5DS-R DR test on DPReview

[neuroanatomist]

Your point wast that you assert neither camera has enough dynamic range to capture that scene in a single shot. I disagree. I believe the A7r has enough dynamic range to capture it either way, but particularly if the blinds are open (the more pleasing shot, as you stated).

You believe? Demonstrate, or demonstrate not. There is no 'I believe'.

 

[jrista]

Your point wast that you assert neither camera has enough dynamic range to capture that scene in a single shot. I disagree. I believe the A7r has enough dynamic range to capture it either way, but particularly if the blinds are open (the more pleasing shot, as you stated).

You believe? Demonstrate, or demonstrate not. There is no 'I believe'.

Heh. Well, we both believe something. You believe one thing, I believe another. I've done some demonstration already...I'm not surprised it's not enough, however. Where are the demonstrations to back up your beliefs? Hmm... ???

I've got some six stop pushes (actually probably more than six stops, maybe seven or so). Once my current integration routine in PixInsight finishes (it's registering and integrating nearly 200 files, so it will probably be a couple of hours), I'll do some more advanced NR on both the 5D III and A7r files.

 

[neuroanatomist]

I'm not sure how advanced NR is going to bring back the blown out windows, but have fun!

 

[jrista]

I'm not sure how advanced NR is going to bring back the blown out windows, but have fun!

The outdoor highlights were purposely clipped in this image, as I stated in my previous post, to ensure full utilization of both cameras dynamic range. It's easy enough to avoid doing that, however it is also beside the point. Highlights aren't going to have any noise. It's the shadows that have noise issues, and the reason we resort to HDR blending in the first place.

I wanted to see what I could do with "normal people tools", since few people even know about PixInsight, and nor many regular photographers would want to spend the time learning it. Here is quick pass of both sets of data in Photoshop CC, from the area of deepest shadows, from 6-stop pushes. I used the best debanding script I know of to clean up the banding on the 5D III crop. I also did what I could with Topaz DeNoise and PS's Reduce Noise filters on both images. I tried to use a PixInsight masking technique, however I am not sure how much PS supports it...the results on the A7r data are pretty good, I couldn't tell if it helped at all with the 5D III data. The 5D III noise seem to have improced in characteristic a bit...it's nicer, cleaner...but there just isn't any real detail there to restore. I also tweaked the contrast a bit, similar to my previous post, to restore some punch to the images.

Original untouched exposures, exposures after +6 stop push, exposures after NR:

PI is still cranking away, so this is without any of it's advanced NR features. PI may be able to do some more with the 5D III's blotching, and will probably be able to preserve some more detail. I'm actually curious to see how much detail I can preserve, or even enhance, while still reducing noise. I am pretty sure I can clean up the blotchy color noise better with PI. I am not sure if I can do much better with PI for the A7r...

Anyway...not looking for a fight. Hope people find these samples interesting and/or useful.

 

[neuroanatomist]

I'm not sure how advanced NR is going to bring back the blown out windows, but have fun!

The outdoor highlights were purposely clipped in this image, as I stated in my previous post, to ensure full utilization of both cameras dynamic range.

Yes, I noticed that. Interesting that the parameters of your demonstration specifically preclude testing your belief that the a7R could capture the full scene DR, but I know that wasn't the point you set out to make.

 

[jrista]

I'm not sure how advanced NR is going to bring back the blown out windows, but have fun!

The outdoor highlights were purposely clipped in this image, as I stated in my previous post, to ensure full utilization of both cameras dynamic range.

Yes, I noticed that. Interesting that the parameters of your demonstration specifically preclude testing your belief that the a7R could capture the full scene DR, but I know that wasn't the point you set out to make.

The highlights aren't clipped all that much. It's maybe a quarter of a stop at most, it shouldn't even be that much. When I was taking the shots, I only upped exposure by a third of a stop to get the highlights to clip, and they had some headroom before I did that. A quarter of a stop isn't going to make a lot of difference to the shadow performance, for either camera. I plan to get the A7r II or maybe A9 (or whatever that higher end model ends up being called...it's supposed to be aimed more at pros, should have better weather sealing, be more rugged, etc.) in the future. I'll be happy to get some more 'realistic' exposures then.

