APS-C Image Quality (7d vs nex7)

[akiskev]

Excellent pics KeithR!!
The images I uploaded were 100% crops though. Nobody said that 7d is not a very good camera iq wise. It surely is one of the best aps-c cameras anyone can buy at this moment.

I just mentioned that Sony's sensor was WAY BETTER in pixel peeping situations. To pixel-peep you need a 100% image. Not a scaled down one

[torger]

However what we do know is that the 7D sensor has at ISO100 pattern noise issues in dark shadows and considerably less dynamic range than compared to the best current APS-C sensors.

Not strictly true - what we know is that the 7D needs the user to apply some intelligence and discernment in his choice of conversion software. I've cranked low ISO 7D shots up by 4 or more stops without any problem in the shadows.

Well it kind of is strictly true ;-). It is correct that you can reduce hardware pattern noise with software noise reduction, and Canon's own software does quite well. But if you want to use third party software to process your RAWs? I like user-freedom in post-processing, and therefore I certainly prefer well-behaving hardware rather than hardware that requires fixups in software.

I do agree though that one should not exaggerate this problem, although I do see it now and then when I work with my images it has so far not given me any practical problem in prints. I actually have had more problems with reduced size web images, when the pattern can occasionally become more visible when reducing to certain sizes, then also in brighter parts of an image, such as a dull uniform sky.

For example this test image: http://torger.dyndns.org/temp/IMG_9045.jpg
has line noise problems in the sky, not so very obvious but when stretching around with the picture it does appear a slight vertical "corduroy" pattern. It is too slight to be visible in a print (and some of you will probably not see it on screen either, requires good screen and perfectionist eye), but I could do without it anyway. Might be more visible in this one:
http://torger.dyndns.org/temp/IMG_8160.jpg (again, the sky).

The image was converted with RawTherapee, without using the line noise filter (which I only use in more severe situations, since it reduces detail too, sometimes I do a separate filter for the sky since there is generally no sharp details there to worry about).

[AJ]

The Sony NEX photo looks oversharpened.

Something's wrong with the 7D photo.  My 7D gives me much sharper results at 100%

[Mt Spokane Photography]

My experience with the 7d is that a careful microfocus adjustment makes a big difference, the high pixel density requires near perfect focus.  There is a learning curve.

Sony cameras are great in jpeg, they use a ton of NR and saturation so that images look wonderful on the web.  Compare raw output, and the tables reverse, give images the same nr and developing treatment, and the noise from the sony sensors suddenly becomes very apparent.

This makes Sony a good choice for a person who justs wants to point and shoot, while those who want to process their own images to get best possible quality will prefer the Canon sensor.

[dr croubie]

Just for an example of lens and 'sharpness' processing differences, here's some shots I was taking to compare a few weeks ago.
I was comparing the Canon EFs 15-85 @ 35mm to the Samyang 35/1.4.
Here's the shots I took under identical conditions on my 7D, firstly I RAW-processed everything with sharpness 3 using DPP. Basically, the samyang kicks butt. (shutter speeds were at 1/200s at f/8, so no motion blur, iso100, 100% crops from the far-right vertically-middle of frame)
Today, i've gone back to the RAW files, and processed the exact same photos again, with the samyang sharpness setting on 0, and the 15-85 sharpness on 5 (i tried up to 7, but it looked too 'over-sharp').
The differences are so much less pronounced than when the sharpness setting are the same. Basically, if this small difference in PP makes such a difference in the output photo, comparing across sensor-types is a lot harder than just looking at sample pics...

[willrobb]

Excellent pics KeithR!!
The images I uploaded were 100% crops though. Nobody said that 7d is not a very good camera iq wise. It surely is one of the best aps-c cameras anyone can buy at this moment.

I just mentioned that Sony's sensor was WAY BETTER in pixel peeping situations. To pixel-peep you need a 100% image. Not a scaled down one

I agree 100%. I think the 7D is a great camera, lots of people get great results with it, I was very happy with mine, but I sold it last weekend as I felt the IQ at high ISO wasn't suited for what I do. Looking at the test results it looks as if the Sony sensor is pretty damn good, I welcome that as it means canon will have to raise it's game a bit which helps us. However, if they made a 7DII sensor too good it would eat into canon FF sensor sales, so I don't expect a massive increase in sensor quality in the 7DII, but I am sure it will be pretty damn good and if it keeps a lot of the great 7D functions in terms of AF, FPS, built in wireless controller etc, the 7DmkII is going to be first choice camera for many people as well as an awesome 2nd camera for a lot as well.

As I said, I loved my 7D. For anything in print or online publication I was extremely happy, but for private wedding and portrait hires where I always give clients some large A3 size prints, I just found my 5DmkII to way outperform on IQ across the board so I decided to sell the 7D for another 5DmkII. I will miss the 7D, but for the price I paidfor it and what it did for me, it really was a great buy and I am happy to have used it.

[dtaylor]

I tried IR's studio scene at ISO 100. After downloading both samples I:

* Re-sized the 7D sample to the NEX-7 sample size using bicubic sharper.

* Applied Smart Sharpen at 100% / 0.7px.

After that the IQ advantage of the NEX-7 was much smaller then what appears in your samples. It's a 24 MP sensor so I would expect some advantage.

