p.7 #1Sony A7Rii compared with Sony A7R and Canon 5DSr
timparkin wrote:
Ermm - the "amendments" were:
1) [...] 2) [...] 3) [...]
None of these are amendments in my eyes, they were additional information I thought was interesting to people.
Not amendments to the test, but amendments to a hastily written article with explanations and discussion that might have been included in the first place.
The update about the exposure problems due to inconsistent lighting was definitely an amendment and rightfully should have been in the article near the top, as a caveat at least.
timparkin wrote:
As for the exposure offset, the ambient conditions changed but that doesn't nullify the test.
Not exactly "nullify" the test, but make it very dubious. When testing technical issues, extraneous variables must be controlled, and exposure & ambient lighting can be controlled.
For proper testing, one shoots under consistent lighting. If it is not steady clear sunlight or electric light or electronic flash, then there are variables between shots and between cameras that are not being properly and adequately controlled.
timparkin wrote:
I never said anything about the difference in exposure - I just explained the methodology.
A well-written test report explains such differences, even when they are embarrassing.
timparkin wrote:
If the light changes what are you supposed to do?
Wait minutes for consistent lighting that covers all the shots of the test. If that is not possible, then come back when lighting is consistent. If that is not possible, then don't call it a "test", call it a "tryout" or an "initial impression". If you have to rush into print (deadlines?), then whatever you do, be upfront about the lighting, contrast, and colour balance problems very early in the article.
Lighting was not a variable under test.
As an uncontrolled variable it introduces lots of problems for these kind of tests: exposure variations, contrast variations, colour balance variations. You made a simple adjustment for exposure, and a post-test adjustment for colour balance. Nothing could compensate for contrast variation. Adjustments for extraneous variables introduce other variations. It is like building a pre-fab house where one of the walls is delivered a meter and a third too short so rocks are piled to form a base to adjust for that. Sure the roofline would be at spec, but there could be no confidence in the house.
It is simply basic experimental design to eliminate variables that are not actually being tested.
timparkin wrote:
Does it make testing in the landscape irrelevant?
No, not the concept of testing in the landscape, but the lack of control of variables greatly reduces the relevance of this particular test. When exposure in daylight on a landscape changes by a stop or a stop and a third, then I know from experience that the contrast almost always changes (which affects Dynamic Range) and the colour balance usually changes too. Perhaps you have noticed similar changes when you have been out.
It is not difficult to make three shots with three cameras in three minutes under steady light, say, clear sunlight between the hours of 10 am and 2 pm. That would be doing a controlled test in the landscape. (I know, difficult to find that in Britain at times, as it has been difficult here the last two weeks.)
timparkin wrote:
I've also uploaded a rawdigger comparison of the files showing that the comparison is good but I don't suppose anybody is interesting in actually working their way to conclusions when jumping is so much more fun!
Thank you, Tim, for participating here on FM as well. (PS the portraits of you on both sites look very different; like two different people.)
I do not understand what the RawDigger comparison is supposed to mean or how to analyze it. I would like to know.
But it seems to me, as someone with a scientific and engineering approach when it comes to the technical issues in photography, that the "test" and the article were rushed, out of an understandable sense of eagerness and wanting a scoop due to the newness of the camera, and they are where the jumping to conclusions occurred. I concur with the impulse but not the results.
It would be interesting to see the test repeated with the lighting controlled (and therefore colour balance and contrast). Also, please read the other responses where people have desired to see additions to the test for ISO and low light.
p.7 #2Sony A7Rii compared with Sony A7R and Canon 5DSr
gdanmitchell wrote:
Most likely because it unfairly maligned the performance of a Canon camera.
Fast6 wroteTest methodology or results aside, it's curious that this topic has gotten 4x the attention in the Canon forum than it has in the Sony forum.
At least the Canon can attain 14 bits in continuous mode. The Sony degrades to 12 bits in continuous mode. The IQ comparison should also be made between the Sony at 12 bits and Canon at 14 bits. It's like the Sony can't walk and chew gum at the same time.
