Just for Jrista: 2014 Market Data

[privatebydesign]

You have fallen into the first trap of comparative capabilities testing, the same exposure settings, you need to bracket 1/3 stop for a full stop either way to get the optimal exposure for highlight retention from each system.

You yourself said your 7D had better highlight capabilities than your 5D MkIII you need to explore that in the 5D MkIII and A7r if you want to conduct a thorough comparison.

You know, you guys keep blaming me for moving the goal posts. You just moved them yourself, from the position you are always preaching from: No one cares about the science, they just care about making photos.

Well, that's the approach I took here. I didn't actually set the exposures to the same thing purposely. Look at the unpushed originals...the highlights are the same. I pointed the cameras, adjusted exposure to preserve the highlights (which was actually easier with the A7r, because of the EVF...which was horribly pixelated and showed a lot of moire... :-\ I took one shot with the A7r based on the EVF. I had to take a few with the 5D III to find the point where the highlights weren't clipped), and I took the shots. Both showed -3EV...I didn't TRY to make it -3EV...that's just what they showed. Not surprising, either, give the fact that both cameras were shooting the exact same scene. If there was a large discrepancy between the cameras metering, I'd have been suspicious that one of them was metering wrong, or metering with a different mode. I'm not generalizing about the EV either...it wasn't -2 2/3 EV in one and -3 EV in the other...both showed -3EV. I can't imagine it being correct any other way.

Oh, I also just looked at the A7r image...it says ISO 80. I did not choose that ISO...I'm not sure if that's it's default, but I don't think the A7r has a native ISO 80. I don't know what that does to DR...using expanded ISOs on Nikon Exmor cameras usually cost you a little DR, so the A7r image could very well not be the best possible exposure. Still, I wasn't trying to make the exposures the same...I was trying to preserve the highlights the same.

Fault me for being scientific. Fault me for not being scientific, and just being a photographer. LOL, it really doesn't matter what I do, it will never be sufficient. But I don't really care. It doesn't really matter either way. I'll be sharing the RAWs soon here. Then you guys can pick em apart to your hearts content.

Dude, don't over react, I made a simple suggestion that makes sense, I haven't changed anything, I have been entirely consistent for years on this subject, you can't expose different cameras the same EV value to get optimal results, it used to drive Mikael mad and I don't know or understand why, but it is true.

I appreciate it is an informal quick "test" example, I was adding to the conversation to help it be more authoritative, you don't see me being all paranoid do you? Relax, if you want the conversation I'll have it, if you don't then I won't.

My honest reaction when I imported your jpegs into LR was that they are both unusably bad, I wasn't surprised at that being the case for the 5D mkIII, I did expect the A7r to be "better" than it appears via jpeg at first glance though, yes it is much better than the 5D mkIII, but it is still unusable.

If you do want to continue the conversation I am interested to see RAW's and "optimal" exposures, something none of the test sites ever do.

 

[takesome1]

Alright. First photos with the A7r. Quick point and shoots. I like the IQ...hmm, going to have to see about the body. It's already demonstrated some weaknesses...well. I'll share my thoughts once I've had more time with it.

Sorry about the delay in getting these up. As soon as I exported the sample JPEGs, imgur decided to quit on me. Using PhotoBucket now.

I'm not going to give any of my own conclusions. Everyone can come up with their own. All I'm going to say is, I exposed to preserve the highlights outside the window. The camera meters showed -3EV. I had to boost exposure +4 to result in the same kind of ambient lighting of the room that my eyes saw. I'll share the RAW files soon, along with a couple GIF images to directly compare.

Oh, and I won't "mislabel" anything.

Original Exposures (1/30s f/4 ISO 100):


Canon 5D III


Sony A7r

Pushed Exposures (+4 EV, -100 Highlights):


Canon 5D III


Sony A7r

Fill size jpeg images for the pushed versions:

Canon: http://i1375.photobucket.com/albums/ag461/jrista/A7rvs5DIII-ExposeforHighlightsPushShadows4stops-2_zps8fed09de.jpg
Sony: http://i1375.photobucket.com/albums/ag461/jrista/A7rvs5DIII-ExposeforHighlightsPushShadows4stops_zps0cca615c.jpg

Let the ridicule parade begin. Just, don't be mean. That goes for people on either side of the debate. It doesn't need to be a war. Those of you who don't care about having more DR, great, wonderful. You already have equipment that satisfies you, move on with your lives. Those who do care about having more DR, well, you can have it. There are cameras out there that offer a lot more, some of them are not extremely expensive and compatible with your Canon lenses thanks to Metabones. Don't cut yourself short...if you have a use for more DR, more DR is easy to have.

