p.1 #1 · 7d vs 5D mkII vs RAW development variation
Following discussions in a couple of threads I decided to do a couple of tests to compare the difference between 7D and 5D MkII compared to the variation that you get when you use different RAW developers .
This is not a test of various RAW developers. The hypothesis being tested is that the process RAW development has a greater impact on image quality than sensor size has.
I found the results interesting so I thought I'd share them.
Set up, subject
I used the 70-200 f/2.8 MkII lens as it has very high optical quality across the zoom range and it allowed me to compensate for the 1.6x crop factor of the 7D. The lens was mounted on a heavy and very steady video tripod (Manfrotto 535 CF + 503HDV head). A cable release and mirror lock down was used:
http://peltarion.eu/img/5D7D/0_setup.jpg
For subject I selected a building that I can see from my balcony. The distance is 190m. Apart from convenience the subject was chosen because it had a nice dynamic range with both highlights and shadows, detailed texture and sharp straight lines and edges. The sky was overcast so the light was very flat and even.
This is what the full scene looked like (with the crop subject marked):
http://peltarion.eu/img/5D7D/0_full_scene.jpg
The tests were done at 200mm f/8 for the 5D MkII and at 125mm f/8 for the 7D. Manual mode, with two different ISOs (100 and 3200). The shutter speed was 1/50s and 1/1600s). Single point auto focus with 10x magnified live view verification was used to guarantee that the subject was not out of focus. All was shot RAW.
RAW processing
I processed the RAW files in Adobe Lightroom 3 Beta 2 (LRb3), Canon Digital Photo Professional (DPP) and DxO Optics Pro 6 (DxO). Default settings were used in all instances with no manual modifications. The files were saved to JPG highest quality and then cropped in Photoshop. The difference in size of the crops is due to the 22 Mpixel of the 5D MkII vs the 18 Mpixel of the 7D. So keep in mind that these are 100% crops and we're comparing by pixel and not how good it will look resized for screen use or printed on equally sized media.
The reason why default settings were used are very simple: I'm testing the performance of FF vs crop sensor size, not RAW development software. The choice of three different procedures give us an idea of how sensitive the quality is by changes in that variable. Default settings were chosen as the operating point for the software used as it is guaranteed not to be biased by me manually changing any settings. This is not something unique - digital photo review uses default settings in DPP as well. The Digital Picture (the king of ISO chart tests) uses default settings with only the sharpness turned down.
Results
http://peltarion.eu/img/5D7D/1_7D_iso100.jpg
http://peltarion.eu/img/5D7D/2_5D_iso100.jpg
http://peltarion.eu/img/5D7D/4_7d_iso3200.jpg
http://peltarion.eu/img/5D7D/3_5D_iso3200.jpg
Follow up tests
I did some prints on my Canon IPR5200R printer which is a high end consumer inkjet. The prints were on good quality semi matte paper. The difference in color between the 5D and the 7D is obvious on the prints but not much more than that. If you look really closely at some details you can see that the 5D has slightly higher contrast - but you really need a to look very closely. I could not spot any difference in sharpness even when using a magnifying glass. I'll do some printouts later on very high quality glossy paper and see if it makes a difference, but I doubt it will.
The difference between the RAW processors on the other hand are plain to see on paper in the ISO 3200 case where both DPP and DxO kill off all detail with too heavy noise reduction.
And now, please, before commenting on the use of default settings in the RAW developers, please keep in mind that this is not a test of various RAW developers. The hypothesis being tested is that the process RAW development has a greater impact on image quality than sensor size has.
Edit: Added the bold part, changed title etc. So many have completely missed the point of this test that I needed to clarify things.
p.1 #3 · 7d vs 5D mkII vs RAW development variation
All RAW processors have noise reduction enabled by default. Lightroom does it well for instance so it would be unfair if I had started to tweak DPP to make it look good. This is a comparison of the default settings.
p.1 #5 · 7d vs 5D mkII vs RAW development variation
The problem I have with these tests is the use of the default settings. If you really want to test various programs, I think it would make more sense to go into each program and process the file the best you can, then see what you get.
Basically, I don't want to see what the each program's defaults do, I would like to see what each program is capable of.