It is ridiculous for a camera at that price to be so crippled. I suspect that when Nikon gets the newer Sony sensor, they would not put up with that limitation. The D810 is still a very viable body in that context.
p.7 #3Sony A7Rii compared with Sony A7R and Canon 5DSr
RustyBug wrote: Sneakyracer wroteNo surprise that the Sony A7RII is THE Best choice for Landscape/Architecture photography at base iso. No question about it.I might revise this a bit to say that the A7R II (predicated upon A7R history) is "best" at showing its strengths in landscape / architectural settings, where its base ISO can be well harnessed via slower exposures and tripod utilization to work with stopped down apertures well with its low base ISO, rewarding the shooter with an opportunity to harness more light for reduced noise files in certain uber-dynamically challenging situations.
The farther you go beyond the base ISO, the strengths of it's difference are decreased accordingly.
I'm inclined to think that folks @ Leica S2, Phase One, Hasselblad, etc. might have a difference of opinion regarding the Sony as having a "no question about it" THE Best choice for ...
I have a PhaseOne IQ160 back and use it with Rodenstock HR-W lenses on an Arca Swiss RM3Di body (and sometimes on a Hasselblad H body). Overall that system is hard to fault but it is slower to use (setting up is not bad but it also requires an LCC exposure and focus is by scale using a table). The lens quality is just amazing though. Dynamic Range wise the newer Sony CMOS sensors are better but the color depth and color differentiation of the Phase back is stunning. The Phase does have significantly better dynamic range compared to the Canons.
What I am seeing in spades with the A7RII files (available so far) which I started to see in the D810 files and afterwards in the Sony 44x33mm CMOS sensor files is a depth and smoothness that make the files incredibly malleable. They resemble somewhat what one gets from a RED cinema camera and an ARRi Alexa camera. Awesome files for color grading and post-production work. There is no bracketing in motion imagery so single frame capture needs to be of the highest quality and the newer breed of Snoy EXMOR sensors deliver just that.
Of course lot's of details/quirks have yet to be discovered about the A7RII's image quality and features/performance pending more rigorous testing. But overall looks like it is a game changing product and miles ahead of the A7R.
For lower light / high iso use and long exposures the 5DSr would be my choice though. The A7RII (like most of Sony EXMOR cameras) is a "isoless" camera with a low base iso. At higher isos the advantage over the Canon sensors is basically nill.
p.7 #4Sony A7Rii compared with Sony A7R and Canon 5DSr
Monito wrote: timparkin wroteErmm - the "amendments" were:
1) [...] 2) [...] 3) [...]
None of these are amendments in my eyes, they were additional information I thought was interesting to people.Not amendments to the test, but amendments to a hastily written article with explanations and discussion that might have been included in the first place.
The update about the exposure problems due to inconsistent lighting was definitely an amendment and rightfully should have been in the article near the top, as a caveat at least.
timparkin wroteAs for the exposure offset, the ambient conditions changed but that doesn't nullify the test.Not exactly "nullify" the test, but make it very dubious. When testing technical issues, extraneous variables must be controlled, and exposure & ambient lighting can be controlled.
For proper testing, one shoots under consistent lighting. If it is not steady clear sunlight or electric light or electronic flash, then there are variables between shots and between cameras that are not being properly and adequately controlled.
A well-written test report explains such differences, even when they are embarrassing.
Wait minutes for consistent lighting that covers all the shots of the test. If that is not possible, then come back when lighting is consistent. If that is not possible, then don't call it a "test", call it a "tryout" or an "initial impression". If you have to rush into print (deadlines?), then whatever you do, be upfront about the lighting, contrast, and colour balance problems very early in the article.
Lighting was not a variable under test.