I see noise in the landscape picture in the upper left hand corner in the A7R pushed picture.
This landscape picture must have been shot with a Canon camera and the A7R is picking up on the detail of the noise.

 

[takesome1]

I see noise in the landscape picture in the upper left hand corner in the A7R pushed picture.
This landscape picture must have been shot with a Canon camera and the A7R is picking up on the detail of the noise.

LOL. ;D Ironically, yeah, all the landscape pictures were taken with Canon cameras. They do have their problems with noise as well. I could share those, if I could find them...I think some of those were taken with my original 450D, which actually wasn't as bad as the 7D or 5D III at low ISO (again, ironically.)

Well on the serious side.

Lets put this test in perspective to the non technical guys that just want to take good shoots.

You have an A7R and the 5D III in hand.
In the digital age memory is cheap, you just underexposed a shot by 4 or 5 stops. You look at your screen and your review shows black. You screwed up. If it is a picture of bigfoot and he just ran off well you try and push the picture and maybe it matters. But landscapes, buildings and front room chairs do not run away. You take another shot and get it right. So no benefit to DR here. So I look at your pictures and say they are meaningless as far as seeing how I would improve. I know how to delete bad pictures, you just click on the trash button.

However as I said before you have an A7R and a 5D III in hand.

So you go to the park mid day with the sun high, you find the edge where there are a few trees and a lake off to the side. Your buddy goes and stands under a tree, you try and capture the clouds and the lake and you look at your viewfinder and either one or the other blows out. Shade or sky take your pick. Use both bodies, like I said both are in hand. When you get home this shot matters but you want to improve it. You put both shots in LR and you work away. At the end of your PP one will be better than the other. That is the winner for you. It may not be the winner in someone else's hand but it will be in yours.

Real life situations, real pictures those are the ones that matter. That is the test. It is not hard to set the test up real life, we all know why we want more DR.

Underexposing, pushing it up and down, putting the lens cap on in the end are meaningless. End result wins.

Don't believe the results? You say here are the Raw files, decide for yourself.

 

[sarangiman]

No it wouldn't, a +100 shadow lift is a +100 shadow lift, it doesn't matter where the highlights are, or even if there are any. The entire image is as irrelevant as it was in sarangiman's crop which all the DR'ers thought was "amazing", see what you are doing here? Trying to make what you want/expect to see fit into what you actually do see.

This is completely misleading. A +100 shadow lift is not just a +100 shadow lift as you're implying. Whether or not you see banding/noise depends on where the tones you're pushing initially resided in the 16-bit Raw file.

It's pointless to have any other discussions until you at least appreciate that.

 

[sarangiman]

Seems you're more reasonable now, in most regards?

Isn't it amazing how people are reasonable when they agree with you? :

Yes, that's it. I'm happy he's agreeing with me. Not happy that he's:

[list type=decimal]
[*]No longer making claims about a camera he hadn't owned or even touched (5D3)
[*]Making claims about highlight headroom without doing a controlled comparison between Nikon and Canon cameras
[*]Irrelevantly claiming that my D800/5D3 results were wrong b/c I didn't give both cameras ~0.5EV more exposure to totally get near clipped highlights in the sky. Yes, b/c that would've changed the end result.
[/list]

It couldn't be that I'm actually happy b/c he's no longer making those illogical statements, right?

Must be some psychological comfort in finding someone who's agreed with me. Yes, that must be it.

 

[privatebydesign]

No it wouldn't, a +100 shadow lift is a +100 shadow lift, it doesn't matter where the highlights are, or even if there are any. The entire image is as irrelevant as it was in sarangiman's crop which all the DR'ers thought was "amazing", see what you are doing here? Trying to make what you want/expect to see fit into what you actually do see.

This is completely misleading. A +100 shadow lift is not just a +100 shadow lift as you're implying. Whether or not you see banding/noise depends on where the tones you're pushing initially resided in the 16-bit Raw file.

It's pointless to have any other discussions until you at least appreciate that.

Unless you can explain better why that is misleading, after all the RGB values increase exactly the same numbers as an exposure slider lift does, then I agree, you and I have nothing to discuss.

What, specifically, is misleading?