Also, one thing you are not catching here is how bad DPP is with hightlight clipping or significantly overexposed non-clipped highlights. That is a big deal for me, and I've always found DPP to suck in that area, and ACR much better. When I switched from DPP to LR I gained, in effect, about 1 stop of overexposure latitude.
p.1 #6 · 7d vs 5D mkII vs RAW development variation
chez wrote:
Nice job of the test, but who would use any of these with default setting?
I would be very surprised if the vast majority RAW files are not developed with default settings.
Typical situation for me: Say I shoot 1000 shots on an occasion. I come home and load them up in Lightroom. I start first by filtering off the rubbish (photos that have neither technical nor artistic merits). I also remove all near duplicates of which there are quite a few. Say I'm left with 500 photos. So I start grading them and I end up with 20 images that I really like. Those are the ones that I manually develop and tweak. The remaining 490 are left with the default settings and any time in the future that I look at them they will still have the default settings. So it's a 1:49 ratio of custom/default settings.
There is one thing more and I should have clarified that. This is not meant to be primarily a test of RAW processors. This test has its origins in people dismissing the 7D as 'soft' and claiming that there is a massive difference between the 5D MkII IQ and 7D IQ. The main point here was to show that RAW processing makes much more difference than the FF vs crop quality.
p.1 #7 · 7d vs 5D mkII vs RAW development variation
Thanks for sharing your work - and interesting results. I still wonder why so many are trying to decide between the latest 1.6 crop camera (7D) or the latest FF (5DMk2). I see these cameras as complimenting each other - not direct competitors.
For me its trying to decide between the 7D or 1DMk4 - that will replace my 1DMk2 (which compliments my FF body).
Dave
Mar 28, 2010 at 10:00 AM
brainiac Offline [X]
p.1 #8 · 7d vs 5D mkII vs RAW development variation
Thanks denoir, but really, a test of these programmes' default settings is probably of questionable interest to people here. For a start, the default settings are what you set them to. Your defaults are not my defaults. Secondly, the kind of people interested in spending several thousands on a camera body, and plenty more on lenses, are not going to be casual about raw development . We may apply default or specific settings en masse, but that doesn't mean they are likely to be the defaults. Your test would look very different if you switched off NR, as most people seem to, in DPP, and set sharpening appropriately, again, as most people seem to. Really, I just don't see the point of this. Maybe you should have used a kit zoom, since that's also what the "vast majority" use.
I also wonder whether f8 is the best aperture to test 7D resolution potential. F5.6 would be safer from a diffraction point of view.
p.1 #9 · 7d vs 5D mkII vs RAW development variation
brainiac wrote:
Thanks denoir, but really, a test of these programmes' default settings is probably of questionable interest to people here. For a start, the default settings are what you set them to. Your defaults are not my defaults.
These are Adobe, Canon and DxO defaults. The point is exactly that these are neither my nor your defaults.
Secondly, the kind of people interested in spending several thousands on a camera body, and plenty more on lenses, are not going to be casual about raw development . We may apply default or specific settings en masse, but that doesn't mean they are likely to be the defaults.
I seriously doubt it. The defaults are set to work with as a wide choice of subjects and lenses. If you deviate from them and then batch process large number of files a lot of them will be crap. You can either tweak the details of a RAW file or batch process them (i.e. keep them as they are and view them with some default settings). There is no middle ground where you half-tweak a file. Well, at least no serious photographer would do it. That means that when you look through the great majority of unprocessed pictures, you will be looking at them with whatever blanket settings you use. Those settings can't be much different from the defaults set by the software. If they are then either the programmers of the software were incompetent or you are.
Your test would look very different if you switched off NR, as most people seem to, in DPP, and set sharpening appropriately, again, as most people seem to.
My test would look much worse for DPP if I had turned NR off. The NR in DPP is not unreasonably strong, it's just bad.
Really, I just don't see the point of this. Maybe you should have used a kit zoom, since that's also what the "vast majority" use.
No, clearly you have missed the point of it.
The point is actually quite simple - variation in quality due to RAW development (within the boundary conditions of the default settings of various RAW-developers) is greater than the variation in quality due to difference in sensor size. The qualifier within the parentheses is to signify that no extreme RAW tweaking is involved.
