Canon 7D and DLA

<p>The DigitaPicture.com website describes DLA (Diffraction Limited Aperture) as the result of a mathematical formula that approximates the aperture where diffraction begins to visibly affect image sharpness at the pixel level. You can read about the DLA of all Canon Digital Cameras at the DLA chart at the bottom of the Specs on this page: <a href="http://www.the-digital-picture.com/Reviews/Canon-EOS-5D-Mark-III-DSLR-Camera-Review.aspx">http://www.the-digital-picture.com/Reviews/Canon-EOS-5D-Mark-III-DSLR-Camera-Review.aspx</a> <br />According to this chart, the Canon 7D has one of the lowest DLA's of any of the semi-pro Canon cameras including some of the earlier models ?</p>

<p> </p>

<p>This is one of the most often misconstrued things in photography, especially when it leads incautious readers to extrapolate nonsensical conclusions such as a) one cropped camera has better diffraction performance than another cropped sensor body, b) higher photo site density cameras are more prone to diffraction blur than lower photo site density cameras, or c) you'll have to shoot at larger aperture with one cropped sensor than with another in order to avoid diffraction blur.</p>

<p>If you make a photograph with a 8MP cropped sensor camera and a photograph with a 18MP cropped sensor, using the same lens and aperture, and then make prints at the same size from both photographs... they will have precisely the same amount of diffraction blur. Diffraction blur is an optical phenomenon caused by lenses, and the point at which it becomes an issue as you stop down is affected by sensor dimensions. Sensor photo site density does not affect diffraction blur.</p>

<p>Dan</p>

<p>The higher the pixel density, the smaller the pixel, the smaller the pixel, the lower the fstop where diffraction starts to kick in(affect your image). Since Canon doesn't offer a DSLR with a higher pixel density than the 7D, it is also stands to reason that it would have the lowest "DLA" number (or close enough anyway).<br>

Different cropped cameras have different levels of this 'DLA' because they're pixels are different sizes. This is one aspect that isn't rocket science. The subjective interpretation of this is of course ... subjective, and of course the amount of diffraction varies with the lens.</p>

<p>G Dan is correct, however, his interpretation ignores a critical point. Since pixel sites are quantified, a larger pixel site yield a smaller amount of <em>recorded</em> blur <em>due to diffraction. </em>In a nutshell<em>, </em>even though pixel density doesn't change the<em> amount</em> of diffraction blur incident upon the sensor, it changes how much of it<em> is recorded</em> into the file, hence the 'DLA' concept.<br>

So, while the 8mp sensor 'sees' the same diffraction as the 18mp one, it <em>records</em> vastly less. </p>

<p> </p>

So, while the 8mp sensor 'sees' the same diffraction as the 18mp

one, it records vastly less.

 

What Marcus says is correct, but at the same time, the 8mp

sensor also sees the same detail but is able to record less than

the 18mp sensor. In the end, the two cancel each other out, which

brings us back to what GDan said

 

Don't lose any sleep over it, it doesn't matter as much as some

literature may make it out to be

<p>"<em>Since Canon doesn't offer a DSLR with a higher pixel density than the 7D</em>"<br>

<br />Except for the 60D of course which has the same exact DLA as the 7D. Does this explain why some users complain about soft-focus issues, although I never had that problem ?</p>

By the way the DLA is not an absolute limit - just a calculation. In the real world you will not see one stop over the DLA

and would have to print very large or pixel peep to really see two stops over. Thus the 7D is fine at F11 but diffraction

does start to show above this.

