Dynamic Range - Try it for yourself, conclude for yourself: 5D III vs. A7r

[jrista]

I received my A7r rental from LensRentals today. I've been working, but I took a few minutes to snap some quick shots of my living room, with the goal of preserving the highlights of the bright sunlit outdoors in the front windows. To do that, I had to underexpose the foreground well below the levels that they appeared it my eyes.

I did not try to make any explicit exposure choices here. The goal was simply to expose for the highlights. These were shot from the hip, no tripod, no real planning. The goal was just to create a couple exposures with highlight preservation in mind, and see how the shadows handled some major pushing. I'll leave it up to each person to draw their own conclusions. I have provided a link to the raw files, so you can all download them, play with them yourself, and verify (or refute, if you prefer) the various claims made about DR.

Here are the original exposures:

Canon 5D III

Sony A7r



Here are the exposures lifted to generally mirror what the room looked like to my bare eyes (+4EV, -100 Highlights):

Canon 5D III

Sony A7r



Just to see how far I could push things, here a couple more versions, not particularly "realistic", but still a good demonstration of what's possible and how the raw data reacts. No interpretation of the data needed here...5D III totally falls apart. A7r...noisy...but manageable.

Maximum push (+5EV, -100 Highlights, -100 Whites, +50 Shadows):

Canon 5D III

Sony A7r



Here are full size versions of the +4 stop pushed examples (zoom in on PhotoBucket for full size):

Canon 5D III

Sony A7r


Here are full size versions of the +5 stop pushed examples (zoom in on PhotoBucket for full size):

Canon 5D III

Sony A7r



Histograms

ETTR? Yup. Too much. On both cameras.

You can download the raw files here:

http://1drv.ms/1ol6Km5




Finally got a chance to process one of the sets of images I got with the A7r and 5D III. This particular set of images I did not really expect to show any difference, however I think the subtleties and small push here might actually be best. I saw this rock with a patchwork of different kinds of mosses, frames in the spotty sunlight shining through some pine trees. I thought the light and shadow made for a decent example of DR.

Here are the original shots:

Canon 5D III

Sony A7r

Here are the shots processed to bring up the shadows:

Canon 5D III (+2 EV, -80 Highlights, +100 Shadows)

Sony A7r (+2 EV, -85 Highlights, +100 Shadows, +35 Contrast)



Here is a quick close-up comparison of the two key shadow areas after the 2-stop push.

Raw files pending.

 

[Lee Jay]

Looks a lot like a shot I took a long, long time ago with my 20D. Guess how I did that...

 

[V8Beast]

Yeah, I can see how that would give any DR junkie a stiffy ;D

Seriously, though, I'm glad to see that you're finally taking a more productive approach to address that nasty case of DR cognitive dissonance that's running rampant around here. It's a good start, and perhaps adding any A7R to your kit will make you a much more happy camper.....or shooter......or keyboard warrior ;D

 

[RLPhoto]

Looks like the a7r will be your landscape cam. Now you can shorten your posts to a link to this thread when the issue rises.

As for Lee jay, I would say a bounce flash.

 

[Lee Jay]

Looks like the a7r will be your landscape cam. Now you can shorten your posts to a link to this thread when the issue rises.

As for Lee jay, I would say a bounce flash.

Yes, as should be quite obvious, a flash is often a great way to control scene contrast. That was a 550EX fired into the ceiling.

 

[jrista]

I dunno yet. There are definitely things I do not like about the A7r. The EVF has massive lag in a lot of situations, shows a LOT of aliasing and moire, and is very pixelated with limited resolution, which just makes the product DOA for me. EVFs have to come a LOOOOOOONG way before I'll be ready to move on from the OVF. A really freakin LONG way. The focus system needs work. It may just be because I'm using an EF lens with an adapter...but it's just...not good. Maybe the A7s will be better. The shutter is indeed very heavy, like people have said. I haven't experienced shutter shock yet, but wow...compared to the 5D III, it's really loud, and it sounds very deliberate. I don't really get why they didn't just use an electronic shutter...Sony has global shutter technology.

Anyway, lots of little things that are just not really there yet. I heard Sony may be releasing new products in January, so I think I'm going to wait for then, see what they release, and give one or two of those a rental. There are things about Sony's implementation I like. Their UI has some nice things about it. I don't care as much for the menu system...it isn't as easy to get around as with the Canon system. But the rest of the UI is pretty cool. There are three main dials, so you can very quickly control aperture, shutter, and ISO with just a simple turn. They also have a dedicated dial for flash compensation, which is kind of cool. There are more features to it, with programmable custom buttons. Need more time to really conclude about button and dial placement, but so far...eh. It could be a lot better.

