Canon EOS 7D tested, retested and reretested

[angel_bocanegra]

<p>What kind of photographer buys 18k worth of lenses to use them on bodies that are disappointing(to Julian) and lower end models? one-the <strong><em>luxuriously wealthy </em> </strong> and two- someone full of <strong>BS</strong> . Julian, you already ruled out one of those two. You have Zero credibility with a big Z, just like this Widgett joker(comparing 7D to a g11 and saying the g11 is better, WTF?). You two probably have Nikon affiliation of some kind and what you are doing is wrong and shameful, yet funny(your lengthy comments gave some of my most memorable laughs from reading post here on photo.net). I already posted sample pictures(previous posts) comparing for sharpness and noise from a 7D and 5dii. What will you post next?</p>

[images_in_light_north_west]

<p>One thing, is the 7D has got emotions going on both sides, its had good reviews and not so good, looks like some sample variation going on. I do believe that Canon have so many DSLR's now that they are having trouble keeping the lower end cameras from being as good at IQ as the top end cameras, and by adding more MP or a stronger filter, may be there way of avoiding this, just speculation of course. I wanted the 7D to be a very good camera to go with my 5D2, I believe Canon could have done this, but chose not to, they want me to buy a 1D4, well thats not going to happen (dont have that kind of cash) so I will have to explore other options. I am not trying to bash Canon, they are after all, a big company that must show profits to the share holders.<br>

Ross</p>

[keith reeder]

<blockquote>

<p>I wanted the 7D to be a very good camera to go with my 5D2, I believe Canon could have done this, but chose not to</p>

</blockquote>

<p>I really haven't got a clue what that's supposed to mean - by what possible measure can the 7D be described as anything other than excellent? It is clearly <em>and without question</em> the best APS-C camera available today.<br>

And that's not emotion speaking - it's an objective reaction to the evidence.</p>

[arie_vandervelden1]

<p>Okay time for a reality check. Here's a photo I took with my 7D last night.<br>

<img src="http://www.arievandervelden.com/Calgary.jpg" alt="" width="800" height="533" /><br>

Canon 7D at 400 iso, Canon 17-55/2.8 IS at f/4. Shutterspeed about 1/8 s or so.</p>

<p>Now, let's pixel-peep, shall we? Here's a collection of 100% crops:<br>

<img src="http://www.arievandervelden.com/100pctCrops.jpg" alt="" width="600" height="400" /></p>

<p>Bonus points if you can identify which part of the photo these crops came from.</p>

<p>Some observations:<br>

- <strong>The detail is there</strong> . I see lines/dots that are 1.5 pixels wide. This is the best we can do with Bayer interpolation, and on par with any other DSLR I've owned. So, the AA filter is not super aggressive or anything like that.<br>

- Is there more information here than in a file from my 10 Mpix XTi? Yes, definitely, conclusively, without the slightest doubt. Is there more information here than in a file from a 12 Mpix XSi? I say, probably. Is there more information here than in a file from a 15 Mpix 50D? I say, who cares, there is no significant difference between 15 Mpix and 18 Mpix. We're talking sqrt(18/15) = 10% more pixels in each dimension.<br>

- I examined in-camera jpeg (standard picture style), RAW processed with DPP, and RAW converted to DNG and processed with ACR/PS. The DPP and in-camera jpeg were indistinguishable. The ACR result was slightly sharper and slightly noisier, but we're splitting hairs.<br>

- <strong>There is noise</strong> . At 400 iso the amount of noise is about the same as my 400D. At 100 iso (not shown) I think my 400D is slightly cleaner, on a pixel-by-pixel basis. At 1600 iso the 7D blows away the 400D.<br>

<strong>- The quality aspects of the noise are interesting. </strong> On my 400D, the shadows tend to be noisy, with bright specks (short-wavelength luminance noise) and blotches of chroma noise (e.g. dark blue sky has greenish blotches). The highlights tend to be pretty clean with the 400D. Shadows get really really noisy when you crank up the iso.<br>

- On the 7D at low iso, the noise levels look higher in the highlights, compared to low noise in the shadows. The highlights have some luminance noise (dark specks). At higher iso, the noise levels between shadow and highlight look very even. Some folks have commented that the noise on the 7D looks more "film-grain like" rather than some digital artifact. I wonder if it is because of these qualities.</p>

<p>I have yet to make a large print from a 7D file. In the past, noisy shadows and blotchy chroma noise have been problematic but some lumina noise in highlights (e.g. sky) has not been as problematic - the dithering takes care of that. I do know that noisy highlights are problematic for screen dispays. My speculation is that more even noise levels between shadows and highlights will produce a nice looking print. Maybe this weekend I'll have time to give it a try. I'll keep you folks posted.</p>

<p>ok enough procrastinating time to get back to work...</p>

<p> </p>

[images_in_light_north_west]

<blockquote>

 

 

<blockquote>

<p>I wanted the 7D to be a very good camera to go with my 5D2, I believe Canon could have done this, but chose not to</p>

