What’s Up?
It was cloudy with a light mist on the morning of Day 3 of the San Diego IPT. We did well at the cliffs with lots of nice pelicans and more chances on the pair of Brown Boobies. We had a great lunch at The Crab Catcher in La Jolla. As reported the Wood Ducks at Santee were nowhere near up to par but we did have lots of drake and hen Ring-necked Ducks along with a (finally) cooperative male Ruddy Duck in winter plumage.
|
|
|
Like yesterday’s peregrine image, this image was created at La Jolla, CA on Day 2 of the 2016 San Diego IPT with the Induro GIT 304L/Mongoose M3.6-mounted Canon EF 500mm f/4L IS II USM lens, the Canon Extender EF 2X III, and the Canon EOS 5DS R. ISO 400: 1/1250 sec. at f/9. AWB. Center AF point (by necessity)/AI Servo Expand/Rear Focus AF as originally framed was active at the moment of exposure. This is a very small crop from the left and the bottom. Click here to see the latest version of the Rear Focus Tutorial. Click on the image to see the incredible fine feather detail in a larger version. Brown Pelican face detail |
Dealing With My New 1000mm Lens and the Canon 5DS R…
Going back to the 7D II and long effective focal lengths, I had the gut feeling that lens shake was more of a factor than with previous camera bodies and assumed that what I was sensing was because of the higher mega-pixel counts. Most folks disagreed but I was still sure that with a 7D II, and more recently with the 5DS R, that one had better keep the lens perfectly still in order to create sharp images.
As it turns out, it looks as if I was both wrong and right. Right about my gut feeling that lens shake was more of a factor with both the 7D II and the 5DS R than with previous camera bodies, and wrong about the reason. Below are selected quotes from an e-mail from good friend and many multiple IPT veteran Alan Lillich.
First a definition. Pixel pitch the distance from the center of one pixel to the center of the next pixel measured in microns, millionths of a millimeter.
Here is my take, derived from basic principles. Remember that pixel pitch is just the spacing of the pixels. All of my body-to-body comparisons assume the same lens–side by side, the same aperture and shutter speed, at the same subject distance. The visibility of shake is measured by viewing at 100%, looking at the sharpness of the finest subject details.
With a given camera body, lens, tripod,tripod head, wind, shutter speed, and level of operator skill there will be a certain amount of shake during the exposure. Physically this shake means the image moves across the sensor during the exposure. How much it moves is a just a distance. How visible it is depends on the pixel spacing. (And probably on the AA filter, Bharat’s comment in the previous post on this subject is interesting.) I have no idea what the actual shake amount is in any situation, nor exactly how much the AA filter (on all the bodies but the 5DS R) smears things. We do know the pixel spacing for specific bodies, but for this discussion all that matters is the ratio of shake to pixel spacing.
If the shake is less than the distance between pixels then it will be invisible. The more pixels it covers, the more visible it is. It is hard to compare given the span of time and changes like the improvements in IS with the Series II super-telephotos. With each new camera the same sensor size with more pixels means tighter pixel spacing (smaller pixel pitch); the same shake will cover more pixels and be more visible.
Let’s compare the 7D II to the 1D IV. The 7D II has a pixel pitch of 4.1 microns, the 1D IV has a pixel pitch of 5.7 microns. So a 12 micron shake would cover about 3 pixels on a 7D II and about 2 pixels on the 1D IV. In other words, the shake should be 50% more visible on the 7D II than on the 1D IV. You have to be that much more careful with the 7D II than with the 1D IV.
The 5DS and the 5DS R have a pixel pitch of 4.14 microns. I’m getting these numbers from quick web searches. So I’m confident they are accurate, but don’t know how precise. The 7D II number I saw was just 4.1, with no hundredths digit. Let’s just say the 7D II and 5DS have the same pixel pitch. So the 12 micron shake mentioned above would also cover 3 pixels on the 5DS and the shake has the same effect as on the 7D II. So you need to use the same care with the 7DF II and the 5DS.