I also tried sharpening the NEX-7 sample a bit and then trying to "catch up" again with the 7D sample. The differences were still small. Basically the largest difference came down to more artifacts on the 7D sample, but on either the artifacts would only be visible while pixel peeping.

That said I expect a 24 MP sensor might yield a better rendition of foliage in landscape shots for larger prints. Not dramatically larger, but another 4-6" for a print which will be critically reviewed. (I consider the 7D to be a 24-30" landscape camera as is. Larger with greater viewing distance, less critical viewers, or less demanding subjects.)

We are reaching the point of diminishing returns in small format sensor resolution, where top notch glass, expert technique, challenging subject matter, and big prints are required to see any difference at all. That said I'm all for more megapixels as long as other characteristics do not suffer as large landscape prints are one of my favorite subjects.

This isn't enough to make me consider jumping ship. But I could end up with a NEX-7 any way because I would like to add a mirrorless system. I would like to see if Fuji releases an interchangeable lens version of the X-100 though. For that matter I would like to see any of the mirrorless systems embrace small, fast, high quality primes. They are too zoom heavy right now, and big slow zooms do not play to the strengths of a compact, mirrorless system.

[papa-razzi]

I have sharpness on my 7D set to 5 or 6, shoot RAW and use L glass. It's tack sharp.

Dumb question, but I'll ask anyway.  Are you talking about the sharpness setting you apply in your post processing, or is there a sharpness setting for the JPEG conversion in-camera?  Is that part of a picture-style?

[dr croubie]

I have sharpness on my 7D set to 5 or 6, shoot RAW and use L glass. It's tack sharp.

Dumb question, but I'll ask anyway.  Are you talking about the sharpness setting you apply in your post processing, or is there a sharpness setting for the JPEG conversion in-camera?  Is that part of a picture-style?

Sort of both.
Go to picture-style settings, and (on my 7D at least), pressing the 'info' button brings up all the options for each style. In the list, they're listed as "standard 3,0,0,0" or something. The first number, 3, is the sharpness.
If you shoot JPG, that's the setting the camera uses for conversion. (I presume it uses the same algorithms as when raw-converting using DPP, but just saves as 99% quality jpg, not 100%).
If you shoot RAW, whatever picture style you choose in the camera is set as the *default* when you raw-convert in DPP. You can shoose anything else you want from the raw-file, i've made multiple copies of a shot from the same RAW file, and they all turned out wildly different. But setting the style in-camera is good if you batch-process and don't want to play around too much.
See my examples a few posts back, they were from RAW files, i processed them with sharpness 3 (because that was the default), then processed them again with sharpness 0 and 5. All were using the 'standard' picture style.

[torger]

[However, if they made a 7DII sensor too good it would eat into canon FF sensor sales, so I don't expect a massive increase in sensor quality in the 7DII

I hope / think that regardless how good quality the sensor is it would not really cut into FF sales. The 2.5 times larger sensor area in FF will make some difference by itself, higher resolution, and you have other depth of field properties and the field of view you get in the nice focal lengths 24, 50, 85, 70-200 glass is more versatile for many. In sports and wildlife APS-C will probably become more popular though, now when APS-H seems to be phased out.

Canon must keep up with the competition too, concerning sensor quality it may be even more important to be competitive on entry level APS-C cameras than on full-frame. Enthusiast amateurs are often technically interested and wants a sensor that performs well in lab tests, even if a lesser sensor would be perfectly good enough for their needs. It would not be good for Canon if they get the reputation to have the less good sensors compared to Sony and Nikon, it would hurt entry level DSLR sales which is an important market.

[briansquibb]

See my examples a few posts back, they were from RAW files, i processed them with sharpness 3 (because that was the default), then processed them again with sharpness 0 and 5. All were using the 'standard' picture style.

If you are shooting in RAW then I would recommend sharpness as 7 as this is what the image on the LCD will show. If you shoot with 3 you will think that your shot is coming out soft or blurred. 7 shows much closer to the pp image

[dilbert]

However, if they made a 7DII sensor too good it would eat into canon FF sensor sales, so I don't expect a massive increase in sensor quality in the 7DII

I disagree with that statement completely.

There are reasons for wanting to use FF cameras that have nothing to do with the difference in sensor IQ.

Being about to mount a 21mm Zeiss lens on your camera and get a 21mm lens and not a 32mm lens is among them.

[moreorless]

I agree 100%. I think the 7D is a great camera, lots of people get great results with it, I was very happy with mine, but I sold it last weekend as I felt the IQ at high ISO wasn't suited for what I do. Looking at the test results it looks as if the Sony sensor is pretty damn good,

Looking at the results I'v seen the Sony seems significantly worse than the 7D at high ISO and noticbley worse all the way down to 200.

[KeithR]

Looking at the results I'v seen the Sony seems significantly worse than the 7D at high ISO and noticbley worse all the way down to 200.

The A77?

Yep, it's terrible - eye-poppingly worse than the 7D at any ISO.

[KeithR]

I disagree with that statement completely.

There are reasons for wanting to use FF cameras that have nothing to do with the difference in sensor IQ.

I'll take that a stage futrther: IQ "superiority" is the least valid reason of all to go FF.

This is something which has been demonstrated time and time again on DPR, Photo.net and other sites: the end result from a modern crop sensor can easily match the end result from a given FF camera to the extent that it's impossible to tell which image came from which camera.