<blockquote>

<p>What Marcus says is correct, but at the same time, the 8mp sensor also sees the same detail but is able to record less than the 18mp sensor. In the end, the two cancel each other out, which brings us back to what GDan said</p>

 

</blockquote>

<p>This. Exactly.</p>

<p>Diffraction is just this year's fashionable <em>bête noire</em> for people who <em>have to have</em> something to complain about.</p>

<blockquote>

<p>Does this explain why some users complain about soft-focus issues</p>

</blockquote>

<p>No, but it does go some way to explain why the Darwin Wiggett 7D "review" is the subject of such ridicule (along with anyone who references it as "proof" of the 7D's failings).</p>

<p>The dozy sod was testing the 7D at <em>f/16</em> - the Airy disc caused by diffraction will be about two 7D pixels across at f/16 - not sharpening the files, and then complaining about softness at 100% view..!</p>

<p>(He also used a 1/60 shutter speed for testing the camera with a 300mm lens outside, but that's another story).</p>

<p>Good explanation of the Real World non-issueness of DLA <a href="http://photography-on-the.net/forum/showthread.php?t=747761">here</a>. This quote from it says all we need to know:</p>

 

<blockquote>

<p>I am by no means a photographer (only use a camera to shoot my own work) or smart enough to understand the technical discussions, but as I posted a couple years ago that I did very controlled comparisons between my 5dc and my brother's 5dii before I bought one.<br /><br /> Table top stuff, using strobes to control variables. Same distance to piece, same lens, etc.<br /><br /> The question in my mind at that time was whether even at f/16 would the extra pixels buy me any more detail. Important if say an editor wants to crop a shot and then make it 8.5 x 11 for a mag.<br /><br /> The results were very clear. Much more detail available with the 5dii, even at f/16. And when I was presenting my results to my brother I also downsized the 5dii images to the same pixels dimensions as the 5dc to compared. Huge difference, with the 5dii coming out on top. I have a number of lenses which I have compared to each other in fairly controlled tests, and the difference between the 5dc and 5dii (in terms of detail) was bigger than the difference between any two of my lenses (all decent lenses).</p>

</blockquote>

"The dozy sod was testing the 7D at f/16 - the Airy disc caused by diffraction will be about two 7D pixels across at f/16 -

not sharpening the files, and then complaining about softness at 100% view..!"

 

Amazing, do these people have no education or physical intuition at all.

To be fair the link that Harry provides does explain what the DLA is, but unfortunatly I have found that people really need

to have the implications spelt out to them; the foot note probably does not do enough spelling out.

<p>For some reason my pictures are not looking as sharp as they used to...</p>

<p>I was reeding all this posts about 7D just for fun and did some digging online. It looks like engineers now have technology, to put unlimited number of pixels on camera sensor.<br>

So I decided to look at this problem with sharpness on high density sensors, from film era point of view, we still keep the same format as film use to be: half frame APS-C, full frame 24x36, medium 4.5x6, 6x6, 6x9( it is in metric system, for those who are not old enough). Those days quality and size of final image, using reasonably good lenses, depended mostly on size of frame used, with some film grain, depending on ISO number, small variations. You normally would not expect print poster size from APS-C film, if you need large print you will drag Mamiya or Rolleiflex from closet.<br>

I think we have the same problem here, those cramped in small sensor megapixels showing all limitations of small format, you have to use best available (most expensive) lens to use that resolution and diffraction is still clawing it back. Computer do not understand difference between formats, computer reeding pixel dimensions. Cellphone, APS-C, full frame cameras all the same for photoshop. So you see overblown to 1.6 time lens deficiency, motion blur and diffraction from 100% magnification. That's why 8 megs Canon 30D seems sharper, than 18 megs 7D, it is the same format. It is law of diminishing returns.</p>

<p>"<em>Amazing, do these people have no education or physical intuition at all</em>."</p>

<p><strong>Most</strong> people buy a car with no implications on how the car works. They are truly not interested on how the gas is mixed with the air(Co2) which produces combustion and makes the pistons go up and down in the engine case, which makes the crank works which spins the wheels. They are just not interested all that stuff. All they want to do is drive back and forth to work. </p>

<p>Yeah, but Wiggett presented himself up as a self-proclaimed expert whose opinion actually meant something because of his "expertise" - that's the point, Harry.</p>