 

[sarangiman]

No it wouldn't, a +100 shadow lift is a +100 shadow lift, it doesn't matter where the highlights are, or even if there are any. The entire image is as irrelevant as it was in sarangiman's crop which all the DR'ers thought was "amazing", see what you are doing here? Trying to make what you want/expect to see fit into what you actually do see.

This is completely misleading. A +100 shadow lift is not just a +100 shadow lift as you're implying. Whether or not you see banding/noise depends on where the tones you're pushing initially resided in the 16-bit Raw file.

It's pointless to have any other discussions until you at least appreciate that.

Unless you can explain better why that is misleading, after all the RGB values increase exactly the same numbers as an exposure slider lift does, then I agree, you and I have nothing to discuss.

What, specifically, is misleading?

It's not just about the amount of the push. It's (the signal of) what you're pushing.

And, no, I'm sorry, I don't have time to write a novel on this right now. Honestly, I'm not being snarky. Perhaps someone else can explain.

 

[privatebydesign]

No it wouldn't, a +100 shadow lift is a +100 shadow lift, it doesn't matter where the highlights are, or even if there are any. The entire image is as irrelevant as it was in sarangiman's crop which all the DR'ers thought was "amazing", see what you are doing here? Trying to make what you want/expect to see fit into what you actually do see.

This is completely misleading. A +100 shadow lift is not just a +100 shadow lift as you're implying. Whether or not you see banding/noise depends on where the tones you're pushing initially resided in the 16-bit Raw file.

It's pointless to have any other discussions until you at least appreciate that.

Unless you can explain better why that is misleading, after all the RGB values increase exactly the same numbers as an exposure slider lift does, then I agree, you and I have nothing to discuss.

What, specifically, is misleading?

It's not just about the amount of the push. It's (the signal of) what you're pushing.

And, no, I'm sorry, I don't have time to write a novel on this right now. Honestly, I'm not being snarky. Perhaps someone else can explain.

I don't need a novel, I need a cohesive thought that can be backed up with supporting information. You think you know but you don't, you think you can explain but you can't, very convincing........

What don't you understand about RGB values in the low single digit percentages being shadow detail?

 

[privatebydesign]

0.2/0.2/0.1% -> 15.1%/14.%/9.6%

Very significant shift in "pixel levels as percentages."

Are these numbers unacceptable because I lifted information that looks black on screen, but wasn't black in real life, just so it would look realistic, as in "what I see with my own eyes", in the final photo? I mean, this is an invalid test of dynamic range, is it not? Oh, BTW, look at the histogram. I was fully ETTRed...there was no room left to "optimize" my exposure. The highlights were already riding up the right-hand wall. Some of them ARE clipped in these photos.

Here are both histograms:



BOTH cameras were fully ETTRed. Note, however, that in the A7r histogram...it moves inward, at BOTH ends.

For those who are interested, even though these are just "from the hip" preliminary examples, here are the RAWs:

http://1drv.ms/1ol6Km5

It isn't unimpressive, but it still has no practical value for me, the window blooming and complete lack of tonal detail in three quarters of the A7r shot makes the image unusable even though you can remove the noise. Obviously the 5D MkIII shot is also unusable.

In this particular scenario, whilst I understand is an off the hip trial, the A7r certainly doesn't convince me to get one for my real estate work as I would still need to bracket and blend, or light the room, or any number of other techniques I currently use.

I look forward to the fuller tests.

 

[sarangiman]

I don't need a novel, I need a cohesive thought that can be backed up with supporting information. You think you know but you don't, you think you can explain but you can't, very convincing........

What don't you understand about RGB values in the low single digit percentages being shadow detail?

Why are you still talking about percentages on a histogram that represents the Raw values converted to ProPhoto RGB internal space and then converted to sRGB output and then converted to a percentage?

I analyze *actual* Raw numbers using RawDigger. Quit convoluting this analysis. I don't even know what your point is anymore.

My point is that you can't just do +3 or +4 EV pushes for these sorts of comparisons on brighter tones to demonstrate the differences. If you don't see a difference, you're not pushing deep enough shadows. If you don't have deep enough shadows while you're exposing for highlights, then your scene doesn't have enough DR for it to matter whether or not you're shooting Exmor or Canon.

I never said it's just about how many stops you can push. It's about what tones you're pushing. You seem to be stuck on this 'but XYZ percentage *is* a shadow'... what? First of all, what's a shadow depends on your exposure, and second, my signal of 7 out of 16,000 on a linear scale would be 0.04%. So, no, single digit percentages don't impress me.