As an uncontrolled variable it introduces lots of problems for these kind of tests: exposure variations, contrast variations, colour balance variations. You made a simple adjustment for exposure, and a post-test adjustment for colour balance. Nothing could compensate for contrast variation. Adjustments for extraneous variables introduce other variations. It is like building a pre-fab house where one of the walls is delivered a meter and a third too short so rocks are piled to form a base to adjust for that. Sure the roofline would be at spec, but there could be no confidence in the house.
It is simply basic experimental design to eliminate variables that are not actually being tested.
No, not the concept of testing in the landscape, but the lack of control of variables greatly reduces the relevance of this particular test. When exposure in daylight on a landscape changes by a stop or a stop and a third, then I know from experience that the contrast almost always changes (which affects Dynamic Range) and the colour balance usually changes too. Perhaps you have noticed similar changes when you have been out.
It is not difficult to make three shots with three cameras in three minutes under steady light, say, clear sunlight between the hours of 10 am and 2 pm. That would be doing a controlled test in the landscape. (I know, difficult to find that in Britain at times, as it has been difficult here the last two weeks.)
Thank you, Tim, for participating here on FM as well. (PS the portraits of you on both sites look very different; like two different people.)
I do not understand what the RawDigger comparison is supposed to mean or how to analyze it. I would like to know.
But it seems to me, as someone with a scientific and engineering approach when it comes to the technical issues in photography, that the "test" and the article were rushed, out of an understandable sense of eagerness and wanting a scoop due to the newness of the camera, and they are where the jumping to conclusions occurred. I concur with the impulse but not the results.
It would be interesting to see the test repeated with the lighting controlled (and therefore colour balance and contrast). Also, please read the other responses where people have desired to see additions to the test for ISO and low light. ...Show more →timparkin wroteI never said anything about the difference in exposure - I just explained the methodology.timparkin wroteIf the light changes what are you supposed to do?timparkin wroteDoes it make testing in the landscape irrelevant?timparkin wroteI've also uploaded a rawdigger comparison of the files showing that the comparison is good but I don't suppose anybody is interesting in actually working their way to conclusions when jumping is so much more fun!
I have a PhD in computational engineering and have done "tests" on live running 660mw generators providing the UK public with their daily power. You don't always get to control variables during a test - nor is it essential. Thanks for your definitions though.
p.7 #5Sony A7Rii compared with Sony A7R and Canon 5DSr
I can run a test of how tire "A" responds while driving over a bed of nails compared to tire "B" while running over the front lawn.
That is somewhat of an uncontrolled "test" compared to running both sets of tires around the same test track to evaluate the differences in the tires. Obviously a gross over-exaggeration, but the point @ controlled vs. uncontrolled is not an insignificant matter in the challenges it presents for both presentation / and interpretation.
Thanks for the info @ Raw Digger ... but, I'll have to defer interpretation of approach / interpretation to others (i.e. Snapsy, etc.), as I'm not up to speed on what I'm looking at with what you've presented (explanation welcome, but as long as others can guide / interpret it properly ... I'll try to follow along).
I still think it would be a very good starting point to simply shoot the same lighting, same subject, same WB, same ISO setting, same manually set exposure ... IDK, maybe just use an incident meter to determine the amount of light falling on the subject and set exposure accordingly ... and then show the straight results. If you want to show the extreme push / pull activities available to those files, sure carry on and add them to the presentation, but I'm not sure why you have chosen to use a seemingly convoluted approach instead?
Is there any particular reason why that can't be done?
(PhD in computational engineering)
Then, you could show the results from subsequent deviation testing(s) of your choosing, i.e. -6 EV or +5 EV or whatever you want from there. Does that make sense, or am I missing something here?
p.7 #6Sony A7Rii compared with Sony A7R and Canon 5DSr
RustyBug wrote:
I can run a test of how tire "A" responds while driving over a bed of nails compared to tire "B" while running over the front lawn.
That is somewhat of an uncontrolled "test" compared to running both sets of tires around the same test track to evaluate the differences in the tires. Obviously a gross over-exaggeration, but the point @ controlled vs. uncontrolled is not an insignificant matter in the challenges it presents for both presentation / and interpretation.