On the IQ front, the A7r does not disappoint! It's a little noisier in the deep shadows than I expected, not quite like many of the D800 images I've worked. I still need to play around with it, but I think that may be due to compression. Still, despite the noise, the shadows of the A7r are WAY easier to work with. I can just shift the NR slider in LR, and it cleans right up. A little bit of sharpening and deconv, maybe a bit of color noise smoothing, and I can have very usable shadows, even with a +5 stop lift (something I don't expect I'll need very often). I can easily get two more stops out of the A7r than I get with my 5D III, at the very least. Banding presents in the 5D III rather rapidly in high DR situations (like when the sun is setting behind mountains, a situation I am often faced with). I am not sure if the weather will hold up tomorrow...I'm going to be going up into the mountains to photograph the last bit of fall (winter is coming fast this year).

Hopefully the A7r will serve me well this weekend. I'll be taking images with both cameras, so I can compare the results. First impressions, the 5D III takes warmer images, with more vibrancy than the A7r. However, I feel the A7r colors are more realistic (this is when using them on Camera Neutral, and importing with Camera Neutral in LR). I'm a big realist...I don't like to hyper saturate my images (although in some cases, I think it works for some photographers, like Marc Adamus). In that respect, the A7r would also serve me better, producing color I'm more pleased with right out of the camera, meaning I don't have to spend time working color on top of working noise and blending HDR and all that.

So yeah, I'm pretty impressed with the IQ. Not so impressed with the camera body, but there are some things I like about it. Not sure I want to buy one, but I'm happy to know there are options out there that can definitely produce better IQ for landscapes.

 

[Lee Jay]

Looks like the a7r will be your landscape cam. Now you can shorten your posts to a link to this thread when the issue rises.

As for Lee jay, I would say a bounce flash.

Yes, as should be quite obvious, a flash is often a great way to control scene contrast. That was a 550EX fired into the ceiling.

You do realize the whole point was to demonstrate DR, right?

Yes...in a situation where it is not needed.

In fact, in all my years of shooting, I've never once come across a situation where I couldn't get enough DR from a Canon camera at base ISO. Well, except once, and in that case I needed around 30 stops or so, and I don't think even the lens could manage that due to flare.

I need more DR at high ISO all the time. Fortunately, Canon delivers there.

 

[jrista]

When you don't have a flash...then flash cannot be used to solve the DR issue. Which means, more DR IS needed. I do not have a flash for the A7r.

 

[unfocused]

Looks like the a7r will be your landscape cam. Now you can shorten your posts to a link to this thread when the issue rises.

As for Lee jay, I would say a bounce flash.

Yes, as should be quite obvious, a flash is often a great way to control scene contrast. That was a 550EX fired into the ceiling.

You do realize the whole point was to demonstrate DR, right? Oh, and um...I don't have a flash for the A7r...so, doing a comparison with flash just wasn't an option.

That may have been your point, but apparently it wasn't his. Don't you just hate it when people hijack your thread to talk about something you don't care about? Sure glad no one else on this thread does that.

Seriously I'm also very glad you've found the perfect camera. Although now it sounds a little like you think there may be more to a camera than the dynamic range of the sensor. Too bad no one else has ever made that point.

 

[jrista]

Well, it's clear this thread will be chock full of sarcasm. :

Have at it, ppl. Real data. That's what it's all about. For those who really want to know what the difference between Canon sensor/ACD IQ vs. Exmore IQ is, I hope the data I provide will help you figure that out. I'll try to provide more example RAW images over the next few days. I don't want to conclude for everyone...just, if your interested, download the RAW images and see for yourself. Draw your own conclusions.

 

[zlatko]

Looks like the a7r will be your landscape cam. Now you can shorten your posts to a link to this thread when the issue rises.

As for Lee jay, I would say a bounce flash.

Yes, as should be quite obvious, a flash is often a great way to control scene contrast. That was a 550EX fired into the ceiling.

You do realize the whole point was to demonstrate DR, right? Oh, and um...I don't have a flash for the A7r...so, doing a comparison with flash just wasn't an option.

That's a good demonstration of the superiority of the Sony sensor in a situation that few photographers encounter, i.e. underexposing a static scene by 4 stops in order to "control" highlights. It's not so much a test of dynamic range as it is a test of fixing really, really bad underexposure. A near pitch black exposure of a room is a textbook example of a bad photo, one that goes in the trash. I mean look how intensely dark the unprocessed photo is ... and you want it to end up like the processed version?!

