Another dissapointing review of the 7D

[inTempus]

Here's another disappointing review of the 7D. This review includes sample pics shot from a Rebel (12mp), 7D (18mp) and 1DsMk3 (21mp).

They used more than one 7D as the results from the initial tests were so disappointing they thought perhaps they had a dud body. Unfortunately the 2nd and 3rd bodies performed the same.

The XSi out performed it in IQ at every turn... and of course the 1Ds3's images were far and away better.

The Canon EOS 7D Review « Darwin Wiggett

Depressing to say the least. If I were in the market today for a mid-level body and I didn't already have a bag full of Canon glass, I would go for the D300s.

 

[Overread]

Ouch!
the only thing that comes to my mind as a possible error in the test ( and I honestly don't know enogh about it to know if its a problem with this test or not) is that the RAW processing was, at some level incorrect. I recall that early 50D images were softer until the RAW processing (or maybe firmwire) was upgraded and then results improved overall and this could be what we are seeing here.

I would be very surprised if canon would allow a midrange body to be beaten so clearly by an entry level body - canon especailly since they are very aggressive with making sure that each level of their camera range has a distinct advantage over the lower teirs.

 

[MrLogic]

The DPReview guys love it, though. Perhaps to make up for their (overly negative) 50D review.

Full review:

Canon EOS 7D Review: 1. Introduction: Digital Photography Review

 

[Overread]

Heh reading the comments to the blog there is quite abit of variety in responces. One weakness I can see in teh test is that it relies on unprocessed RAWS only with no mention of processed RAWs. I mention this as whilst a RAW shot should have "no incamera processing" this is in actuality a false fact and some sharpening and noise reduction will be performed - so I wonder if the rebel series camera is not simply applying more of this incamera sharpening to its RAW shot than the 7D is (or perhaps less incamera noise reduction) ?

Ever since the 50D I have worried that canons move toward more MP would have led to softer images as noise became harder to control without having to use stronger noise reduction (And thus more image quality loss in the finer details).

The other thing that is worrying is the mentions of diffration starting a lot earlier - f13 is where I like to work with macro, I would hate that with more MP the diffraction effect would start sooner and thus reduce the possible smaller apertures I could use for macro work.

 

[inTempus]

I think the biggest problem is that other cameras with similar processing produce far better files.

As for sharpening, if shot in RAW there shouldn't be any sharpening applied. That's done by DPP after the RAW is downloaded. If it's disabled, there won't be any sharpening done.

The test is of the basic RAW files produced by the cameras. That's what you have to work with right out of camera. If you start off with a better file (more detail, better shadow detail, etc.) your end result should be better once processing is applied.

No matter how you slice it the RAW files from the 7D don't look much better than a G11 point and shoot.

 

[Overread]

As for sharpening, if shot in RAW there shouldn't be any sharpening applied. That's done by DPP after the RAW is downloaded. If it's disabled, there won't be any sharpening done.

Noise reduction should also not be applied but it is. IF they are willing to apply noise reduction then why not sharpening to. It something I would suspect more in the G11 though over the DSLRs (the sharpening).

The test is of the basic RAW files produced by the cameras. That's what you have to work with right out of camera. If you start off with a better file (more detail, better shadow detail, etc.) your end result should be better once processing is applied.

No matter how you slice it the RAW files from the 7D don't look much better than a G11 point and shoot.

I can't disagree against that save to say that some are saying that processed RAWs from the 7D are better looking than processed from the rebel series. The only other thing I have found is debates on teh best RAW processor to use - some are saying that canon DP is not the best to use and that better results are seen in the Lightroom 2.5 beta support for the camera.

 

[MrLogic]

This pretty much sums it up, IMO:

de Fresz Says:
November 12, 2009 at 8:23 AM You missed few quite important technical aspects in Your review, that are good to know, when writting review (and generaly for seroius amateur and pros).

First: pixel density (pixel size)
7D – 5.4 MP/cm2
450D – 3.7 MP/cm2
1Ds mk3 – 2.4 MP/cm2

Two things link to that – diffraction limit and lens sharpness.
For 7D diffraction kicks in about f/6, for 450D in about f/8. The more You close the lens past this point, the image is less sharp (unsharpness of detail is better visible).