Thanks for the info @ Raw Digger ... but, I'll have to defer interpretation of approach / interpretation to others (i.e. Snapsy, etc.), as I'm not up to speed on what I'm looking at with what you've presented (explanation welcome, but as long as others can guide / interpret it properly ... I'll try to follow along).
I still think it would be a very good starting point to simply shoot the same lighting, same subject, same WB, same ISO setting, same manually set exposure ... IDK, maybe just use an incident meter to determine the amount of light falling on the subject and set exposure accordingly ... and then show the straight results. If you want to show the extreme push / pull activities available to those files, sure carry on and add them to the presentation, but I'm not sure why you have chosen to use a seemingly convoluted approach instead?
Is there any particular reason why that can't be done?
(PhD in computational engineering)
Then, you could show the results from subsequent deviation testing(s) of your choosing, i.e. -6 EV or +5 EV or whatever you want from there. Does that make sense, or am I missing something here?...Show more →
Always intended to do that and it's been posted since last night. I think your comparison is ridiculous and I really don't think you understood the test (especially if you think tint affects raw files). I'd really like you to explain why a 1/60s exposure with 10ev is so radically different than a 1/15s exposure with 8ev because that is the difference you are complaining about. Yes a studio comparison is great and would have been easier than walking 6 miles with 8 cameras but I run a landscape photography magazine and our readers want to see real outdoor pictures. I
p.7 #7Sony A7Rii compared with Sony A7R and Canon 5DSr
Tim,
My comparison wasn't a comparison, it was hyperbole, noted as a gross exaggeration ... it was supposed to be ridiculous to illustrate the diff between a controlled test and an uncontrolled one.
As to 10EV @ 1/60 vs. 8EV @ 1/15 ... I'm good with that being equitable, but what you posted didn't reveal such an equitability from what I could see. Instead, giving a difference in exposure by more than a full stop, then pushing in post to try to compensate for the difference as a means of equalizing the unequal exposure at a pseudo ETTR. That's a different thing, from what I can tell. If I'm the one missing it here ... apologies for such and I welcome a better understanding of how it is I'm missing it.
Maybe, I (and many others) don't understand the test, so again, apologies if it's on my end. If your test served your readers faithfully (their interests and fair testing), well, you certainly have no reason to answer to the likes of me ... but, I am trying to understand how your methodology was fair testing. I mean, from what you showed they had two different exposures (noting both exposure math & image tonal values @ desat). To me, that suggests either A) a variance in exposure (not equitably compensated for as in your 10 EV vs. 8 EV example), or B) a variance in the camera's response (i.e. true ISO rating).
I'll take a look at your updates and see if I can now better understand, but I'm presently still scratching my head on how you felt that the original was equitable? Again, maybe its just on me (et al) ... but, there were several others (whom I respect their knowledge). If TheSuede were to chime in and clarify such that your testing was equitable and explain to us how we are "missing it", I'm sure that many would garner a renewed perspective, but currently, I'm at a loss to understand. Sorry for any angst this may have caused ... just trying to understand what you have shown (and are trying to show) us.
p.7 #8Sony A7Rii compared with Sony A7R and Canon 5DSr
snapsy wrote:
In all fairness to him it's difficult to do outdoor comparisons when there is cloud cover and changing LV. Since he accommodated the ambient light changes with exposure adjustments his original results are within the ballpark and valid,
+1 @ changing conditions challenges
So, why then did the images he posted have such a variance in tonal value ... if they had actually been equitably accommodated for the ambient variance? My take is that while there is a proposed compensation ... it wasn't an equitable one, based on the images he presented (or the true ISO's of the cameras aren't equitable) The tonal values of the images clearly showed disparity ... just trying to understand how that equates to proper accommodation / compensation to each sensor receiving the same amount of light as presented.
p.7 #9Sony A7Rii compared with Sony A7R and Canon 5DSr
timparkin wrote:
I have a PhD in computational engineering and have done "tests" on live running 660mw generators providing the UK public with their daily power. You don't always get to control variables during a test - nor is it essential.
