This new review of the Mark IV served to remind me of a question that's been puzzling me, which is that of purportedly needing higher shutter speeds to avoid motion blur as we move to higher resolution cameras. I've seen this mentioned any number of places, including Canon's Mark IV white paper.
This review puts it this way:
The high pixel density of the 1DmkIV has two negative implications - it puts great demands on lens quality and also the camera is more susceptible to camera shake/motion blur. I find that I have to raise my shutterspeed a little to ensure getting a sharp image on the mkIV. To give you an idea of this, I took some images with the three cameras (plus 500mm f4 on a tripod trained on a distant target) reducing the shutterspeed progressively until the results were no longer critically sharp when viewed at 100%. I found that I could get away with 1/30th sec on the 1DsmkII, 1/80th sec on the 1DmkIII and 1/125th on the 1DmkIV. This is quite significant and would necessitate increasing the ISO setting over the other cameras by up to 2 stops in low light conditions to compensate.
The key seems to me to be the "viewed at 100%" part. Isn't this nothing more nor less than a matter of greater magnification? A Mark IV file viewed at 100% on a given monitor will be magnified to a greater degree than will a Mark III file — kind of like looking at an 11x14 print instead of an 8x10. And we all know that the larger you blow up your files, the more critical each of the various components that go into "sharpness" — lens quality, precision of focus, camera motion, subject motion, etc. — becomes. So yes, if we're going to magnify our images to a greater degree, then we have to be more careful about these things.
But the fact is, aside from when we're pixel-peeping at 100%, we're not (necessarily) going to be enlarging our images any more than we are already. Just because we can doesn't mean we will — I don't think I'll start selling less 5x7's and more 8x10's just because I bought a Mark IV, nice as that would be.
Put another way — and here's where the "Am I missing something?" part comes in — does "pixel level sharpness" have anything to do with overall perceived image sharpness when we're viewing the image not at the pixel level, but at "normal" print or onscreen viewing sizes? IOW — do I really need to bump up my shutter speeds if I only care what my images look like at those magnifications, not at 100%?
What makes me think I'm missing something is the difference in his findings between the 1DsMkII and the MkIV — both 16MP bodies. I don't understand how the difference in crop factor alone could lead to that much of a difference — two stops worth! — in perception of motion blur.
The higher resolution and resulting requirement for greater care to maintain sharpness is a factor if you're selling 5x7 and 8x10 prints that are significant crops of the original capture. Maybe that's not your habit coming from the IIN. Having the high resolution means it's now more of an option to crop into images rather than having to compose primarily in-camera.
If you continue shooting the way you did with the Mark IIN and simply down-rez IV files to a common print size without cropping, then you can get get away with the same settings and techniques that were suitable for the IIN.
The thing is, because it's now possible with the IV to 'zoom in' more on an image due to the higher resolution, people will do so.
The difference between the IDsII and IV is that the former is a full frame camera with a certain pixel density while the IV is an APS-H camera with an equivalent 27MP FF density (if I remember this correctly). If you crop the 1DsII to APS-H, it ends up being 10MP. I'm not sure why the person in your quote found they could shoot at 1.3 stops slower shutter speed with the 1DsII vs. the IDIII because images viewed at 100% should have similar subject detail. You can see a list of Canon cameras with comparative pixel density in a list near the top of this page: http://www.the-digital-picture.com/Reviews/Canon-EOS-1D-Mark-IV-Digital-SLR-Camera-Review.aspx The 1DsII and 1DIII have the same pixel density.
Nill Toulme wrote:
What makes me think I'm missing something is the difference in his findings between the 1DsMkII and the MkIV — both 16MP bodies.
Both 16MP bodies but different sensor sizes.
I don't understand how the difference in crop factor alone could lead to that much of a difference — two stops worth! — in perception of motion blur.
The greater the crop factor the greater the magnification. And the higher the MP count the greater the magnification when viewed at 100% on a computer monitor. If you understand why a longer focal length requires a higher shutter speed to get a clear shot, then you understand this because it's basically the same thing. The more you magnify the greater the impact of even small movements.
Thanks Ron, that makes sense. And you're absolutely right, I have already found that I can crop deeper into a Mark IV frame than a Mark II image, to an even greater degree than I think I was expecting. (Or maybe actually seeing it is just more impressive than expecting it.)
And yes, his different result with the Mark III vs. the 1Ds2 is baffling. If it were the 1D2 it would make perfect sense.
FWIW, back when the original 1D first came out, and I started shooting with it, I was pleasantly surprised that I could very easily hand hold wide angle shots at 1/4 or 1/8 and get images that didn't seem to have camera shake blur. Recently I was making low light tests with the IV with non-IS lenses and virtually none of the images were acceptably sharp at marginal shutter speeds like 1/4.
