canon has better lenses and a better designed body. E.g. an always available EFCS which is implemented on the D810 in a clumsy way. But the D810 is a significant improvement over the D800E.

[Hans Kruse]

2-3 stops of DR is a huge difference. It's the equivalent of a 2-3 stop ND filter. Many otherwise-impossible shots on a 5D2 or 5D3 were saved by the ability to use such a filter, due to the transition zone being linear enough to allow it. But many others were missed, or taken suboptimally, because the horizon didn't allow it. With the extra DR of the A7r or D810, these shots are not a problem.

I agree that 2-3 stops is a big difference and many photographers loose a big part of that by not exposing optimally. It's not true that the D810 sensor does away with the need for blending. But it certainly does make the cases fewer.

[Bart_van_der_Wolf]

In my view many landscape shooters do not realize that they will underexpose and not get the optimal exposure unless they bracket and choose the optimal exposure from looking at the exposure in post processing (Lightroom, Capture One and Raw Digger). If we had RAW histograms one could determine this while shooting although at the expense of time and attention to this while shooting.

+1

I couldn't agree more.

In addition, the resulting shadow noise can be reduced further with noise reduction. It's only when trying to significantly boost the brightness of shadows that the remaining pattern noise can rear its ugly head, and that is supposed to be improved in the 5DS / 5DS R.

A noise reduction plugin like Topaz Denoise has a horizontal / vertical banding reduction control that can be useful for cameras which have too much pattern noise. Subtraction of Dark frames would also allow to address issues like that and clean up shadow patterns, but there is only little support from Raw converter producers, RawTherapee being a positive exception.

Ideally there should not only be RAW histograms but also a setting in the camera to get the optimal exposure and with margins for clipping.

Yes, it's amazing that camera makers have not picked up this requirement, which has been requested for a long time already.

Cheers,
Bart

[shadowblade]

I'm a full time professional photographer organizing photo workshops and shoot more than 100 days per year and landscapes almost exclusively. Although I do agree about the benefits of the Exmor sensors and I have this in my Nikon D810 I still shoot Canon and only a few percent of the shots need blending. In my view many landscape shooters do not realize that they will underexpose and not get the optimal exposure unless they bracket and choose the optimal exposure from looking at the exposure in post processing (Lightroom, Capture One and Raw Digger). If we had RAW histograms one could determine this while shooting although at the expense of time and attention to this while shooting.

I thought everyone knew this. And I'm still puzzled as to why no-one's bothered implementing it, when it's really just a firmware update.

Do you know of any way to set a custom curve as the 'default' in-camera jpeg shooting mode? Because, if you could set a linear curve going from the noise floor to full well saturation, that would give you a RAW histogram anyway.

I do not use ND grad filters because in many cases they do not fit the landscape I'm shooting and they are time consuming to use. As mentioned my stats is around 98% on Canon do not need blending and on Nikon around 99,5%. I shoot both systems to known them in detail so that I can help my workshop students with any detailed setting needed. The Exmor sensor has for me also made less blending needed. The problem is that Canon has better lenses and a better designed body. E.g. an always available EFCS which is implemented on the D810 in a clumsy way. But the D810 is a significant improvement over the D800E.

Being used to shooting with MF/LF film bodies, any digital setup is an improvement.

Looking through your photos, I suspect the reason you need to blend much less often than me comes down to where and when we shoot. The photos on your website are almost exclusively above 35 degrees from the equator, mostly greater than 40 degrees. Here, even direct sunlight is much less strong and produces less contrast than at locations closer to the equator. Also, you seem to shoot a fair bit in ovecast conditions. Looking through my collection, the majority of my shots were taken between 30 degrees south and 30 degrees north of the equator; virtually all the photos that didn't require blending or filters (not counting night-time shots) were taken at higher latitudes, such as Mongolia, Patagonia or parts of New Zealand.

[Hans Kruse]

I thought everyone knew this. And I'm still puzzled as to why no-one's bothered implementing it, when it's really just a firmware update.

Do you know of any way to set a custom curve as the 'default' in-camera jpeg shooting mode? Because, if you could set a linear curve going from the noise floor to full well saturation, that would give you a RAW histogram anyway.

Being used to shooting with MF/LF film bodies, any digital setup is an improvement.

Looking through your photos, I suspect the reason you need to blend much less often than me comes down to where and when we shoot. The photos on your website are almost exclusively above 35 degrees from the equator, mostly greater than 40 degrees. Here, even direct sunlight is much less strong and produces less contrast than at locations closer to the equator. Also, you seem to shoot a fair bit in ovecast conditions. Looking through my collection, the majority of my shots were taken between 30 degrees south and 30 degrees north of the equator; virtually all the photos that didn't require blending or filters (not counting night-time shots) were taken at higher latitudes, such as Mongolia, Patagonia or parts of New Zealand.