Congratulations.
A test of a brand new camera does not have the constraints that a live power test does, nor the urgency.
It is essential to control variables as much as possible.
You traded off some of the integrity of the test in order to achieve speed of reporting. It was a choice you made. Simply waiting for consistent lighting would have made a less dubious test, one on a firmer foundation.
p.7 #10Sony A7Rii compared with Sony A7R and Canon 5DSr
timparkin wrote:
(especially if you think tint affects raw files)
"tint"? You mean the White Balance / Colour Balance settings, right?
The colour balance setting does not affect Raw files.
Changing lighting between shots in a test does affect Raw files, particularly in the differences between shadows and highlights. Brighter sun is generally warmer than dimmer sun as clouds thicken. The actual light recorded in the Raw files is affected, as is the contrast.
When light changes by a stop or a stop and a third, the contrast changes between highlights and shadows and locally. That is recorded in the Raw files. This has a big effect on the Dynamic Range and can't be compensated for in post-processing in any kind of meaningful way that wouldn't skew the results with introduced confounding variables.
Doing the test in 3 minutes (3 shots one each on 3 cameras) under constant lighting does not seem to be difficult.
p.7 #11Sony A7Rii compared with Sony A7R and Canon 5DSr
RustyBug wrote:
Tim,
My comparison wasn't a comparison, it was hyperbole, noted as a gross exaggeration ... it was supposed to be ridiculous to illustrate the diff between a controlled test and an uncontrolled one.
As to 10EV @ 1/60 vs. 8EV @ 1/15 ... I'm good with that being equitable, but what you posted didn't reveal such an equitability from what I could see. Instead, giving a difference in exposure by more than a full stop, then pushing in post to try to compensate for the difference as a means of equalizing the unequal exposure at a pseudo ETTR. That's a different thing, from what I can tell. If I'm the one missing it here ... apologies for such and I welcome a better understanding of how it is I'm missing it.
Maybe, I (and many others) don't understand the test, so again, apologies if it's on my end. If your test served your readers faithfully (their interests and fair testing), well, you certainly have no reason to answer to the likes of me ... but, I am trying to understand how your methodology was fair testing. I mean, from what you showed they had two different exposures (noting both exposure math & image tonal values @ desat). To me, that suggests either A) a variance in exposure (not equitably compensated for as in your 10 EV vs. 8 EV example), or B) a variance in the camera's response (i.e. true ISO rating).
I'll take a look at your updates and see if I can now better understand, but I'm presently still scratching my head on how you felt that the original was equitable? Again, maybe its just on me (et al) ... but, there were several others (whom I respect their knowledge). If TheSuede were to chime in and clarify such that your testing was equitable and explain to us how we are "missing it", I'm sure that many would garner a renewed perspective, but currently, I'm at a loss to understand. Sorry for any angst this may have caused ... just trying to understand what you have shown (and are trying to show) us....Show more →
I don't think you understand the test then. Here it is in simple steps:
1) Take a set of photos around the 'clipping point' of the camera (metered plus three approx - between a 1/10 and 1/40
2) Stop down six stops and take a set of pictures around that point (because we didn't know which of the photos from step one would be the clipping one
3) When we get back we find the picture from step one which is just clipping in the non-sky area and then choose the photo from step 2 which is 6 stops below
Now for whatever reason, and I'm presuming light changes here, the exposures for the Sony and Canon differed and by about a stop and a 1/3. However because we used a six stop difference based on the clipping point the relative difference in EV of the file exposure is the same even if the absolute EV is different.
I don't really understand that from being much different from my mentioned " 1/60s exposure with 10ev compared with a 1/15s exposure with 8ev ". Yes the light flux is different but I wouldn't expect that to make any absolute difference.
The tint differences are a red herring (or a magenta one?) because they are ignored by Rawdigger when calculating clipping point.