Sure, it's also a test of dynamic range. But for whom? For the photographer who can't be bothered to use even a single light or blend bracketed exposures or shoot earlier/later in the day, etc.? For the photographer whose only notion of controlling contrast is the Lightroom shadow & exposure sliders?

For ~ 150 years photographers have faced the same problem of controlling contrast, and that example is not how they did it. The fact that you can do so now is excellent — if you're hellbent on doing it that way. But what happens when you encounter a situation that is even beyond the Exmor sensor? Do you just give up? Look for a newer sensor? Or do you perhaps think about some technique other than the shadow & exposure sliders, such as common photographic techniques that have worked with virtually any film/sensor ever made?

But if you need to do this — if no other method is acceptable — the Exmor is your thing. Adopt it, use it, love it. In that case, it's the only way to go right now. If this method is such a priority for someone, why in the world would they wait for Canon when they can have it *right now* and presumably solve all of they DR headaches?

 

[KeithBreazeal]

5D Mark III got the job done. I'm happy with my choice of "tools".
Mustang Sunrise Photo Tour CCA 2014 Sun1099 © Keith Breazeal by Keith Breazeal Photography, on Flickr

B-29 FiFi Cockpit Mather 7 June 2014 3334 © Keith Breazeal by Keith Breazeal Photography, on Flickr

F-86 Hell-Er Bust X Sunset CCA 2013 Sat 1331 web 2000 © Keith Breazeal-2 by Keith Breazeal Photography, on Flickr

 

[zlatko]

5D Mark III got the job done. I'm happy with my choice of "tools".

Nice work! I'm happy too. Some people claim the 1D / 5D series sensors don't meet their very high image quality standards. Presumably they have higher image quality standards than a long list of highly skilled and successful Canon photographers around the world.

 

[jrista]

@zlatko: It's not about any of that. It's simply data. There are lots of debates that go on about DR. There aren't very many RAW files that people can download and play with themselves. That's all this is about. I'm not trying to push anything with this thread. I'm just trying to provide data. People make decisions. It's useful to have data to back up those decisions.

YES, this is an extreme example. It's not intended to be a totally realistic demonstration. All it is supposed to do is give people who may have questions about what it really means to have more DR the ability to see for themselves. To actually download RAW files that they can open up in their preferred editor, work with themselves, and...see what's what. If some people conclude that more DR does nothing for them, fine. I don't care.

I simply care about providing some concrete data. The DR debate is never going to end until the playing field is level. So it's going to rage on. At least I can provide something people can reference. That's all this is. I am hoping I'll have the opportunity of photographing some landscape scenes tomorrow to provide some more realistic and balanced examples than this. Personally, I'm impressed with how well the A7r holds up under a 5-stop push (especially given it's compression). I don't think that's a particularly common scenario...but the data held up very well, all things considered. For smaller prints, like an 8x10, it's entirely viable.

It won't be any surprise to me if most of the responses to this thread are like yours, this is a Canon community, so it makes sense that people will decry the value of having more DR, defend their preferred brand, and even get hostile. That seems to be the nature of this community (sadly). Well, so be it. There's the data. Shred it as you will.

 

[weixing]

@zlatko: It's not about any of that. It's simply data. There are lots of debates that go on about DR. There aren't very many RAW files that people can download and play with themselves. That's all this is about. I'm not trying to push anything with this thread. I'm just trying to provide data. People make decisions. It's useful to have data to back up those decisions.

YES, this is an extreme example. It's not intended to be a totally realistic demonstration. All it is supposed to do is give people who may have questions about what it really means to have more DR the ability to see for themselves. To actually download RAW files that they can open up in their preferred editor, work with themselves, and...see what's what. If some people conclude that more DR does nothing for them, fine. I don't care.

I simply care about providing some concrete data. The DR debate is never going to end until the playing field is level. So it's going to rage on. At least I can provide something people can reference. That's all this is. I am hoping I'll have the opportunity of photographing some landscape scenes tomorrow to provide some more realistic and balanced examples than this. Personally, I'm impressed with how well the A7r holds up under a 5-stop push (especially given it's compression). I don't think that's a particularly common scenario...but the data held up very well, all things considered. For smaller prints, like an 8x10, it's entirely viable.

It won't be any surprise to me if most of the responses to this thread are like yours, this is a Canon community, so it makes sense that people will decry the value of having more DR, defend their preferred brand, and even get hostile. That seems to be the nature of this community (sadly). Well, so be it. There's the data. Shred it as you will.