My photos are mostly from my workshop locations in Italy from Sicily to the Dolomites. There is definitely strong sunlight in these areas and clear air in the mountains. I shot all kinds of angles to the sun from directly towards the sun, side lighting and the sun behind me. I like side lighting a lot as it gives structure to the landscapes. I shoot much less strong sun against me with deep shadows. Sometimes there can be great photos like this but mostly uninteresting.
Not many of my photos are in overcast situations. Many have clouds but with sun light. Besides Italy I have been shooting in Peru, South West USA, India, China, Israel, South Africa, Namibia, Botswana, Spain, Scotland, Denmark, Sweden, France and not any of these location have changed much from what I generally experience in Italy in terms of dynamic range. I do look for what I find to be interesting landscapes and mostly they need to be lit in some way. I seriously doubt that I would find a difference even close to the equator where I was recently on the Seychelles Islands. So I don't buy into your theory

Unless you show me some of your galleries I don't think can get any closer to a reason for the vast difference in needs for blending of shots from a Canon camera.

[Slobodan Blagojevic]

Seriously, people!? Page after page after page, thread after thread after thread, debating the undebatable: that having more DR is better than having less!?

[albedo13]

Sorry to veer the topic away from blending and DR ( ), but thanks Bart for the info on RawDigger. I have never used it, is there a link you could point me to (for Windows 7)? Is this a utility for just being able to read the header information, or can you actually either view the image itself or save it into a format that can then be read via PS or DPP? Thanks

Jim

[Hans Kruse]

Seriously, people!? Page after page after page, thread after thread after thread, debating the undebatable: that having more DR is better than having less!?

Seriously Slobodan, that's not what the discussion is about 

[BJL]

... debating the undebatable: that having more DR is better than having less!?

It is a bit more nuanced than that; these days sensor performance comparisons are all about edge cases, and how relevant a particular edge case is to a particular photographer, when weighed against other factors like preferring a particular lens system.  

As fas as I can tell, a wide array of camera systems now offer enough resolution, dynamic range and low light performance that the great majority of the work of even "serious" (professional and dedicated amateur) photographers will not benefit noticeably from any more.  So the question becomes _which_ photographic situations and styles, and _which_ photographers can still benefit _significantly_ from gear that offers even better performance in one or other of these measures.

For example, I am happy to believe that further DR improvements over what Canon's best cameras currently offers are of little benefit even to some professional landscape photographers such as Hans Kruse (and if so, the same is likely true for a great majority of all "serious" photography), but are highly desirable to some other small but significant group of photographers.

[Slobodan Blagojevic]

... Unless you show me some of your galleries...

Some of us remember Shadowblade images in this forum. Most of them fall into the holly-crap-you-got-to-be-kidding-me category (a compliment). As a public service, I compiled some of them here:

[Slobodan Blagojevic]

...I am happy to believe that further DR improvements over what Canon's best cameras currently offers are of little benefit...

Yes, indeed... that is exactly what we, Canon hostages, say every night, crying ourselves to sleep

[NancyP]

Not what this Canon hostage says when shooting astrolandscapes with the 6D!
Yes, this is another thread about "Why can't manufacturers give us RAW histograms?". This just seems like a no-brainer.

Speaking of "no-brainers", why am I manually bracketing when the camera will do it for me? I started out 5 years ago using 1-stop x 3 bracketing on all non-action shots shot in aperture-priority. At some point I just went to M for everything, made a guesstimate of the right central exposure ( + or - from metered), and hand-bracketed, just as I did 40 years ago.

[Bart_van_der_Wolf]

Sorry to veer the topic away from blending and DR ( ), but thanks Bart for the info on RawDigger. I have never used it, is there a link you could point me to (for Windows 7)? Is this a utility for just being able to read the header information, or can you actually either view the image itself or save it into a format that can then be read via PS or DPP? Thanks

Hi Jim,

It's a tool for the inspection of Raw file data, and a few other options for the generation of files for profiling. This is its home.

Cheers,
Bart

[dwswager]

Seriously, people!? Page after page after page, thread after thread after thread, debating the undebatable: that having more DR is better than having less!?

We can add less noise is better than more and easily corrected noise is better than difficult to correct noise. 

It's about options and forgiveness.  Very few shots require extreme performance, but having it is sometimes a safety net.  I can shoot most sports with 6fps, but it certainly would be easier with faster frame rates to make up for user error or incompetence.

[Hans Kruse]

Some of us remember Shadowblade images in this forum. Most of them fall into the holly-crap-you-got-to-be-kidding-me category (a compliment). As a public service, I compiled some of them here:

Thanks and I do not remember to have seen any of them before.

Btw. they make Peter Lik look naturalistic