Testing outside is inherently difficult and you have to make sure you're making 'relative' changes - not setting absolute numbers. The only time it's possible to use absolute exposure values is when you've got the sun out (and then you need to be careful that there isn't a sensitivity difference in between the cameras).
p.7 #12Sony A7Rii compared with Sony A7R and Canon 5DSr
Monito wrote: timparkin wrote(especially if you think tint affects raw files)"tint"? You mean the White Balance / Colour Balance settings, right?
The colour balance setting does not affect Raw files.
Changing lighting between shots in a test does affect Raw files, particularly in the differences between shadows and highlights. Brighter sun is generally warmer than dimmer sun as clouds thicken. The actual light recorded in the Raw files is affected, as is the contrast.
When light changes by a stop or a stop and a third, the contrast changes between highlights and shadows and locally. That is recorded in the Raw files.
Well I'll quietly disagree with you here. Thick cloud doesn't affect white balance enough to make a darn difference and the diffusion of thick cloud won't affect the shadow highlight imbalance. I don't know where you're getting this information from but if you can prove it that would be great. As far as I'm concerned, a shadowless cloudy day is just like a great big softbox and although the light level may change this is just like changing the power on the softbox.
I suggest a little challenge. Go outside on a shadowless cloudy day and record two files within 20-30 mins that show a significant difference in shadow highlight balance/contrast/etc and also in tint (enough to throw results off) and I'll eat my hat. Until then I accept you don't like the test - you're welcome to your opinion.
p.7 #13Sony A7Rii compared with Sony A7R and Canon 5DSr
Tim,
Thanks for the augmented explanation. I think I follow your process a bit better. It sounds like though that you are discounting the impact of the lighting / exposure variation as significant vs. insignificant. That's pretty much where we are on separate perspectives, as you feel that the "clipping point" made things equitable ... yet, the images revealed unequal values (both sky and quarry).
Anyway, thanks for the explanation and the updates in your second article. I think the fact that your additional testing revealed the A7R II to have 2/3 stop LESS DR than the A7R introduces (at least somewhat) a component of a different result from your initial presentation. So, two diff tests, two diff outcomes ... which lends me to think that your first test contained more significance to the variation than you've given credence to.
Thus, which test should your readers being giving more credence to ... the one that renders a conclusion of the A7R II having better DR or the one that renders a conclusion that the A7R has better DR?
p.7 #14Sony A7Rii compared with Sony A7R and Canon 5DSr
Yes, now I'm even more confused. Is the A7r2 really showing 2/3 stop less dynamic range than the A7r-- but better color tonality so it looks, in actual images, like more?
p.7 #15Sony A7Rii compared with Sony A7R and Canon 5DSr
I mostly look forward to the price a year from now.
We can wait for Fred and some others to provide assessments. From the pool of testing we'll mostly likely see that it is a fine camera with a few livable quirks like every other camera.
p.7 #16Sony A7Rii compared with Sony A7R and Canon 5DSr
justruss wrote:
Yes, now I'm even more confused. Is the A7r2 really showing 2/3 stop less dynamic range than the A7r-- but better color tonality so it looks, in actual images, like more?
It's very borderline and seems to depend on quality. I've just run another three imatest series as I was a little dubious about that and have averaged the results to get the following.
average values of 10.9 stops for the Canon 5DS, 12.2 stops for the Sony A7R and 12.5 stops for the Sony A7Rii. Imatest was quite variable from run to run but these averages seems sensible when I compare the test with checking clipping in rawdigger and looking at the tones in the deep shadows.
p.7 #17Sony A7Rii compared with Sony A7R and Canon 5DSr
RustyBug wrote:
Tim,
Thanks for the augmented explanation. I think I follow your process a bit better. It sounds like though that you are discounting the impact of the lighting / exposure variation as significant vs. insignificant. That's pretty much where we are on separate perspectives, as you feel that the "clipping point" made things equitable ... yet, the images revealed unequal values (both sky and quarry).