Before that even, I plan to get the A6000, and I could do more testing with that.

BTW, the parameters of my testing were actually set by members of this forum, so many months ago when I actually rented the camera and did the test. There were several members that insisted that a DR test was invalid if the highlights weren't just slightly clipped, as there was a rather extensive argument at that time over the proper way to expose a step wedge for DR testing at the time, and that bled over into my simple testing. My personal parameters would have been to keep the highlights under the clipping point, as that's how I actually shoot.

 

[3kramd5]

These threads are both tiresome and paradoxically amusing. Over and over again, people agreeing about facts and disagreeing about implications.

Can we just once and for all agree that canon's current sensor/signal chain is noisier than its main competitors, but that there are several other parameters about which canon bests its main competitors and move on?

Here are some facts:

1) canon is noisier and this doesn't capture as much usable DR;
2) for many people, the differences between canon architecture and sony/Samsung/toshiba/whatever DR is not a significant factor, but,
3) none of those people would complain (and indeed most would appreciate) if the next generation of canon architecture had less noise.

I have a few 5D3s, a 5D2, a D7000, and an A7R. I have a 5DS on order that I plan to replace the Sony with. The Sony I find preposterous. It feels like it goes out of its way to make using it difficult. I'd much rather have high res in a canon body than the Sony, regardless of the DR deficit. However if the supposed 5DC can do say 18MP stills at 16-stops, I'd gladly cancel my order, return, or re-sell, depending on timeline.

A purchase is a balancing act.

 

[Valvebounce]

Hi Jrista, Neuro.
I find it incredible that two people, whom I would consider to have a far greater comprehension of the technical sides of each other's arguments than I, can continue to intentionally misunderstand each other's statements during these debates.
I'm fairly sure that I understand the point each of you is trying to make, right up to the point when you start this nonsense and confuse me. What is so sad is that I kind of get the impression that you both would want the same thing in the end.

Cheers, Graham.

 

[K]

The comparison, starting with a 3 stop push, then 4 stop push, then 5 stop push:

It's an effective demonstration of the Exmor shadow lifting capability at low ISO, and makes the point.

Of course, in a common real-world application of the full shot posted above, the idea would be to open the window blinds completely and see the view of outside world and a pleasingly-lit interior. The fact that neither camera has sufficient DR to accomplish that makes a very different point...and one that many people find more compelling than the point you seem to have intended to make.

Fixed your above statement to include "at low ISO" ...because the shadow lifting isn't anything special at higher ISO.

Overall, I agree with your commentary, and has been a point I've also been making all along.


It's truly awesome the shadow and exposure lifting capabilities of the Exmor. No doubt about it. However, as JRista's photos show - the result is an ugly photo. While this sensor mops the floor with the Canon in avoiding gruesome noise, banding, blotches and other image degradation in the shadow areas, the resulting image is still sub-par and only less-awful than Canon. Less crappy, is still crappy.

Thus, my point again - while it's wonderful to be able to lift these shadows that much, there just isn't that much practical application. Not to the extent that it has become the war cry of the Nikonians across the internet. This single aspect gets entirely too much attention, and only because it is the only area Nikon has an edge in.


"Exmored" images all have that same ugly look to them. They lack that nice rich contrast. They lack detail they should have had at that ISO level. The colors downright suck. And overall, the image looks grungy and is far from presentable.Worst of all, they have this bizarre, unrealistic look to them due to ares that should be darker and contrasty being unnaturally brighter. At least in my opinion. It's like there's this undetectable glow in the room lighting things up in a low contrast way.


In order to achieve a great looking image in high DR scenes, one is forced to use good lighting technique or good HDR technique - Exmor or not. Practical and reasonable shadow and exposure lifts are minor corrections. Canon, while having an inferior sensor in this regard, is luckily for them, still within the tolerance of 1-2 stops push without mangling the image much. So for minor corrections, Canon is right up there in the game. They are no where near the Exmor on crazy stuff like 5 stops. But again, what use is that?

In another one of these crazy Exmor DR threads, someone posted a link to a blogger who is a total Exmor fanboy doing wedding photography. This photog showed off several Exmored images of actual weddings on their website. These images, perhaps to someone used to cell phone pics, are decent and perhaps even impressive. However, I think they are awful and a great opportunity to capture truly pro images was lost because this person thinks they can Exmor everything and get away with it.