All lens shortcomings in matter of resolution are also better visible with densier sensor. Probably there is very few lenses (even those with good reputation) that are able to assure enough resolution for such dense sensor as in 7D.


Second – because of this technical shortcomings it would make sense to compare 7D with other cameras when images from first are downsampled to their resolution (12 MPx in this case). The real 18 MPx of tis camera will be very hard to achive in real life (sharp prime in narrow range of appertures aprox. f/4-6.3, with short shutter). Anything else will be “overpixeled”.



Third – few other things have influence on detail sharpness – antalising filter, software profile of the camera, in-camera file processing (yes, even RAWs are processed by camera engine).
Finaly – IMHO decision to put 18 MPx on aps-c size sensor is driven by marketing, not common sense or real needs.
Regards

The Canon EOS 7D Review « Darwin Wiggett

 

[NateWagner]

Also though, if the DPP they were using didn't have the appropriate camera profile it would screw things up wouldn't it? I mean, that explains to me why the DPreview was so much better (they at least had the beta version of the updated raw processor)

 

[Overread]

Finaly – IMHO decision to put 18 MPx on aps-c size sensor is driven by marketing, not common sense or real needs.
Regards

I'm pretty sure there have been canon camera designers who have confirmed this very fact. That they even would like to push the ISO, dynamic range and such but that marketing says go for MP.
Hopefully the Nikon push for ISO will eventually lead Canon to follow - I think we are seeing the end of the MP war and the start of the ISO war - for which Nikon got the first shot at this time around.

 

[MrLogic]

Someone did a diffraction "test" here:

7D diffraction test: good through f/16: Canon EOS 7D / 50D - 10D Forum: Digital Photography Review

"Not a problem until after f/11, and even f/16 is still very usable." Hmmm... can't say I agree with that assessment.

 

[Overread]

Based on the linked diffraction test I would agree with the testers in that after f11 is where it starts showing up but that f16 is usable. Of course the blog review counteracts this assessment but then we have to decide if their review is complete enough. Certainly the results they have found are sounding a little strong on one side and whilst I'm not questioning their authenticity it might be that some difference in the processing software used (even with nothing done to the RAWs) might be responcible for this difference

 

[Derrel]

I at first thought that maybe Darwin had stopped the lens down too far, such as f/16 in the hay bales in the field test scene. But in the comments below, he stated that his personal copy of the Canon 45mm TS/E lens is best at f/5.6 to f/13 and that he used those apertures as well in his testing procedure. He also commented that the 7D did well on in-studio test shots, "like those done by dPreview", but that in distant and mid-distance scenes, like those that he tends to shoot, the 7D bodies, all three of them, looked mushy.

He used Live View for critical focusing...the tests, over and over, up-sampled and down-sampled, D300s versus 7D versus 1Ds Mark III versus Rebel versus D300s,ad nauseum--they all showed the same things. The comparisons between the Canon pocket cameras and the 7D were interesting.

From what I have read from highly technical types, diffraction seems to kick in on the 7D at around f/5, which seems reasonable, given that it has the smallest pixel size of any d-slr ever made, and the most-densely packed sensor as well; smaller apertures would likely show a LOSS of sharpness, so even f/5.6 would be affected by diffraction. I have also read that the 7D has some loss of color depth at base ISO, which is 100, and was optimized for higher ISO shooting. My D2x has a 12.3 MP sensor on 1.5x,and diffraction kicks in at f/5.6, so f/5 at 18MP seems perfectly in line with what I have read, namely that at f/5 (as in f/5.0) the 7D's image lose sharpness at smaller openings.

I personally think that perhaps the 7D's unusual sort of 4-color imager is causing some problems; Canon has some hard to explain secondary green-yellow color RGB +G-Y demosaicing of the RAW data off the sensor, which is as I understand it, how they are measuring the color of the light in their color-aware light metering. *Properly* converting the RAW sensor data using this RGB+G=Y system seems to be causing a maze artifacting pattern for the few RAW converters on the market capable of handling 7D RAW files.
I personally think the RAW conversion software for the 7D might not yet be optimized to deal with their weird new sensor's working method, and that better image quality will be possible in the future.