<blockquote>

<p>What Marcus says is correct, but at the same time, the 8mp sensor also sees the same detail but is able to record less than the 18mp sensor. In the end, the two cancel each other out, which brings us back to what GDan said</p>

 

</blockquote>

<p>One is not quite completely unlike the other ;-). Yes, I had written a short paragraph about that, but then I deleted that part from my original post because I thought that it was pretty elementary. <br>

...but the question is... does the difference <em>actually</em> cancel out completely? My intuition says that despite it's abyssmal (f6.9!) predicted performance, it will actually out resolve an 8mp sensor, even at redonkulously high f-stops... I could be wrong (it's happened once or twice), but we won't know until we test it out. Perhaps a 30D vs. 7D 'DLA' shootout is in order? ;-)</p>

<p>Marcus wrote:</p>

<blockquote>

<p>"G Dan is correct, however, his interpretation ignores a critical point. Since pixel sites are quantified, a larger pixel site yield a smaller amount of <em>recorded</em> blur <em>due to diffraction. </em>In a nutshell<em>, </em>even though pixel density doesn't change the<em> amount</em> of diffraction blur incident upon the sensor, it changes how much of it<em> is recorded</em> into the file, hence the 'DLA' concept.<br /> So, while the 8mp sensor 'sees' the same diffraction as the 18mp one, it <em>records</em> vastly less."</p>

</blockquote>

<p>Actually, what Marcus says is wrong. It is time to put this nonsensical misunderstanding of the effect of diffraction blur on digital sensors away for good.</p>

<p>There is no more or less diffraction blur recorded by a system with larger or smaller pixels. To assume otherwise is to imagine that all image elements line up in perfect vertical and horizontal rows that somehow magically fall entirely within the area of individual photo sites - and none are port of objects with curved edges, tilted on a diagonal, or contain more or less random textures.</p>

<p>What I wrote about the photograph at a given size from the two cameras, same lens, same aperture and printed at a given size having the same amount of diffraction blur happens to be exactly correct. At best, you might be able to logically (but photographically speaking, irrelevantly) claim that the higher photo site density system provides a more accurate image of the same diffraction blur that you would see in the lower photosite density system, but there is no more or less diffraction.</p>

<p>In fact, it turns out that there is <em>no bad news in terms of diffraction at smaller apertures</em> yet there is potential <em>good news at larger apertures</em> if your lens happens to be of very hiqh quality. In the latter case (larger apertures) you might be able to produce a very slightly higher resolution image that might be printed slightly larger if everything worked out perfectly.</p>

<p>Bottom line: There is no way in which higher photosite density has a diffraction blur downside in a photograph of a given size.</p>

<p>Dan</p>

<p>BTW: The classic goof that people make when trying to "prove" that diffraction gets worse with higher photo site density is to compare 100% crops from the two systems. If you understand why that is a big mistake that produces misleading results, you get it. If you don't understand, let me know and I'll explain it.</p>

<p>This is <em>deja vu</em> all over again. I remember discussions like this on the letters pages of magazines back in the 'sixties. The only difference was that they were arguing about Pan F versus Tri-X and throwing in the speed difference to confuse things even further.</p>

<p>I do remember that a very widely published photographer called Kevin MacDonnell summed it all up rather nicely by saying something to the effect of "if you put your camera on a tripod and focus carefully it all becomes irrelevant."</p>

<p>Even though I read this review prior to purchasing my 7D I never noticed the part about DLA. Tell you the truth I sort of had my mind made up before reading this review. I just noticed the DLA chart when browing through the 5D III review. The reason why I jumped to conclusions was because of the Heated post a couple of days ago concerning the 7D and sharpness. Today I went over the 7D Review from that site one more time and they do mention that the 7D is not as sharp as the 50D. One way to alleviate that is to bump up the sharpness by 2 notches.</p>