Whatever, who am I to say? Beauty is in the eye of the beholder. If people like that look, go for it. If there are people willing to pay for images that look that way, so be it. Their money, not mine.

The pros that I know in person, and the pros that I've followed online, not a single one of them - Nikon or Canon, does any kind of radical shadow lifting. They use lighting techniques as it should be. If it is natural light, they use their professional skills, artistic eye and creativity to setup the scene to optimize in the natural light.

Exmorites just want to snap shot away in any light with no care for the fundamentals, and then just crank on sliders in LR and think they are putting out Pro quality.

If Canon released a sensor with more DR than Sony, I will never come to this forum or elsewhere and brag about or claim the usefulness of being able to do a 6-7 stop lift on an image. Canon, Nikon, Sony ...matters not - that's not a technique. The result will be inferior to a properly executed photograph.


8)

 

[jrista]

Hi Jrista, Neuro.
I find it incredible that two people, whom I would consider to have a far greater comprehension of the technical sides of each other's arguments than I, can continue to intentionally misunderstand each other's statements during these debates.
I'm fairly sure that I understand the point each of you is trying to make, right up to the point when you start this nonsense and confuse me. What is so sad is that I kind of get the impression that you both would want the same thing in the end.

Cheers, Graham.

I have never intentionally misunderstood anything.

 

[Eldar]

Hi Jrista, Neuro.
I find it incredible that two people, whom I would consider to have a far greater comprehension of the technical sides of each other's arguments than I, can continue to intentionally misunderstand each other's statements during these debates.
I'm fairly sure that I understand the point each of you is trying to make, right up to the point when you start this nonsense and confuse me. What is so sad is that I kind of get the impression that you both would want the same thing in the end.

Cheers, Graham.

I have never intentionally misunderstood anything.

I believe you. If everyone substantiated their view like you, many threads would have been less boring.

 

[Lee Jay]

You're very lost.

If you cram 18 stops into 6 stops, you do it by reducing contrast. Stretching the resulting 6-stop file doesn't change it back to an 18 stop file, it's still just 6 stops with the tones pushed toward the edges.

There are two extremes.

You can keep all 18 stops by cramming them into a 6 stop file. This costs you contrast. The darkest tones in the file are no more than 6 stops darker than the brightest tones. If all 18 stops are in their, you've lost 12 stops of contrast no matter how you distribute those tones.

The other extreme is to crop the histogram. Here, you choose 6 stops of the original scene and keep only those, discarding the rest. This preserves original scene contrast, but costs you all those darks and lights you drove to saturation.

There are no other choices except all those between those two. You cannot preserve both the DR and the contrast, by definition.

Well, I beg to differ. I'll demonstrate.

BTW, I never said you could "restore dynamic range." I also never said "preserve contrast", which I take to mean preserve the original contrast of the image. I said you could "restore contrast."

You're wrong either way. Once you choose to cram 18 stops into a 6-stop output, you have lost contrast. Nothing can ever be done about that. You can restore a bit of local and apparent contrast with sliders, but the original scene contrast is lost, forever. That's actually what you *want* to happen when you use shadow lifting and highlight pulling - you want to reduce contrast to fit it into the box of the output device you are using, be it a print or a screen.

 

[jrista]

You're very lost.

If you cram 18 stops into 6 stops, you do it by reducing contrast. Stretching the resulting 6-stop file doesn't change it back to an 18 stop file, it's still just 6 stops with the tones pushed toward the edges.

There are two extremes.

You can keep all 18 stops by cramming them into a 6 stop file. This costs you contrast. The darkest tones in the file are no more than 6 stops darker than the brightest tones. If all 18 stops are in their, you've lost 12 stops of contrast no matter how you distribute those tones.

The other extreme is to crop the histogram. Here, you choose 6 stops of the original scene and keep only those, discarding the rest. This preserves original scene contrast, but costs you all those darks and lights you drove to saturation.

There are no other choices except all those between those two. You cannot preserve both the DR and the contrast, by definition.

Well, I beg to differ. I'll demonstrate.