Again, this is with same lens, side by side, same subject. Another way to look at this, especially relevant to 100% examination, is the effect on subject detail. The projected image on the sensor has a size determined by the lens and subject distance. A bird’s eye or feather details will be a certain physical size on the sensor. So a certain amount of shake will affect a certain fraction of the most detailed features. How visible those features are depends on the pixel size.
Now let’s consider the anti-aliasing (AA) filter on all cameras but the 5DS R. As Bharat’s comment pointed out, it blurs things so will hide some degree of shake.
What this means is that you need more care with a 5DS R than with a 5DS or 7D II.
Bottom line, better technique will get you better photos in general. The smaller the pixels and the tighter the pixel pitch, the better technique you need. Pixel count alone is not relevant. All of this of course in a theoretical same lens side by side situation. Suppose you and I are sitting side by side on the lower shelf at the Cave Store cliffs shooting tight portraits of pelicans. We have the same lens, but you have a 5DS and I have a 7D II. You can’t get closer without scaring the birds, I can’t move back because of the cliff. So we can’t get the same framing but do get the same pixel coverage of subject features. So we need equal technique for equal sharpness as seen at 100%. With you owning and using the 5DS R, you actually need to take a bit more care as there is no AA filter to blur the effects of lens shake.
Though you many not previously have understood all the technical stuff above, I think you have grasped all of it inherently given the magic you do with super-telephoto lenses used with either teleconverter at slow shutter speeds…
So Yeah…
So yeah, when working with the 5DS R at long effective focal lengths, I do need to take extra care in order to create super sharp images. With today’s featured image, a fast shutter speed did the trick. With yesterday’s peregrine image made at a much slower shutter speed, I made sure that the tripod legs were firmly planted in the soft hillside earth, made sure to press my face against the back of the camera, made sure to tighten both the horizontal and vertical panning knobs on the Mongoose M3.6, and made sure to steady the lens with my left hand. Most of the time I support the lens barrel from below with my left hand while my forearm rests near the tripod platform; at other times I place my left hand atop the lens barrel. Which I do depends on the situation and is not anything that I could put into words. It is more of that gut feeling stuff.
The San Diego Site Guide
Whether you are visiting San Diego for photography for the first time or live in the area and have done the pelicans many dozens of times, you will learn a ton by studying the San Diego Site Guide. Why spend days stumbling around when you can know exactly where and when to be depending on the wind direction and sky conditions? In addition to the pelican primer, there is great info on the best beaches for the gorgeous gulls, on Marbled Godwit, on the lower cliffs, Lesser Scaup, and Wood and Ring-necked Ducks as well.
Learn more or purchase your copy here.
Please Remember to use our Affiliate Links 🙂
To show your appreciation for my continuing efforts here, we ask, as always, that you get in the habit of using my B&H affiliate links on the right side of the blog for all of your photo and electronics purchases. Please check the availability of all photographic accessories in the BIRDS AS ART Online Store, especially the Mongoose M3.6 tripod heads, Gitzo tripods, Wimberley heads and plates, LensCoats and accessories, and the like. We sell only what I have used, have tested, and can depend on. We will not sell you junk. We know what you need to make creating great images easy and fun. And we are always glad to answer your gear questions via e-mail. I just learned that my account was suspended during my absence; it should be up and running by Monday at the latest.
I would of course appreciate your using our B&H affiliate links for all of your major gear, video, and electronic purchases. For the photographic stuff mentioned in the paragraph above we, meaning BAA, would of course greatly appreciate your business. Here is a huge thank you to the many who have been using our links on a regular basis and visiting the BAA Online store as well.
Be sure to like and follow BAA on Facebook by clicking on the logo link upper right. Tanks a stack!