I'm skeptical whenever I read about someone moving to a new camera, like the IV, especially if it's an upgrade from a two generations or older camera and complaining that images from the new camera a soft. Or claims that the AF from the original 1D was always perfect and the III or IV sucks in comparison (problems with the III excepted). Well, yeah... If you're using the same technique that got you acceptable results on a 4 or 8 MP camera and shift that over to 16MP without any change, you're going to see issues when zoomed in at 100%. Resize those images down to 4 or 8 MP and it should look similar to what you were familiar with previously.
Gains from higher sensor resolution are definitely not without consequences.
Ken, yes, I get that... but two stops worth of difference in required shutter speed resulting solely from FF vs. 1.3x crop? That doesn't seem to compute.
Neither does the 1.5 stop difference he found between the 1D3 and 1Ds2, whose pixel densities are pretty close to identical.
And at bottom, isn't my original premise still correct — i.e., that we're really talking about nothing more than greater magnification requiring more careful technique?
I agree that it is effectively giving us more magnification when pixel peeping, and magnifying any slight camera shake/motion blur.
The advantage of these latest generation cameras is that we can have very clean images at higher ISO settings. So maybe we re-think our ISO baseline. Upping the ISO by one stop will automatically double the shutter speed for a given situation. Problem solved.
Nill Toulme wrote:
And at bottom, isn't my original premise still correct — i.e., that we're really talking about nothing more than greater magnification requiring more careful technique?
I found that I could get away with 1/30th sec on the 1DsmkII, 1/80th sec on the 1DmkIII and 1/125th on the 1DmkIV. This is quite significant and would necessitate increasing the ISO setting over the other cameras by up to 2 stops in low light conditions to compensate.
IMHO this is way overblown- I live at 700mm on a 50D & shoot at 1/50th with good results. When it gets darker I have problems with both my 20 & 50D-
I shoot dragonflys at 420mm in dark conditions, now that is tough- when ss gets to 1/20 or so it is over- In really low light I shoot manual at ss250 & use flash as main light.
YMMV
rscheffler wrote:
FWIW, back when the original 1D first came out, and I started shooting with it, I was pleasantly surprised that I could very easily hand hold wide angle shots at 1/4 or 1/8 and get images that didn't seem to have camera shake blur. Recently I was making low light tests with the IV with non-IS lenses and virtually none of the images were acceptably sharp at marginal shutter speeds like 1/4.
I'm skeptical whenever I read about someone moving to a new camera, like the IV, especially if it's an upgrade from a two generations or older camera and complaining that images from the new camera a soft. Or claims that the AF from the original 1D was always perfect and the III or IV sucks in comparison (problems with the III excepted). Well, yeah... If you're using the same technique that got you acceptable results on a 4 or 8 MP camera and shift that over to 16MP without any change, you're going to see issues when zoomed in at 100%. Resize those images down to 4 or 8 MP and it should look similar to what you were familiar with previously.
Gains from higher sensor resolution are definitely not without consequences.
I have found exactly this moving from my 1D2 to a 7D. The fact that the 7D is not only much higher resolution but a higher crop factor than that of an APS-H complicates things even more.
Nil, it's a little nostalgic to hear this topic surfacing with the 1d4. About 5 yrs ago d2x owners were asking the exact same questions ... (with a pixel density fairly equal to the 1d4.) I read all the science, all the opinions, and as you seem to be heading, finally concluded that it's simply about magnification.
For a while my 100% crops seemed to fall down a little, at any focal length. But as I spent more and more time learning long glass techniques, I guess I got steadier and steadier ...because I found the old shutter speed rules (1/focal length)
started working again. It seemed the longer the glass I got used to, the better all focal lengths got. So I've kinda concluded magnification is more about steady, than it is about shutter speed (once a certain shutter speed is met).
Anyway, about all I really know is there seems to be a learning curve to pixel density, and even with the extra magnification pixel density offers we seem to be able to adapt our shooting skills to it.
Nill Toulme wrote:
[snip]
Put another way — and here's where the "Am I missing something?" part comes in — does "pixel level sharpness" have anything to do with overall perceived image sharpness when we're viewing the image not at the pixel level, but at "normal" print or onscreen viewing sizes? IOW — do I really need to bump up my shutter speeds if I only care what my images look like at those magnifications, not at 100%?
In short, no
What makes me think I'm missing something is the difference in his findings between the 1DsMkII and the MkIV — both 16MP bodies. I don't understand how the difference in crop factor alone could lead to that much of a difference — two stops worth! — in perception of motion blur.
I would suspect the photographer. No offense to guy and his blog but when I see claims like this my spidey-sense for user error starts tingling
So, what am I missing?
Nill
I don't think you are missing anything. I have now been shooting with a 1D4 for 2 weeks and in terms of the shutter speeds required to get critically sharp images, I find its not appreciably different from my 1Ds3 (both when filling the frame and in focal length limited shooting).