Anyway, thanks for the explanation and the updates in your second article. I think the fact that your additional testing revealed the A7R II to have 2/3 stop LESS DR than the A7R introduces (at least somewhat) a component of a different result from your initial presentation. So, two diff tests, two diff outcomes ... which lends me to think that your first test contained more significance to the variation than you've given credence to.
Thus, which test should your readers being giving more credence to ... the one that renders a conclusion of the A7R II having better DR or the one that renders a conclusion that the A7R has better DR?...Show more →
Given the re-testing of the dynamic range results I would say that the A7Rii will give most people better results as it needed a lot less post processing to get good shadows. They're pretty damned close though in general.
What is interesting is that the Sony A7Rii red filter is very thick. This is why the Sony A7Rii doesn't look like it's clipping as easily in lightroom. It is in fact clipping at the same time in the green channel but the red channel doesn't clip until later. This can be confirmed looking at the ratio's of the green, red and blue channel in the Rawdigger screenshots..
Personally I think the A7R is going to be the bargain of the century as people start to sell it and it becomes a £600 to £800 second hand camera. The A7Rii is more refined for me and does seem to be getting better colour - like I remember from the A900 - this could be due to the thicker filters..
p.7 #18Sony A7Rii compared with Sony A7R and Canon 5DSr
timparkin wrote: justruss wroteYes, now I'm even more confused. Is the A7r2 really showing 2/3 stop less dynamic range than the A7r-- but better color tonality so it looks, in actual images, like more?
It's very borderline and seems to depend on quality. I've just run another three imatest series as I was a little dubious about that and have averaged the results to get the following.
average values of 10.9 stops for the Canon 5DS, 12.2 stops for the Sony A7R and 12.5 stops for the Sony A7Rii. Imatest was quite variable from run to run but these averages seems sensible when I compare the test with checking clipping in rawdigger and looking at the tones in the deep shadows.
That's interesting. Everyone claims the A7R2 loses it's DR advantage to the 5DS after ISO 400, but your tests show different. Seems like the DR advantage carrys on into the higher ISO as well.
p.7 #19Sony A7Rii compared with Sony A7R and Canon 5DSr
timparkin wrote:
What is interesting is that the Sony A7Rii red filter is very thick. This is why the Sony A7Rii doesn't look like it's clipping as easily in lightroom. It is in fact clipping at the same time in the green channel but the red channel doesn't clip until later. This can be confirmed looking at the ratio's of the green, red and blue channel in the Rawdigger screenshots..
Here are the 3x3 color matrices from the EXIF data of the A7r vs A7rII raw files:
A7r:
1326 -311 9
-22 1356 -310
62 -213 1175
A7rII:
1243 -205 -15
-31 1209 -154
47 -47 1024
For an explanation, see this link and the "A is the 3x3 or 4x3 color correction matrix" section. Another reference is the DxO-measured color matrix data - not yet available for the A7rII but in the past their CFA measurements have been very consistent for each of Sony's Exmor sensors. Might very well be that the A7rII is different. For their A7r measurements, see this and click the color response tab. A measure of each RGB primary's "purity"/denseness can be evaluated by looking at the diagonal from top left to bottom right; the lower the number, the less gain needed for that color channel and thus the most pure/dense.
p.7 #20Sony A7Rii compared with Sony A7R and Canon 5DSr
Thanks for the follow up, we're starting to build out a better picture of how this things operates.
I totally agree on the A7r being a bargain on the used market-- particularly for folks who work from a tripod, at shorter focal lengths, with mf lenses (or who can take their time to mf before shooting). They're already popping up in the $850 to $950 range pretty regularly, and if you can negate the AF and IBIS advantages (mf + tripod) that makes it a super compelling compared to the 3x price for a new A7r2.
I'm going to grab the mk2 as a primary camera... but as a UWA, slow-use cam I'll probably get the original A7r a little later at a bargain price.