Typos
In all blog posts and Bulletins, feel free to e-mail or to leave a comment regarding any typos or errors. Just be right 🙂
Hi, Artie and others. This has been a very interesting discussion. I have come to the conclusion that the important factor is resolution. Very simply put, if the shake is small, then a low-resolution camera won’t detect it and a high-resolution camera will. You can talk about pixel pitch, but what determines resolution is the number of pixels on the image, all else equal; and the number of pixels is inversely proportional to pixel pitch (given similar designs). My Canon 6D and 7D2 both are about 20 MP. If camera shake produces a displacement of the image that occupies the same linear portion of each sensor, then the shake will be equally visible in each camera (discounting differences in noise), because they have roughly equal resolution. If each camera has a lens that covers the same field of view (same effective focal length), then the same angular displacement (shake) on each camera will be equally visible on each camera. So while pixel pitch might be slightly more accurate technically–and I’m not saying it is–thinking about resolution is easier and more intuitive and thus more useful in this case.
Only for the same sensor size. Read my comment to Bill Richardson about Olympus versus Phase One, 16MP versus 100MP. Or think about the 7D II versus 5DS, same shake effects but very different resolutions. If your cameras are all one size then your simplification is great. But be careful using that as a rule for others who might have two or more sizes.
For me the takeaway is that we just ought to get in the habit of being more careful if our images are not as sharp as we like.
Alan: the 6D is “full frame” and the 7D2 is APS-C (“crop”), so their sensors are not the same size, although they have the same number of pixels (resolution). That’s why I talked about lenses of the same effective focal lengths. The “full frame” 5Dr/s has the same pixel density (number of pixels per area) as the 7D2, so with the same lens mounted, the two cameras would show the same shake. I completely agree with you that the main message is to take care, and I think we agree that increasing resolution or increasing the focal length of the lens both increase sensitivity to shake, all else equal.
Thanks for the interesting discussion. It begs a related question. I expect if you have an image made with a 5DS or 5DS R that has minor degradation from image shake you would need to blow it up substantially (on screen on in print) to actually see the blurring of fine detail. So if you have, say, two rigs that are identical in every way except one has a 5DS and the other has a 5D III set up side by side and use them to make matching images with the same amount of cameral shake (enough to cause degradation on the 5DS but not on the 5D III) then blow both images up to the point that the 5DS image’s degradation is visible how would the images actually compare? That is, would you still be in the range where the 5D III image remains crisp, or would it be getting soft from pushing the limits of how far it can be enlarged? If the latter, which would have better subjective image quality at that magnification? This would be an interesting experiment, but would require controlled conditions to perform accurately.
Regarding the need for extra care with the 5DS R vs the 5DS, a very slight use of Gaussian Blur should (I have not actually confirmed this) be able to replicate the anti-aliasing filter when needed (to address mild image shake or moiré).
It’s this kind of techno reading that keeps me coming back to this blog…..I’m just new and you can’t learn this type of jazz this easily anywhere else. Keep on rocking Artie!!!!!!
Yeah I hear so many people complain about technical information on photography sites, but they fail to realize that photography is both technical and creative. When we have a problem or something that doesn’t make sense, sometimes it is best to look for technical issues.
Hey Ward, Thanks for your kind words. I am a techno-phone as well as being pretty much technically illiterate but I am fortunate enough to have friends like Alan Lillich who are technically brilliant. I struggle to understand all of this stuff but with Alan’s help I have been getting it 🙂
a
Well, this sure explains why my 1D Mark II always appeared sharper with lower shutter speeds than my 7D. I knew it had to have something to do with pixel density, but I couldn’t connect the dots (pixels in this case).
Thanks for pointing this out to us Mr. Morris.
YAW Matt. artie
How is pixel pitch related to noise? Or is it? The noise level in my MkIV is less than my 7D. The former has a larger sensor with a slightly smaller number of pixels and a smaller pixel pitch.
The annoyance factor of noise is in the beholder. I don’t like it. It’s one reason I sold the 7D. There are denoise programs but as the saying goes – Noise degrades an image. When you denoise, you get a degraded image with less noise.