I think I know? Not even going to respond to such a patronizing comment considering I do full SNR analyses of sensors...

 

[raptor3x]

So it seems the reason I've been seeing posterization issues and Sarangiman hasn't when doing the emulated high ISO trick is due to Sony's compression algorithm. Good to know I'm not the only person seeing this.

 

[sarangiman]

So it seems the reason I've been seeing posterization issues and Sarangiman hasn't when doing the emulated high ISO trick is due to Sony's compression algorithm. Good to know I'm not the only person seeing this.

To be fair, I rarely see it with my A7R, so I think some of it may be due to your use of the A7 (IIRC)?

But, yes, Sony's compression is very, very annoying under certain circumstances. And that's a great reference/study you linked to, by the way.

 

[Loswr]

I fully plan to create some "optimal" exposures, that was actually the entire point of renting the A7r in the first place. It also so happens that this weekend is the last weekend for fall colors in the mountains. It might be raining, not sure yet, but I'm going to try to get some landscape photos, every scene with both cameras, bracketed, etc. I don't have a lot of time to go hunting for awesome landscape scenes, so don't expect any kind of impressive artwork, but I always planned to do a more rigorous comparison between the two.

I've seen lengthy geometric proofs for the fact that a square is the largest rectangle that can be inscribed in a circle, and even lengthier algebraic proofs that 1 + 1 = 2. At the end of your testing, I'd be quite surprised if don't conclude that the a7R has more DR than the 5DIII and you can push the shadows harder.

Still, there's value in demonstrating to ourselves that which we expect to be true. Enjoy!

 

[takesome1]

I fully plan to create some "optimal" exposures, that was actually the entire point of renting the A7r in the first place. It also so happens that this weekend is the last weekend for fall colors in the mountains. It might be raining, not sure yet, but I'm going to try to get some landscape photos, every scene with both cameras, bracketed, etc. I don't have a lot of time to go hunting for awesome landscape scenes, so don't expect any kind of impressive artwork, but I always planned to do a more rigorous comparison between the two.

I've seen lengthy geometric proofs for the fact that a square is the largest rectangle that can be inscribed in a circle, and even lengthier algebraic proofs that 1 + 1 = 2. At the end of your testing, I'd be quite surprised if don't conclude that the a7R has more DR than the 5DIII and you can push the shadows harder.

Still, there's value in demonstrating to ourselves that which we expect to be true. Enjoy!

I think he will have that conclusion as we'll, that is the easy part.
The hard part of the question is there enough difference to matter.

I for one would be interested in seeing comparisons of pics taken in real situations that photographers face.
The shooter draws a conclusion, provides the RAW files and anyone can see if they draw the same conclusion.

This would be much better than the pointless technical banter that never proves anything or goes anywhere.

Also, the circle square exercise sounds like military funded peg research.

 

[sarangiman]

I for one would be interested in seeing comparisons of pics taken in real situations that photographers face.
The shooter draws a conclusion, provides the RAW files and anyone can see if they draw the same conclusion.

This would be much better than the pointless technical banter that never proves anything or goes anywhere.

This is a very fair, valid request. A little difficult to do, but very worthwhile. What makes it hard is that it's actually very difficult to find sunsets/sunrises that do bottom out a, say D810. So it's hard to show the real difference, i.e. 'what's possible'. And if you do find the right high dynamic range scene, you're probably a landscape photographer who woke up at 2:30 am to shoot a sunrise at a beautiful location, not do a head-to-head test which is fairly challenging to do with the quickly changing light of a good sunrise/sunset. You also have to bracket both cameras all over the place so that you can go back home and then find the one where the highlights are just short of clipping, or where ACR can recover detail/color to taste.

I'm not saying it's impossible, it's usually just hard to do well. Hopefully someone will do it (well). I'll try at some point, maybe, before I sell off my 5D3.

...so here I am. Trying to be a man of my word.

I respect that. Good luck!

 

[msm]

Oh, I agree about the blooming. That's pretty nasty. There are other problems with the A7r as well, if you pull down the RAWs. It has some pretty wicked aliasing and moire in the EVF. When I pointed at the blinds, it was pretty bad.

The focus is also NOT fast. There is only one out of several AF modes available for use with EF lenses, AF-S, which seems to be contrast detect. In better light, there seems to be a quick initial shift, then a slow contrast drive. It's difficult to see what's in focus in the EVF...but, I haven't messed with all the EVF features. Focusing is not intuitive, though.
...

