Luminous Landscape Forum
Equipment & Techniques => Digital Cameras & Shooting Techniques => Topic started by: Roger_Cavanagh on November 28, 2002, 10:39:26 am
-
[font color=\'#000000\']Tim,
I believe that the CoC has to be reduced for the D30 because the smaller sensor size means greater enlargement to achieve a 10x8 compared to 35mm film.
http://dfleming.ameranet.com/digital_coc.html (http://dfleming.ameranet.com/digital_coc.html) This page shows a CoC for the D30/60 of 0.019mm down from the 0.030 for standard 35mm. There's a bunch of other stuff about DOF on the site too.
I agree with you that Nick Rains' article gave a misleading impression.
Regards,[/font]
-
[font color=\'#000000\']The dfleming link gives a .019 COC based on 1.6 of .030. My recollection is that the .030 number is from calculations/measurements done in the 1920's or 30's. Are folks satisfied that something between .030 and .025 is still appropriate? Also, what is the assumption regarding print size behind the .030 number? I guess it's not controversial that the larger the print, the smaller the coc needs to be (assuming the same viewing distance?) But what ranges make sense?
see: http://www.trenholm.org/hmmerk/SHBG01.pdf (http://www.trenholm.org/hmmerk/SHBG01.pdf) and the follow-up 2, 3 and 4 pdf's for a totally different approach to DOF - I'll have to read it 2 or 3 more times before I start to get a glimmer.....
Gosh, I guess understanding DOF really is simple when you think about it.... not.[/font]
-
[font color=\'#000000\']Mhhhhnnn. Goof, factor of ten out on my above post. 0.3 mm CoC in an 8x10" print translates to 0.02mm on a D30/D60 sensor, etc[/font]
-
[font color=\'#000000\']CoC really is purely optical. Any edge narrower than ~3 pixels when projected onto the imager will be captured as a sharp edge. Once captured 3 pixels wide, it can be stretched to whatever length is wanted, and remains 3 pixels. Optical enlargements behave differently. CoC is useful for calculating DoF though, since that's in the optical domain. Scanning film OTOH magnifies images optically, so the higher the scan magnification, the softer edges become. If you scan for the final print size, you've effectively made an optical enlargement.[/font]
-
[font color=\'#000000\']What circle of confusion is reasonable on a D30 for 8x10 prints? The article seems to imply that the 1.6 crop isn't a factor that needs to be considered in setting the COC. So far I've been using .030 but I've seen lots of discussion to the effect that the crop factor needs to be applied to the COC. Just playing with the math - you get the same result if you adjust the flocal length by the square root of 1.6 as you do if you adjust the COC by 1.6.[/font]
-
[font color=\'#000000\']Maybe the article should be titled "Circle of Confusion". :D[/font]
-
[font color=\'#000000\']There's lots of software out there for calculating the Depth of Field for a given CoC, but indeed the head of the nail is "What CoC do YOU accept in a given sized print?" To my eye a CoC of 0.03 mm in a hand held 8x12" glossy print is rather generous (perhaps OK in a print on rough lustre paper?). That's fuzz a third of a mm across! Or just 1.5 lp/mm! Generally accepted "Tack Sharp" is 4 to 5 lp/mm. If we use 4 lp/mm that's 0.0125 CoC on the print. A D30/D60 frame has to be enlarged 13x to fill an 8x12" print. So on the sensor that makes a CoC of 0.001, which just happens to be about the pixel width in a D30. So even if one wishes to chase larger, sharper prints, it's pretty pointless to set an even smaller CoC in a D30, but viable in a D60. Reality of course forces us to compromise and accept a larger CoC. Personally I think 0.03 in an 8x12" print is sort of OK when we are desperate for maximum DoF (on lustre paper) but should not be used when we can afford to be tighter. Strictly speaking, of course, DoF is nil if one demands maximum lens resolution. The lens is focused at one distance, all other distances are at decreasing defocus...[/font]
-
[font color=\'#000000\']I think Nick Rains has made a fair attempt at explaining the issue. Can I just ask one question. If I use a standard 50mm lens, designed for full frame 35mm, with the D60 at a specific aperture, say F8, then that would be equivalent to using an 80mm lens with full frame 35mm at the exact same distance from the subject. Understood. But what should the f stop setting of the 80mm lens be to get the same DOF and the same sized circles of confusion?
To put the question another way, if Canon were to design a standard lens specifically for the D60 with an appropriately smaller image circle, it would be described as a 31mm lens (probably 30mm), but it would be different from the 31mm lens designed for the 35mm format that people are currently using. As a consequence of the smaller image circle the lens would project, the F stop markings on the lens would be different, would they not?[/font]
-
[font color=\'#000000\']Here is my attempt to answer Ray, who asked:
>If I use a standard 50mm lens, designed for full frame 35mm, with >the D60 at a specific aperture, say F8, then that would be equivalent >to using an 80mm lens with full frame 35mm at the exact same >distance from the subject. Understood. But what should the f stop >setting of the 80mm lens be to get the same DOF and the same sized >circles of confusion?
To get the same DOF, meaning the same COC size on prints of the same size, for different camera formats (image sizes) and lenses giving the same angular field of view (so focal length proportional to image size):
a) Use the same aperture DIAMETER (not aperture ratio)
or equivalently
Adjust the f-stop in proportion to the camera image size or "focal length factor".
So in your case
D60 format with f=50, aperture f/8 = 6.25mm
is equivalent to
35mm format, f= 80mm (=1.6x50mm) f-stop of f/13 since 1.6x8=13, which still gives aperture diameter of 80/13 mm = 6.25mm =50/8 mm (all calculations approximate!)
Why?
The underlying optical fact (simple enough, but not so often mentioned in camera literature) is that:
To get a COC whose diameter is the same fraction of the image (diagonal) length, and so enlarges to the same size COC on prints of the same (diagonal) size, when recording images of the same angular field of view with camera of various formats, an aperture opening of the same diameter should be used.
Since focal length then must be varied in proportion to the diagonal length of the image size of the camera, the f-stop (ratio of focal length to aperture diameter) must vary in proportion to focal length and so in proportion to image diagonal length. I could dig up the actual COC diameter formula if anyone desparately wants to see it.[/font]