In my opinion the most important thing is sensor technology. For example, the pixel pich of the 1d mark III sensor is larger than the ones of 1dx, 6d and 5d mark III, however the sensors of these cameras have a better dynamic range, and signal to noise ratio.
As a limit case, if you take the first eos 1d, probably the high iso noise performance is comparable to some of the current point-and-shoot cameras, even if the pixel pitch of the 1d sensor is about 11,5 micron…
In every case, consider that canon sensors includes an internal circuitry to reduce noise of the electrical signal, before that the analog to digital conversion is made. Even if such circuitry has been improved during the years, my guess is that at high iso you lose informations as well as noise, and this happens before that the raw file is created…
I started writing a long reply about noise theory, but it is taking longer than expected and I am busier this week than expected. I will write something and see if I can get Artie to include it in a future post.
I generally agree with Matteo that sensor technology is probably more important than pixel size in practice – so much for the old saw about bigger pixels (always) being better.
Thanks. Very good information. This may explain why I always get sharp images with the 1DX than with the 7DII.
Hi, Artie, and thanks for the explanation about shake. But I still am puzzled. As has been said, when the lens/camera combination shakes, the image moves across the sensor. The bigger the movement across the sensor, the more visible the shake is. A longer lens will produce more visible shake for the same angular movement than a shorter lens because it’s a greater portion of the field of view. I wonder how much pixel pitch actually affects shake in practice. I guess it matters theoretically because each pixel is recording its own signal. In any case, I find I get the most remarkably sharp images with my Canon 6D and Canon EF 28-70 F/2.8L lens.
A micron is one millionth of a meter, not one millionth of a millimeter.
Thanks, that was my mistake not Artie’s.
Interesting discussion, thanks. I would have thought placing your face on the back and hands on the lens would introduce shake rather than remove it. Do we need to increase the old maxim of shutter speed faster than focal length? Though that only applied to hand holding anyway.
Also are you saying that people with current generation cameras really ought to have newer lenses? I have the old 500 & can’t see being able to replace it for a long while.
Hi Gary, I have been teaching folks to damp the camera body with their face and damp the lens with their left hand/forearm since as far back as the original ABP 🙂
With the 5DS R you simply need to take more gear in reducing lens shake. And yes, higher shutter speeds help.
a
I have a signed copy, clearly time I have another read.
Excellent explanation. This + the higher noise levels of mega pixel cameras at higher ISOs + slower frame rate gives me pause on purchasing one for my kind of shooting–active, non studio.
Remember, it is higher noise levels of small pixel cameras, not high pixel count cameras. These are the same only if the sensor size is the same, which often gets lost in discussion. My Olympus E-M5 has just 16 megapixels. But with it’s 2x crop factor that is the same as 64MP in 35mm full frame. The new Phase One 100MP back is surely high megapixel. But its sensor has about 2.5 times the area of 35mm full frame, so it equates to about 40MP, just 2/3 that of the 16MP Olympus in terms of pixel denisity.
P.S. The point of higher noise with smaller pixels is a complex topic. I’m writing a long reply to Leonard.
Alan Lillich: thanks. Both are complex topics. On pixel pitch and amount of movement, it seems to me that if the movement from shake is say half the size of the pixel pitch, it still affects all the pixels, not so?
Yes, all the pixels are affected, but the effect on subject features – the things that define sharpness – is smaller for smaller movement. Think about a fine grid like a window screen. The image of a wire covers some small number of pixels, let’s pick 10 as an arbitrary example. A half pixel smear won’t do much to the image. But a 5 pixel smear will make a wire’s image half again as wide, ruining edge contrast.
Ot think about taking a sharp image and average each pixel with its immediate neighbors. The image is still pretty clear. Now average each pixel with 100 neighbors, serious softening, and with 1000 neighbors just mush.