The 7D is a different beast and its extremely high pixel density necessitates more care. However even when comparing it to the 1Ds3 or the 1D4 I don't need anywhere near 2 stops higher shutter speeds to get critically sharp images.
Crop makes a big difference. A full frame sensor with the same pixel density as the 1DIV's 16mp sensor (1.3 crop) would be 27mp. Think of the extreme difference in magnification when you look at it that way.
yep its that crucial line "when viewed at 100%"
of course if you have more pixels then 100% will magnify the image more.
what the tester should have done is view at the same fov and then they would have seen no different. or even printed the results. Its not rocket science to know that an accepable image printed at 6x4 could/will look unacceptable at A3 or bigger
Dawei Ye wrote:
Back in the old days with a 20D, 30D or whatnot, you had to multiply by 1.6X for the 1/FL rule
Explain that, please, because that's one I've never gotten. With a crop camera, you aren't actually extending a longer lens and thereby amplifying any len-shake due to rules of triangulation (or whatever it may be called -- geometry isn't my thing).
Mitchell Carter wrote:
Explain that, please, because that's one I've never gotten. With a crop camera, you aren't actually extending a longer lens and thereby amplifying any len-shake due to rules of triangulation (or whatever it may be called -- geometry isn't my thing).
True, but you are capturing a smaller image, and when scaled up to a print or on screen display of equivalent size you will need to magnify it by a factor 1.6X greater to achieve the same size. Therefore you will magnify blur/shake by a factor 1.6X greater too. In order to limit the amount of shake/blur to compensate you will need shutter speeds 1.6X faster.
Furthermore, that rule applies when enlarging to around 12x8 max. If you enlarge to 2X or 4X that size, as you might when pixel peeping, then you will need to double those shutter speeds and then double them again. That's why you can soon find yourself "requiring" shutter speeds of 1/2500 minimum to handhold a 400mm lens on a 7D, if you want sharp pixels. If you only need sharp images, at "normal" print sizes then 1/640 or thereabouts should be OK.
If you are going to start cropping (heavily) then the rules change again. It's all about the enlargement factor you apply to the image, not the number of pixels. That said, for a sports/action camera I find the pixel density of the 7D to be needlessly high. The 1D4 doesn't even come close to that level of pixel density, thank God. It has a density equivalent to a 40D. I have no idea why the 7D needed 18MP. Lovely for well lit static subjects and scenes. A complete waste of storage for action shooting in anything but excellent light.
I make a distinction between "magnification" and "enlargement." I use magnification to describe the lens' magnification ratio; i.e.,
M = b/v,
where b is the image distance and v is the object distance. (From Toothwalker's excellent DOF pages.) Thus, the EF 100/2.8 macro is capable of M = 1.0x at minimum focusing distance.
By contrast, I reserve the term "enlargement" to refer to the ratio of the viewed print dimensions to the sensor dimensions. So a 24 x 36 cm print made from a standard 35mm "full-frame" sensor has undergone an enlargement ratio of exactly 10x.
The reason why I make this distinction is because a lot of people confuse the two, and it can lead to misunderstandings when we talk about pixel density and what it means to view a print (or a displayed image on a monitor).
As it relates to slowest acceptable shutter speed, the pixel density of the sensor array is what counts, not the actual physical dimensions of the array. This makes logical sense because if you have a 35mm sensor and you simply "cut out" the central APS-C sized rectangle, they should have the same susceptibility to (or tolerance of) camera shake.
The reason why camera shake is more visible on high-density sensors is because the shake manifests itself as a change in the framing of the subject as a function of time. The displacement as a function of time therefore can be reduced either by decreasing the amount of time the sensor is exposed to light, or by decreasing the sensor's ability to discriminate between a fixed point and a moving point of the image. As an extreme example, suppose your sensor was made of only one single pixel. You would have to move the camera quite a bit in order for it to register a change in the light falling upon it. That is because a larger pixel simply averages the luminous flux over a proportionately larger area.
Conversely, if you make the pixels very, very small, then it takes much less displacement of the camera for them to register a change in the light falling upon them; thus the sensor becomes more sensitive to camera shake.
The result of these scenarios, however, is analogous to diffraction--higher density sensors are merely better able to observe the phenomenon that was already there. After all, the camera (theoretically) shakes equally in either high- or low-density situations. Your shooting technique is no different. It is merely that the low-density sensor is unable to resolve the effect of that shake.
Therefore, when you take a high-density image which facilitates a print with high enlargement ratio, you will be able to resolve more but in doing so you have a greater likelihood of revealing sharpness losses due to camera shake.
If you have two sensors, both of the same total area, but one with very high density and the other of low density, and if you make equal-sized prints for which the low-density sensor's print is acceptably sharp, you should not be able to observe loss of sharpness due to camera shake in the high-density sensor.
