Well, I can't shoot it at 1/320, but I can at 1/400. ;-) That's one of the reasons I'm excited about Auto ISO (I think). I can avoid inadvertently slipping below 1/400 when I'm shooting that tricky half hour just before sunset.
And I'm sitting here banging on C1 as I type. I just love it. It's simply wonderful. The shadow and highlight recovery tools are amazing. It's great. Rush right out and buy it today. ;-)
tdodd wrote:
... you need to take account of camera shake, subject motion blur and final enlargement factor. None are irrelevant.
Then all this really comes down to is the obvious point that if you intend to magnify your original capture more, you must be more careful about all factors that affect sharpness. In other words, it isn't that denser photosites require a faster shutter speed per se - it is that a larger print than you might have made previously might benefit from using a faster shutter speed and from increased care about all factors that affect sharpness.
Nill Toulme wrote:
Yes but... I think we've agreed that we will only need higher shutter speeds due to higher resolution to the extent the higher resolution is causing us (allowing us) to enlarge more. Yes? (See Dan's "thought experiment" above.")
Just to try tying up these issues. - most of these points have been covered but not all together
Blur is blur and greater enlargement means a given degree of blur at some point becomes perceivable, no matter where the blur came from. Even from a "perfect lens" the blur can come from
1) Out of focus
2) Diffraction
3) Camera shake
4) Subject motion.
Then there's
(5) +Lens imperfections)
With smaller pixels and the ability to enlarge an acquired image more, all 4 are affected by the option for increase in enlargement.
Imagine a pinpoint light source that, when perfectly focused is smaller than the smallest pixel. In an ideal situation it will register as a point (at or below the viewer's ability to resolve) in the print.
1) If it gets blurred by imperfect focus it will be bigger than it was and if it is printed large enough, the viewer will see it as "bigger" than the one that was in focus and it will not be perceived as sharp. This is how DoF is affected by enlargement. (If it's not enlarged enough it will still appear to be in focus)
2) Diffraction is always happening but at a certain aperture the effect of it becomes noticeable. The "point" becomes perceived as a disk. If that disk is enlarged enough the viewer will notice the effect of the diffraction. If not, it will appear as a point. (The print size and aperture are obviously intimately related)
3) If the camera is moving while the shutter is open and it moves far enough that this point "draws" a shape that is larger than the pixel it used to be, the image will have that information in it. If the image is not enlarged much, this shape may still be perceived as a point. However, if it is enlarged enough the viewer will notice that it is no longer a "point" and the blur will be perceived.
4) If the subject moves during the opening of the shutter (whether a small distance for someone close to camera, or a larger distance for the guy downfield) the point will be drawn into a streak/blur over some number of pixels. If this blur is not enlarged much, the image may still look like a point, but if it is enlarged enough the viewer will be able to detect that it's not a point.
If you use the extra MP to make a bigger print or to make the same size print from a deeper crop, all of these causes for blur need to be considered and the taking technique needs to keep up. Thus it becomes a choice between "using your old tried and trusted technique and getting images that show its imperfections" and "I need to improve my technique to take the greatest advantage of the technology improvements I just paid for".
By this token, a 1D is 3x more forgiving in all of these areas (including lens induced blur) than a 7D
For those who print bigger because they have the extra MP, a non-trivial number of folks I suspect (whether more MP or deeper crop)
In a 300 ppi print,
a 1D image is enlarged 7x
(a typical 35mm print at ~8x10 was about 8x)
a 5D image is enlarged 10x
a 1Ds3 ( or a 20D) image is enlarged 13x
a 7D image is enlarged ~20x.
With regard to the (related) teleconverter thread
If the image coming through e.g., a supertele is already blurred in a perceivable way (i.e. any "point", when printed, is already perceived as blurred) then the converter will just blow that blur up and ~no gain will be had. Similarly if the blur is already greater than the ~size of the pixels, more pixels won't produce a better resolution image (just a more accurate image of the blur). However, if the lens has produced an image with more resolution than the pixel size, then more pixels can capture more detail - at least until the pixels are smaller than the "point" as our reference (above) - or a converter can increase the size of the "point" until it reaches ~pixel size and still obtain more detail. (Thus a really good lens such as 300 2.8 can "take" a converter or two or three and still yield more detail, while others just don't do well - they're already at some limit)
When you combine the issues above about blur, with the challenges of keeping the "point" sharp in the first place in "telephotus extremis" like BIF, it's easy to see how some (with extremely good technique to minimize ALL sources of blur) can get the very good results one expects are possible, while many complain that "My 7D telephoto shots are sooooffftt " and "My teleconverter shot is no better than PS interpolation "
A bit like getting a new faster car - we have to improve our driving skills to get the most out of it
Any time to move to a greater magnification of the source capture you must pay increasing attention to all issues that might affect sharpness to the extent that not doing so might cause problems. (It is also worth noting that the need for "better sharpness" does not increase in a linear manner along with print size. Look at large prints close up and you'll understand.)