The choice of the default values is in order to have a fixed repeatable point of reference. It's called "controlling the variables" of the experiment.It is exactly the same way for instance Bryan at the digital picture does when he processes the images - he has one RAW processing template that he applies to all charts. If I started to subjectively tweak the RAW files the whole test would be completely pointless. It's for the same reason why I use the exact same aperture, shutter speed, glass, tripod etc etc
The choice of the default settings as opposed to any other arbitrary choice should be obvious as well. It guarantees that the test was not biased - at least not by me.
I also wonder whether f8 is the best aperture to test 7D resolution potential. F5.6 would be safer from a diffraction point of view.
f/8 and f/5.6 are indistinguishable as far as sharpness gos with a 7D using a 70-200 MkII. It was not chosen at random but to keep a couple of trees in front of the building in focus.
p.1 #10 · 7d vs 5D mkII vs RAW development variation
Thanks for the test Denoir, and I agree with your approach. When processing a large amount of images I also want a default batch setting to start my selection with. After pulling out the images I wish to PP I can than do the tweaks on them.
Paul
p.1 #11 · 7d vs 5D mkII vs RAW development variation
dbehrens wrote:
I still wonder why so many are trying to decide between the latest 1.6 crop camera (7D) or the latest FF (5DMk2). I see these cameras as complimenting each other - not direct competitors.
I completely agree - I have both.I also think though that not everyone finds it practical or cost efficient to have two cameras. So after they pick one they try to assure themselves by assuring everyone else that it's the best possible choice. It's similar to the whole Canon vs Nikon debate that always seems to come up on a regular basis.
p.1 #12 · 7d vs 5D mkII vs RAW development variation
Lightroom 3 beta 2 (at least my installation) has luminance NR turned off by default. Hardly a fair comparison when the other two RAW processors are clearly applying luminance NR...
I have to agree with some of the other posters that I would find a comparison of "best effort" NR between the 3 far more interesting.
p.1 #13 · 7d vs 5D mkII vs RAW development variation
denoir wrote:
Following discussions in a couple of threads I decided to do a couple of tests to compare different RAW processors and the results one can get from a 7D and a 5D MkII. I found the results interesting so I thought I'd share them.
the tricky thing is you used default settings which is not at all the best that one could get and each company has their own viewpoint as to what best even is.
you really need to play around with all the sliders
granted that is a tricky job it's hard to find the (personal) ideal slider set for each program
it's a very long process and tricky to really compare converters
not sure if LR3 beta 2 fixed the 7D but beta 1 had lots of mazing artificats since it didn't deal with the 4color CFA array of the 7D
not sure either if LR3beta2 has any luma NR at all no matter how you set things, i thought they were leaving that out until final, but maybe beta 2 put some in
p.1 #14 · 7d vs 5D mkII vs RAW development variation
denoir wrote:
These are Adobe, Canon and DxO defaults. The point is exactly that these are neither my nor your defaults.
which doesn't make it a good point
I seriously doubt it. The defaults are set to work with as a wide choice of subjects and lenses. If you deviate from them and then batch process large number of files a lot of them will be crap. You can either tweak the details of a RAW file or batch process them (i.e. keep them as they are and view them with some default settings). There is no middle ground where you half-tweak a file. Well, at least no serious photographer would do it. That means that when you look through the great majority of unprocessed pictures, you will be looking at them with whatever blanket settings you use. Those settings can't be much different from the defaults set by the software. If they are then either the programmers of the software were incompetent or you are. ...Show more →
I don't doubt it at all. The defaults for DPP, IMO, are almost always too smeary and artificial looking. They are the low noise company want to stress low noise and they are set for the masses, the same ones, who complain that a grainy film shows grain on blu-ray, perhaps.
The defaults ACR sets are not the one I use a general base starting point either.
I wasn't entirely happy with the C1 defaults either.
Never used DxO, but they seem to really, really go for the smear affect at high ISO as default, I would never, ever use the way they have it set for ISO1600 even as a starting point.
I think it pretty absurd to say if you dare deviate from the defaults and were to batch a huge bunch that they would mostly look like crap! Heck your DxO photo makes it clear that just about any photo will look like crap at high ISO if you user their DEFAULTS.
And ACR doesn't even tune by lens at all, so default means zero vignetting control and zero CA control of any sort! Granted those lens specific defaults from teh programs that are set to handle specific lenses and if you enable those settings, such as DPP and DxO, are likely to be about as well as you can do, but those settings weren't even under investigation in this particular test.