BTW, I never said you could "restore dynamic range." I also never said "preserve contrast", which I take to mean preserve the original contrast of the image. I said you could "restore contrast."

You're wrong either way. Once you choose to cram 18 stops into a 6-stop output, you have lost contrast. Nothing can ever be done about that. You can restore a bit of local and apparent contrast with sliders, but the original scene contrast is lost, forever. That's actually what you *want* to happen when you use shadow lifting and highlight pulling - you want to reduce contrast to fit it into the box of the output device you are using, be it a print or a screen.

Now your just debating semantics. I've demonstrated what I meant, there is no longer any ambiguity. I was never talking about preserving "original scene contrast", I made that very clear. I also made the meaning of contrast that I was referring to quite clear, both through words and images. Contrast is contrast...it doesn't have to be "original scene contrast"...local and apparent contrast are still contrast. Your imposing a more limited definition of the word contrast upon what I was saying. Either you are just looking for a fight, or you did not read or understand everything I wrote.

Lee Jay, you, like many others here, seem to have an innate need to prove people wrong. I feel sorry that's how you have to spend your time on these forums, proving people wrong. I disagree with you, however I never set out to prove you wrong, I only set out to demonstrate my point from the angle I see it from...and I made it clear that the vast bulk of my post was simply educational and for the benefit of everyone reading, not a rebuttal, certainly not an attempt to prove you wrong.

This whole proving people wrong thing...it's really a travesty on these forums. Just as much a travesty as people like K, who seem to think they are the ultimate authority on what is or isn't photography, what makes good photography and what makes bad photography, or that everyone who uses an Exmor is just a slider junkie.

We are all adults here. Is there no way we can all find a way to have reasonable discourse about these subjects? Why does everything have to boil down to insults and discredit and proving everyone around you wrong? That is truly a sad way to go about life, insulting and deriding and discrediting everyone so you can prove them wrong...

Lee Jay, you and I are not talking about the same thing. I made my point very clear, I made the meaning of the words I'm using in the context I am using them very clear. I even demonstrated what I meant by contrast with visual aids. I'm no longer interested in debating semantics.

Well, I have clearly disrupted this forum long enough. Yes, I'm a crazy insane exmorite. I like having more dynamic range. I love the technical engineering feat that many new modern sensors are, not just Exmor. I'm fascinated by image signals and the mathematics involved in describing them, as well as processing them. I love that stuff. I like to share my knowledge about such things. But damn...these forums are one of the more hateful places on the net, I tell ya. I really hate coming here. It's just a wonderfully distasteful experience every single time.

 

[Valvebounce]

Hi Jrista.
My apologies, intentionally misunderstand was a bit strong. I would also say that I find many of your posts most informative, including those here, it is just that the information starts getting lost in the noise!

Cheers, Graham.

Hi Jrista, Neuro.
I find it incredible that two people, whom I would consider to have a far greater comprehension of the technical sides of each other's arguments than I, can continue to intentionally misunderstand each other's statements during these debates.
I'm fairly sure that I understand the point each of you is trying to make, right up to the point when you start this nonsense and confuse me. What is so sad is that I kind of get the impression that you both would want the same thing in the end.

Cheers, Graham.

I have never intentionally misunderstood anything.

 

[Sporgon]

Hi Jrista.
My apologies, intentionally misunderstand was a bit strong.

Perhaps you meant obtuse

 

[jrista]

Hi Jrista.
My apologies, intentionally misunderstand was a bit strong. I would also say that I find many of your posts most informative, including those here, it is just that the information starts getting lost in the noise!

Cheers, Graham.

Hi Jrista, Neuro.
I find it incredible that two people, whom I would consider to have a far greater comprehension of the technical sides of each other's arguments than I, can continue to intentionally misunderstand each other's statements during these debates.
I'm fairly sure that I understand the point each of you is trying to make, right up to the point when you start this nonsense and confuse me. What is so sad is that I kind of get the impression that you both would want the same thing in the end.

Cheers, Graham.

I have never intentionally misunderstood anything.

I agree, there is a lot of noise. I've restore a number of people to my ignore list. Hopefully that will cut down on it. I can't help what you guys may see, but it should be easier for me to ignore all the noise and not bother responding to it.