If your reasons for moving to, say, full frame and to, say 21MP instead of 12MP are not related to moving to even larger prints than you have made in the past then sources of blur such as camera motion, subject motion, and diffraction will have exactly the same effect in the 21MP images that they had in the 12MP images if you print at the same size and all other variables are controlled.
It is also really important to "step away from the theorizing" and make some photographs and print them. A lot of this stuff turns out to be pretty much completely academic in the real world.
So does this all explain why many of my friendly Canon sports shooters are complaining that 1/1000 shutter speed when shooting digital does not freeze the action like it did when we shot film at 1/1000. I had heard it has to do with the pixel pitch that Canon is using on the sensor as compared to the pitch of a Nikon D3.
gdanmitchell wrote:
AJSJones, to make it really simple...
Dan
Different folks learn in different ways.
All I was saying was - think about what causes blur in your image, and think about how much it's going to be enlarged when you print it. Couldn't be simpler !
jprisching wrote:
So does this all explain why many of my friendly Canon sports shooters are complaining that 1/1000 shutter speed when shooting digital does not freeze the action like it did when we shot film at 1/1000. I had heard it has to do with the pixel pitch that Canon is using on the sensor as compared to the pitch of a Nikon D3.
When phrased like that it's hard to know quite what they meant. However, I would say that when shooting action in difficult lighting it is quite possible to have a camera with more pixel density than is useful, unless you have a particular desire to faithfully record blur (and noise) with more precision than other cameras permit. But in those situations you do always have the option to downsize/merge pixels, which restores the status quo.
On the other hand, in favourable conditions, a camera with higher pixel density, and/or greater absolute pixel count, is likely to pull out detail that other cameras will simply fail to resolve.Which camera would you prefer to use - the one that give you the options, or the one that takes them away?
I have some what of a different background than a lot of you guys because the first DSLR I used was a Sigma SD10. I still have it, but am using my SD14 more than the SD10. Sigma bodies use a much different sensor than any other digital camera I know of. If you want to get up to speed on the Foveon sensor wiki is a good place to start.
The reason I moved to Canon bodies was the AF, hi ISO performance, burst rate, lens selection, and other factors allowed me to more consistently produce images with IQ acceptable to me; but I still find some uses for my Sigma bodies (mostly macro).
What does all this have to do with this thread. Well anyone who has read John Williams book "Image Clarity"
will recognize the term roll off. For those of you who missed class the day you were suppose to read this book roll off is a term from film days that describes what pixel peepers did before there were pixels, they looked at blow ups that showed the individual grains of silver halide on a print.
Back in the old days instead of fancy resolution charts with tiny numbers and fat black lines morphing into skinny black lines it was the custom to take a pix of a knife blade to determine roll off, and the sharper the transition from a black knife edge to the white background was the better the roll off was.
Relating to the current discussion a Bayer sensor (wiki is your friend on this) used by Canon, and every one else but Sigma, has an inherent problem with roll off. Because the RGB mask over the sensor prevents light from reaching about 60% of the detectors the firmware in the camera has to interpolate (a fancy word for making a good guess) what light would have reached the sensor if the mask was not there.
Currently Bayer sensored cameras taking pix of a resolution chart produce an image of the black lines on a white background with about 6-8 pixels of various shades of gray going from the black line to the white background. When the lines get very skinny the whole mess just turns into gray mush. Bringing color into the picture (nice pun) makes things even more complicated. There have been resolution charts that take advantages of both the strong points and the weak points of the RGB mask used on a Bayer sensor making the 6-8 pixel roll off change wildly.
Of course other factors can make roll off better or worse. Williams was old school (same as me) and claims that unless you are shooting a 35mm camera on a tripod with a 50mm lens at 1/250 and f8 and MLU you will not get the best possible roll off. But even if you get the best possible roll off you are still faced with the 6-8 pixel blur inherent to a Bayer sensor.
jprisching wrote:
So does this all explain why many of my friendly Canon sports shooters are complaining that 1/1000 shutter speed when shooting digital does not freeze the action like it did when we shot film at 1/1000. I had heard it has to do with the pixel pitch that Canon is using on the sensor as compared to the pitch of a Nikon D3.
I think that has a lot more to do with looking at images blown up on computer screens instead of contact sheets. ;-)
Nill Toulme wrote:
I think that has a lot more to do with looking at images blown up on computer screens instead of contact sheets. ;-)
Nill
If you take the numbers I quoted and multiply them by at least 3 you get the enlargement factors for on-screen at 100% so, yeaaa-aaahhh
1D @100% = 21x, 5D = 30x, 1DsIII= 40x and 7D = 60x enlargements
In a 300 ppi print,
a 1D image is enlarged 7x
(a typical 35mm print at ~8x10 was about 8x)
a 5D image is enlarged 10x
a 1Ds3 ( or a 20D) image is enlarged 13x
a 7D image is enlarged ~20x.