I wonder if maybe Darwin needed to have applied MUCH higher USM, like maybe 500% at perhaps somewhere between .22 and .30 pixels Radius at 0 Threshold to overcome the strong anti-aliasing filter of the 7D.

But the fact remains--with that high-density 5.4 Megapixel per Square Centimeter sensor density, the LENS performance that I have seen in several reviews does not look good when using the 17-55 f/2.8 L EF-S lens,and frankly it does not look very good on the 24-105-L zoom either. Looking very closely at full-sized samples shot with the 17-55, I see a LOT of chromatic aberration problems and a slightly 'fuzzy' quality at the edges of most frames shot with the 17-55 f/2.8. With the premium,single focal length telephotos the bird guys are using, the 7D looks quite good, which is to be expected from a premium lens in the $7,000 price range.

The review is what it is. It's a two-person, practical, field test, involving three different 7D bodies, a top FF Canon body, a Nikon D300s, and some Canon P&S captures,and lots of cross-comparing. On landscapes. Using a pretty new camera, that uses a rather odd demosaicing system that "most" RAW converter software is not handling very well. And the sensor's pixel density is clearly beginning to factor into lens quality, and even lens f/stop in use. This camera seems to be pushing the boundaries quite hard...

 

[Derrel]

Here is another EOS 7D review

Canon EOS 7D Review « The World According to Roland

What is probably most disappointing is how significantly better EOS 5D MArk II images look when re-sampled and shrunk down to the same size as 7D images.

Also, he has a diffraction test that is better than the one referenced above from a dPreview user who photographed some book bindings. While images are always subject to personal impressions and prejudices when the images are 'close' to one another, I personally think I see diffraction setting in quite early in this test on the teddy bear; to me, the f/2.8 shot looks actually a tiny bit better than the f/8 shot. The f/8 shot is on the right hand side,and the f/2.8 shot crop is on the let hand side: I see slightly,and I mean ever so slightly better contrast in the f/2.8 capture. http://rolandlim.files.wordpress.com/2009/11/7d-f8vf2-8.jpg?w=500&h=800

Roland's review also has a good DPP versus Lightroom Beta 3 comparison, showing just how critical the RAW converter is.

 

[IgsEMT]

WOW
well, b/n various things I've read on it, here, other forums and blogs, it is a camera of "flavor". Some people LOVE IT, others - not really.

 

[musicaleCA]

I have two, no regrets.

"We processed the photos in Canon’s DPP exactly the same with no sharpening or noise reduction."

Yes, and that'd have something to do with it. In the technical notes they noted how the AA filter is probably one of the strongest ever on a Canon body. Well, yeah, it really is. This can be corrected later with sharpening, and in my own processing I've noticed that I've had to apply quite a bit more sharpening to my 7D images over my 450D images. (Say, 75 on the sharpening slider instead of to 45.)

The weird colour rendition changes between the P&S and the 7D are a case of "duh, did you calibrate it?" Until you calibrate the camera body by creating a profile for it using a munsell colour card and something like Adobe's DNG Profile Editor, you're going to have some odd colour shifts happening. The software in-camera likely corrects for much of this on in-camera JPEGs; RAW files need to be fixed yourself.

I think Royce's comments at the end of that blog post are the most informative there, in terms of explaining WHY there are these differences. They look different, yes, but that means you need to process differently.

If what we're looking for here is apples-to-apples comparisons, I must agree, shoot JPEG. That may solve some of these oddities right off the bat. RAW is not really a hard-set image format; it's interpreted, and that interpretation can vary between programs, and how one should interpret the data to get the best image out of the RAW file will change from camera to camera. Remember everyone, Adobe is still in beta when it comes to 7D support. Profiling is still iffy (though AFAIK it works; at least well enough for me). When they get LR3 and the next version of ACR out, they might very well have figured out what to do to a 7D's RAW files to get the most out of them. $10 says there's some crafty engineer working on the problem right now.