And surely the ACR default color mapping are not the ideal and used to be terrible in fact for many years.
My test would look much worse for DPP if I had turned NR off. The NR in DPP is not unreasonably strong, it's just bad.
IMO it would look better, by far for the ISO100 shot had you just turned it to zero where it needs to be. And at higher ISO I'd turn it down a notch or two as well. That said I don't like how it works too much so I will agree that it is just bad.
No, clearly you have missed the point of it.
The point is actually quite simple - variation in quality due to RAW development (within the boundary conditions of the default settings of various RAW-developers) is greater than the variation in quality due to difference in sensor size. The qualifier within the parentheses is to signify that no extreme RAW tweaking is involved.
who uses things like that? everyone would be randomly saying their 20D or the 5D2 did better then
The choice of the default values is in order to have a fixed repeatable point of reference. It's called "controlling the variables" of the experiment.
yeah so you controlled them at that level, but so what?
the fact is that on the underlying levels the variables have not been controlled at all, since it's like DPP calls 1ft to be 2m and ACR calls 1ft to be 3m and DxO calls 1ft to be 0.05m.
you need to try to do isolation detailed tests at ISO100 and see if you like the forms of detail extraction and compare artifacting under a wide variety of inputs and then try to isolate more on NR at higher ISO and see how it acts and how it fights off things to a certain level with what sort of look and detail retention
and then maybe try to level both at once and compare and then try a best possible, these latter get to be really tricky to figure out
anyway if you just want to see what happens on default, sure a good test then and not lacking in worth, but it doesn't tell one which to use if they are going to get into the detail; the best defaults you can come up with could potentially radically change how you'd rank the converters and the ideal per picture settings might perhaps change it again, although it is a bit less likely, it might mix it up a bit more
Mar 28, 2010 at 01:10 PM
brainiac Offline [X]
p.1 #15 · 7d vs 5D mkII vs RAW development variation
The choice of the default values is in order to have a fixed repeatable point of reference. It's called "controlling the variables" of the experiment.
It would be better to control the variables in such a way that the most similar outcomes arose, instead of deliberately setting the raw convertors to produce widely differing outcomes. You might as well have set saturation to max and contrast to min in one convertor, and the converse in another. All this test shows us is how the manufacturers configure their defaults. Not very useful information, even to those who don't already know it.
p.1 #16 · 7d vs 5D mkII vs RAW development variation
Am I so bad at communicating what this test is about? I have edited the original post where I have tried in as a simple way as possible explain the things I've been saying in the last posts.
* This is not a test of various RAW developers. The hypothesis being tested is that the process RAW development has a greater impact on image quality than sensor size has.
*The reason why default settings were used are very simple: I'm testing the performance of FF vs crop sensor size, not RAW development software. The choice of three different procedures give us an idea of how sensitive the quality is by changes in that variable. Default settings were chosen as the operating point for the software used as it is guaranteed not to be biased by me manually changing any settings.
the best defaults you can come up with could potentially radically change how you'd rank the converters and the ideal per picture settings might perhaps change it again
Yes! Almost there. Not quite but almost. It could potentially change how you rank the cameras which was the hypothesis for this whole test. The question that this test tries to answer is: If I put 5D2 and 7D images through different RAW developers will the results be most impacted by the choice of camera or how the software develops the RAW? The point is that with default settings which are both conservative in nature and not biased by me we see greater variation due to differences in RAW processing than due to sensor size.
For me, having both a 7D and 5D MkII this is test forms a preliminary answer to if I should worry about inferior IQ when using the 7D or not. The print tests were especially important for me. The very preliminary answer of this first test seems to indicate that no, I don't have to worry about the 7D. Where it is suitable for its speed and its use of lenses I should not hesitate to use it. How much I spend tweaking the RAWs will have a much more significant effect on IQ than the difference between the two cameras. The 7D will do quite nicely for wildlife photography and action shots while I'll use the 5DMkII for portraits and landscapes or any situation where I want the particular shallow DOF/wider FOV look that it can give me.
p.1 #17 · 7d vs 5D mkII vs RAW development variation
brainiac wrote:
It would be better to control the variables in such a way that the most similar outcomes arose, instead of deliberately setting the raw convertors to produce widely differing outcomes. You might as well have set saturation to max and contrast to min in one convertor, and the converse in another. All this test shows us is how the manufacturers configure their defaults. Not very useful information, even to those who don't already know it.