</blockquote>

<p>I really haven't got a clue what that's supposed to mean - by what possible measure can the 7D be described as anything other than excellent? It is clearly <em>and without question</em> the best APS-C camera available today.<br /> And that's not emotion speaking - it's an objective reaction to the evidence.</p>

 

 

</blockquote>

<p>It means they could have done better but chose not to, think about it, they cant make there APS-C camera better in IQ than the 1Dmk4 or the 5Dmk2 for that matter, that would kill sales, its part of buisness. So I choose not to buy the 7D, I didn't say it was a bad camera, but I want IQ that is very close to my 5D2 and I have not seen that from the 7D, so maybe a used 1D3 will be in my future.</p>

[keith reeder]

<p>Arie,</p>

<p>which converson software did you use? There is significant variation in end results as far as noise is concerned, depending on how the files were converted, and neither ACR nor DPP are optimal.</p>

[julian_robinson4]

<p>Thanks for the very interersting and informative photo Ariel.</p>

<p>The lower right crop sample nicely demonstrates your point about the noise in shadows vs noise in highlights. I wonder whether the problem is actually not that the bright parts are noisier but that the blue channel is noisy, since it is the sky that is most noisy and the sky is most of the brighter parts of the image? I don't see much noise in the yellow bright parts. </p>

<p>This amount of sky noise is similar to the 50D and a level I find unacceptable. As mentioned before, it's difficult to see how Canon have managed to improve high-ISO noise substantially - amazingly in fact - and not changed lower-ISO noise at all. The 7D should be a significant improvement over the 400D since it's at least 2 models newer AND is 2 levels up the range and price, yet it still has the same noise as the 400D. The maths would suggest that if high-ISO noise is improved this must be reflected in an equivalent improvement at low ISO's. The fact that this hasn't happened would seem to indicate that it's a deliberate policy by Canon for the reasons mentioned before and by Ross above. Not good reasons in my opinion.</p>

<p>I'm not sure I share your enthusiasm for the level of detail and resolution. It's certainly better than on some other web examples including Darwin's, but I don't think it demonstrates the ability to deliver good feather detail in often heavily cropped shots (which is my holy grail). The 5D does do that. But I'd agree it's hard to tell from this very different shot, and hard to separate the camera's abilities from the lens, especially in the presence of so many pixels.</p>

<p>From which point of view is there any chance of a similar (comparative) photo at f5.6 and f8 which I believe would reassure the diffraction phobics who believe that anything shot at f8 with the 7D will look soft. (Alternatively, I might learn that this is true). Yours is the perfect lens for such a test because it should be as sharp as at f4 for comparison. It certainly looks the goods here.</p>

<p>re the previous post, another illustration of Groob's law, that the ability to read and understand forum posts is inversely proportional to the desire by the member to post. If people comment without reading properly, it's no wonder we waste so much time on forums answering things that have already been answered. And engaging in aggressive personal abuse because we don't even know what the previous person has said, rather than trying to learn and share information.</p>

<p>As for DPP, if it delivers results similar to in camera processing (as stated, and as you would imagine it should) I think it's very unlikely that a 3rd party converter could do significantly better. A 3rd party has to reverse-engineer and guess things that the manufacturer already knows so is not likely to do a better job of conversion than e.g. Canon. I'm happy to be proven wrong though so look forward to a demonstration. People have said the Capture One is better than DPP in some ways, though I don't know if this includes detail and noise, and I haven't seen an example to demonstrate.</p>

[keith reeder]

<p>Julian,</p>

<p>instead of regurgitating suitably supportive soundbites you've tripped over on the internet, you might - <strong>as I have done exhaustively</strong> - source a range of 7D files and convert them in a variety of programs.</p>

<p>Do that and you'll find <em>quite unequivocally </em>that several converters are better than DPP, and pretty much all are better than ACR/Lr: Cap One is very obviously better than DPP, but not the best <em>for 7D files</em>.</p>

<p>Interestingly, the free Raw Therapee is<em> excellent</em>, producing squeaky-clean files across the board at lower ISOs; and Bibble 5 is, if anything, better again.</p>

<p> Why? Because their demosaicing algorithms suit the 7D's characteristics.</p>

<p>Bibble 5 will actually automatically switch to the "best" algorithm based on the camera profile; and Raw Therapee gives the user a selectable choice of algorithms, and while its EAHD algorithm is very good, VNG-4 is even better - not a surprise, given the 7D sensor's RGGB CFA. </p>

[arie_vandervelden1]

<p>Julian: the detail is there. Have a look at the 100% crops - blow them up to 400% or 800%. You'll see. I think the 5D2 is a better camera for you. It's got less noise. Sounds like you're passionate about low noise, and the extra cost of the FF system will be worth it for you.</p>

<p>Keith: thanks for the tips on raw conversion. I'll evaluate Raw Therapee and Bibble this weekend.</p>

[py-photography]

<p>Keith, I'll second the thanks..<br>

I'll be giving themm both a look on the weekend as well.</p>

 

[py-photography]

<p>Release candidate lightroom 2.6 a try right now, I have some high ISO shots of my son playing indoor soccer...</p>