Once you used it a while you won't even notice the aliasing and moire in the viewfinder . Also I don't know if you already have done this but it is hard to see things in the deep shadows through the VF unless you enable DRO.

I never even use AF through the metabones adapter, it just doesn't seem reliable enough. Anyways in landscapes I often want everything in focus and that is often impossible when viewing 36mp files at 100%, often the best compromise can be to step down far into diffraction land. Manual focus with the loupe seems to me to be the best way to find best balance between focus between foreground and background, on Canon that involves live view and suddenly the battery life on the Canon is just as bad as on A7R . With the evf I can also actually see what I am doing even in bright light.

I think that autofocus with native lenses is surprisingly fast for a contrast detect system however.

I won't get involved in these pointless discussions but I find the discussion about optimal exposure quite symptomatic. With exmor sensors you don't even have to bother thinking about getting the optimal exposure, you can focus on the other stuff and just shoot.

 

[msm]

Yeah, I enabled DRO. Still messing with all the settings. I do find the EVF to be pretty useless once you get down to around 1/30th second frame rate. I'm not sure why, but the EVF doesn't seem to use simulated exposures like the LCD screen on the back. The LCD screen maintains a high frame rate, but when I use the EVF, it only refreshes at the rate of your shutter speed...so a 1/4s shutter results in truly hideous EVF performance.

That doesn't happen with me, it gets slow in low light when I do multiple second exposures though but then it is both the evf and screen lagging.

 

[msm]

Yeah, I enabled DRO. Still messing with all the settings. I do find the EVF to be pretty useless once you get down to around 1/30th second frame rate. I'm not sure why, but the EVF doesn't seem to use simulated exposures like the LCD screen on the back. The LCD screen maintains a high frame rate, but when I use the EVF, it only refreshes at the rate of your shutter speed...so a 1/4s shutter results in truly hideous EVF performance.

That doesn't happen with me, it gets slow in low light when I do multiple second exposures though but then it is both the evf and screen lagging.

Oh is it related to half pressing shutter button and having the "Live View Display" menuitem in second menu submenu 2 set to "Setting Effect Off"? It seems to work much better with that set to on.

 

[sarangiman]

With exmor sensors you don't even have to bother thinking about getting the optimal exposure, you can focus on the other stuff and just shoot.

Yeah, I totally agree. I always have to get just the right amount of ETTR with the 5D III, so I'll usually (with landscapes, I work it a bit different with birds and wildlife) take a couple test shots to make sure I have the exposure right. With the A7r, it just isn't an issue...get the exposure generally right, take the shot. If your off by a third of a stop, either over or under exposed, it doesn't matter. You can usually recover it either way.

Now this is just crazy talk. Stop making excuses for being lazy and sloppy. You should learn to focus on everything all at once. You should even manually focus, like the pros did decades ago. And make sure you don't leave any more than 0.5 EV highlight headroom and then try to show there's noise in the shadows - you'll be lambasted.

Ok, on a more serious note: isn't that what I've been trying to say all along? That equipment getting out of the way opens up creative potential?

Anyway, make sure you don't take 'don't have to bother thinking about getting the optimal exposure' too far... actually depriving the sensor of light (via faster shutter speed or smaller aperture) does cost you image quality, no matter what sensor.

But, yes, it's a very 'free-ing' thing to not have to worry about excessive read noise - generally the only cost you'll pay by underexposing with Exmor is image detail that just looks like it was shot at a higher ISO. And that's usually not exactly devastating, given the ISOs people willingly accept shooting at these days.

 

[Aglet]

from a few pages back:

@PBD likes to make points demonstrating with his 1DS 3 but he is, in a way, CHEATING.
Not because of misdirection with his example, but because the 1ds3 is likely the best FF camera, IQ-wise, that Canon's ever made.

It has the best color response I know of, other than the 1000D
It has marginally better dynamic range than most other Canon's.
But, IMO, most importantly, it was PRE-Digic 4!
the FPN issues became a serious problem AFTER digic 3

PBD, you can't make those same examples as effectively with a 5d2, 5d3 or 1dx.
FPN read noise issues are worse with Digic 4 and Digic 5 than they were with the old Digic 3.

You don't see me making bad stripey noise examples with my 40D by pushing shadows.. because it doesn't have the same level of read noise banding problems bodies like the 7D have.
And that's why my 40D is still in my stable, it works, it's cleaner than subsequent models from Canon up until the 70D, and hopefully the 7d2.

Shamefully unfair comparisons. new cameras do not perform as well as those older ones.