No, it tells you the variation between the 7D and the 5DMkII compared to the variation of the different software and their default settings.
You may not find it useful, but they I'm guessing you are not finding dpreview tests or the digital picture tests very interesting either. They use a standardized set of RAW processing templates that they don't change from one camera or lens to another. They do not try to get the best image quality possible by tweaking the RAWs in each case but use the same settings for all. TDP uses a slight modification of the default DPP settings. DPreview uses the default settings for DPP.
he supplied software, DPP, provides the best 'out of box' detail (visibly better than the JPEG) using the default settings, with Adobe ACR noticeably softer. However the difference between ACR and DPP is mostly down to sharpening (ACR by default uses very light sharpening). The bottom example below shows this (it's way over-sharpened in ACR, but it illustrates the point).
...
Digital Photo Professional -> TIFF (Default settings, manual WB)
ISO 100 studio scene 100% crops
...
Adobe ACR 5.6 RAW ->JPEG (Camera Standard Profile, manual WB)
ISO 100 studio scene 100% crops
How dare they doing such useless reviews.. And the digital picture? You know with all those great ISO charts that people use when choosing their glass. Default DPP settings with only the sharpness slider turned down a notch. Truly useless, right?
Anyway, I'm sorry you found the test useless but I can see that at least some people did not. You will however have a difficult time finding any proper reviews of photo gear if you insist that advanced Jedi camera masters such as yourself never could consider to look at images processed in that primitive way.
p.1 #18 · 7d vs 5D mkII vs RAW development variation
denoir wrote:
For me, having both a 7D and 5D MkII this is test forms a preliminary answer to if I should worry about inferior IQ when using the 7D or not. The print tests were especially important for me. The very preliminary answer of this first test seems to indicate that no, I don't have to worry about the 7D. Where it is suitable for its speed and its use of lenses I should not hesitate to use it. How much I spend tweaking the RAWs will have a much more significant effect on IQ than the difference between the two cameras. The 7D will do quite nicely for wildlife photography and action shots while I'll use the 5DMkII for portraits and landscapes or any situation where I want the particular shallow DOF/wider FOV look that it can give me. ...Show more →
Sry for my newbie question? I have a 7D now but anything in 5D II better than 7D except the shallow DOF and wider FOV u did talk about?
Because i feel fine with my 7D and i want to know if there is anything better than my cam if i dont need shallow DOF and wider view
p.1 #19 · 7d vs 5D mkII vs RAW development variation
trung hoang wrote:
Sry for my newbie question? I have a 7D now but anything in 5D II better than 7D except the shallow DOF and wider FOV u did talk about?
Because i feel fine with my 7D and i want to know if there is anything better than my cam if i dont need shallow DOF and wider view
It has somewhat less noise with high ISO, and the images when looking at 100% will in some cases be a tad sharper. Beyond that, no. The 7D is the better camera as it is much faster (8 fps), has much better auto focus and metering systems, better live view and video controls (if you care about that), better button layout, better build quality etc etc The 5D2 has a better sensor.
So no, if you don't care about the DOF/FOV differences the 7D is the better choice.
But, having said that, I'm afraid there is something better than your cam - it's the 1D MkIV which is better in all the points that I listed for the 7D.
p.1 #20 · 7d vs 5D mkII vs RAW development variation
denoir wrote:
It has somewhat less noise with high ISO, and the images when looking at 100% will in some cases be a tad sharper. Beyond that, no. The 7D is the better camera as it is much faster (8 fps), has much better auto focus and metering systems, better live view and video controls (if you care about that), better button layout, better build quality etc etc The 5D2 has a better sensor.
So no, if you don't care about the DOF/FOV differences the 7D is the better choice.
But, having said that, I'm afraid there is something better than your cam - it's the 1D MkIV which is better in all the points that I listed for the 7D....Show more →
Thanks for your answer. Now i think i stay with my 7D. 1D IV is rly nice but so far away from my budget, and its too heavy for me to walk around or travel
I just wanna to ask about 5D because i shoot mainly portraits and all says 5D II is perfect for that. So i want to know how much my cam ist worse