...that really left me scratching my head. I don't understand what you're saying there at all.
Sorry - (magnification is reserved to mean the size of the image on the sensor related to its size in real life, so) enlargement is from the size of the optical image falling on the sensor to the size of the print (or screen). It is the number that tells you how much the blur or imperfections have been enlarged - and therefore whether they are likely to be an issue. The prints get bigger and bigger if you fix ppi but increase MP - not always what folks do. I like to make a large a print as I can for landscapes and need as much croppability as possible for birds that don't fill the frame on my longest lens/converter combo. SO this is important to me. If you always have made and continue to make a given sized print such as 8x10 or 11x14 etc, then the MP increase means little once you've exceeded your printer's limit and has no effect on the issues you raised..
Nill Toulme wrote:
I think that has a lot more to do with looking at images blown up on computer screens instead of contact sheets. ;-)
Nill
I am not talking about viewing images on a screen. My point is that many of us who shoot Pro sports have realized that shutter speeds we once used to stop or freeze action is no longer stopping the action. For instance the bat, on a swing, or arm movement on players that we never saw before while shooting in the film days at the same shutter speeds. If there is blur in a certain part of an image from movement, like a bat swinging or arms moving a high shutter speed should stop that. many of us are finding that with digital you actually need a higher shutter speed to stop the movement than what we shot before. There was a blogger who tested the mark IV and if I remember correctly claimed that because of the pitch of the sensor that Canon is using, you needed a higher shutter speed to freeze the action than if you shot a NIkon D3. I have friends who shoot both systems and with the same shutter speed selected the D3 will have less or no movement in areas that the Canon cameras show movement in. In fact sometimes at 1/2000 of a second on canon it is still not freezing the action and you might see arm movement in a player running , and 1/2000 should easily stop that. That movement appears whether it's a 5x7 or 16x20 it's still there.
jprisching wrote:
I am not talking about viewing images on a screen. My point is that many of us who shoot Pro sports have realized that shutter speeds we once used to stop or freeze action is no longer stopping the action. For instance the bat, on a swing, or arm movement on players that we never saw before while shooting in the film days at the same shutter speeds. If there is blur in a certain part of an image from movement, like a bat swinging or arms moving a high shutter speed should stop that. many of us are finding that with digital you actually need a higher shutter speed to stop the movement than what we shot before. There was a blogger who tested the mark IV and if I remember correctly claimed that because of the pitch of the sensor that Canon is using, you needed a higher shutter speed to freeze the action than if you shot a NIkon D3. I have friends who shoot both systems and with the same shutter speed selected the D3 will have less or no movement in areas that the Canon cameras show movement in. In fact sometimes at 1/2000 of a second on canon it is still not freezing the action and you might see arm movement in a player running , and 1/2000 should easily stop that. That movement appears whether it's a 5x7 or 16x20 it's still there....Show more →
That's exactly what I'm interested in. If that's true... why? Intuitively, it makes little sense...
I haven't shot enough games at lower shutter speeds with the Mark IV yet to have a good sense of whether it is true in fact for my sort of shooting. So far, on what limited experience I have, I'm inclined to believe it's not... but that's also what I want to believe. ;-)
Nill Toulme wrote:
That's exactly what I'm interested in. If that's true... why? Intuitively, it makes little sense...
I haven't shot enough games at lower shutter speeds with the Mark IV yet to have a good sense of whether it is true in fact for my sort of shooting. So far, on what limited experience I have, I'm inclined to believe it's not... but that's also what I want to believe. ;-)
Nill
Nill, here is one of the quotes from one photographer Jens Dresling, and I have read this exact fact from others, I too would like to know why.
"However, it is difficult to determine how the camera's autofocus deal for sport. It is slightly complicated. Canon has chosen 1.3 extension and large files. Ie the pixels are very small and therefore it is recommended you use faster closing times than usual. Otherwise, there is motion blur.
On the beta I used the camera I shot JPEG only and the first football game I drove 1 / 800. I should perhaps have been running at 1 / 2000 or thereabouts. There is some motion blur in some of the images and it makes it difficult to judge accurately autofocus."
Seems to me you guys are talking about angular resolution; something astronomy guys talk about all the time. If your FOV is say 10 degrees and the bat motion goes across say 5 seconds of the 10 degree FOV during a 1/1000 second exposure that five seconds of the FOV will reflect more photons on the detectors of a Canon sensor than a Nikon sensor cuz the Canon sensor is 18mp and the Nikon sensor is 12mp. I would bet the roll off in the Canon image would be worse as well, since blur is basically poor roll off.
" and "My teleconverter shot is no better than PS interpolation 