<p>"Today I went over the 7D Review from that site one more time and they do mention that the 7D is not as sharp as the 50D. One way to alleviate that is to bump up the sharpness by 2 notches."<br>

There are several possible reasons that someone might erroneously conclude that the 7D is "not as sharp" as the 50D.</p>

<ul>

<li>the person compared 100% crops and failed to realize that they were looking at a smaller section of the image on the higher MP body.</li>

<li>the test lens did not function the same on the two cameras. AF systems are not always uniform from body to body, and you would need to test a number of samples to rule this out, or else rely on manual focus.</li>

<li>in-camera sharpening settings were modified by Canon between the two camera models, and the difference reflects a software change rather than the capabilities of the camera.</li>

<li>the person used raw files and converted them identically, not realizing that different settings are appropriate for different cameras</li>

</ul>

<p>Notice that "DLA" is not on the list. ;-)</p>

<blockquote>

<p>Today I went over the 7D Review from that site one more time and they do mention that the 7D is not as sharp as the 50D.</p>

</blockquote>

<p>Ho Hum...</p>

<p>The end result from my 7D's a <em>helluva</em> lot sharper than that of my shooting pal's 50D...</p>

<p>G Dan might also have included "different AA filters" as a possible cause for a perceived sharpness difference between two different cameras, but there's no inherent lack of sharpness with the 7D.</p>

<p> </p>

<p>"<em>There are several possible reasons that someone might erroneously conclude that the 7D is "not as sharp" as the 50D."</em><br>

First of all the 50D is a very sharp camera and has very good dynamic range. I just love the images that come out of that camera. Overall it is a very underrated camera sort of like the 30D. I'm not to crazy about the 60D however.</p>

<p>Just a few weeks ago I do remember taking a shot with my 7D at f16 and 1/100. I shot at f16 because this was a scenic shot and I needed the DOF. I was a little confused why I wasn't getting the sharpness I usually get at that aperture ? I could not put my finger on it, so I wrote-it off as operator error or atmospheric conditions. Now I'm thinking that maybe it's time to do some test at smaller apertures with that camera.</p>

<p>Harry, a great way to actually find out how to make aperture choices in light of the very real effects of diffraction blur is to make a series of test exposures at various apertures. </p>

<p>Some years ago I had believed a lot of the silliness spouted as fact about the effects of aperture selection on image sharpness, notably that there was some ideal aperture smaller than which one should never shoot. On my then-new 5D I was religiously avoiding apertures much larger than f/8 because I had read more than once in various discussions that the diffraction limited aperture (sound familiar?) for this camera was something like f/8.</p>

<p>Being the inquisitive sort, I went out with the camera and a favorite lens, set everything up on a tripod, and simply ran through the whole aperture values in a series of shots of a very boring subject. What I discovered was that on full frame there is essentially no visible difference between f/8 and f/11 and that the difference at f/16 is so absurdly small that you would be very hard pressed to see it even in side-by-side 100% crops. I print large and I now never hesitate to use f/16 if I need it. I'll even go to f/22 if the subject can stand a tiny bit of softness in exchange for a longer exposure or slightly larger DoF.</p>

<p>I don't currently have a cropped sensor body, but if I did I would generally shoot at f/5.6 or f/8 (depending on a variety of factors), occasionally push things to perhaps f/11, but pretty much steer clear of f/16 and smaller.</p>

<p>I'm often somewhat astonished that people will go on at length about stuff like this on an entirely theoretical basis and never stop to just make a few darn test shots to see how it really works. </p>

<p>Dan</p>

<blockquote>

<p>I'm often somewhat astonished that people will go on at length about stuff like this on an entirely theoretical basis and never stop to just make a few darn test shots to see how it really works.</p>

</blockquote>

<p>Just "often" astonished, Dan?</p>

<p>It depresses me on a daily basis that almost the entire body of "knowledge" on the internet these days comprises untested opinions and unfounded rumours regurgitated <em>ad nauseam</em> on internet forums.</p>