Hi,
IMHO, I think a lot of Canon users know that DR of Canon camera is not the best and know the importance of DR, but they also try to point out that there is a lot of aspect that make up of a good camera not just DR... there is no perfect camera out there, so there is always some compromise when choosing a camera or camera system.... may be many user just don't put DR high in their priority list...

Also, IMHO, many users here get annoyed not because they want to defend their preferred brand (in this case Canon), but the fact that some of you keep bring up the same thing again and again in thread that are not related to DR.

Just my S$0.02 and enjoy your new toy... I mean camera...

Have a nice weekend.

 

[emko]

since i got a A7r i don't need to do bracketing and merging in PS on most pictures anymore to complete my pictures, i understand some people shoot different things that don't have the need for more DR but there are people who do need it. For example i shoot a street at night i under expose so i don't get blown out light bubbles on the street lights while the 5D3 can bring up the shadows/exposure it gets noise a lot faster then the A7r. I had problems also with shooting city sunsets where i expose for the sky and end up with dark city when i bring up the shadows A7r is amazing.

 

[LetTheRightLensIn]

Looks like the a7r will be your landscape cam. Now you can shorten your posts to a link to this thread when the issue rises.

As for Lee jay, I would say a bounce flash.

Yes, as should be quite obvious, a flash is often a great way to control scene contrast. That was a 550EX fired into the ceiling.

it is, but it doesn't work when out in nature and shooting scenics

 

[LetTheRightLensIn]

Looks like the a7r will be your landscape cam. Now you can shorten your posts to a link to this thread when the issue rises.

As for Lee jay, I would say a bounce flash.

Yes, as should be quite obvious, a flash is often a great way to control scene contrast. That was a 550EX fired into the ceiling.

You do realize the whole point was to demonstrate DR, right? Oh, and um...I don't have a flash for the A7r...so, doing a comparison with flash just wasn't an option.

That's a good demonstration of the superiority of the Sony sensor in a situation that few photographers encounter, i.e. underexposing a static scene by 4 stops in order to "control" highlights. It's not so much a test of dynamic range as it is a test of fixing really, really bad underexposure. A near pitch black exposure of a room is a textbook example of a bad photo, one that goes in the trash. I mean look how intensely dark the unprocessed photo is ... and you want it to end up like the processed version?!

Sure, it's also a test of dynamic range. But for whom? For the photographer who can't be bothered to use even a single light or blend bracketed exposures or shoot earlier/later in the day, etc.? For the photographer whose only notion of controlling contrast is the Lightroom shadow & exposure sliders?

Go to a forest, shoot at the edge where there is direct sun and shade. Go inside of a forest where sunbeams are coming in.

And if you hit something beyond even exmor, well them you struggle in PP or skip the scene. But at least there are now less scenes you need to skip. Why fight so hard to make that seem like a bad thing??

 

[Viggo]

Appriciate the data Jrista, after all, a picture is worth a thousand words ;D

I think this is a much better and more constructive approach. A friend bought the A7r and showed me a few shots, he HAVE to expose for highlights in his landscape shots, and even showed me +3,73 push from a 4 min exposure and it looked fantastic. But as you mention, there are other things to a camera and other things to shoot than landscape which I would rather have over more DR. But if the 1dX is what is and 2,5 stops more range, would I not buy it or be even happier?

 

[LetTheRightLensIn]

5D Mark III got the job done. I'm happy with my choice of "tools".
Mustang Sunrise Photo Tour CCA 2014 Sun1099 © Keith Breazeal by Keith Breazeal Photography, on Flickr

B-29 FiFi Cockpit Mather 7 June 2014 3334 © Keith Breazeal by Keith Breazeal Photography, on Flickr

F-86 Hell-Er Bust X Sunset CCA 2013 Sat 1331 web 2000 © Keith Breazeal-2 by Keith Breazeal Photography, on Flickr

1. of course it can do well for many scenes
2. a couple of those you chose to keep in high contrast, which is fine and looks good for those shots, but what about when you don't want that look?
3. the middle either looks far worse larger than 1/2MP downscale or you used multiple shots and combined in PP (most likely) or you used lighting inside the plane

EDIT: and looking at larger sizes (still far from original, only 2MP vs 20ish MP), you can actually already see a lot of loss of quality in the first image, even though it's kept pretty hi contrast, you can get away with it, but on a big print or UHD large monitor it would show somewhat and that didn't even make shadows get very bright