The angle of view is about 56 to 58 degrees clearly less than the 64 degrees expected from 18 mm on a 1.5 crop. So while I think the camera did introduce barrel distortion in both axes, I'm not sure that it would be all that dramatic when applied to an image of this field of view.Įdit: I checked the angles with The Photographers Ephemeris. I know that location (DuPont Circle in Washington DC) and I would guess the horizontal angle of view as not much more than 60 degrees, while an 18 mm on APS-C would be about 64 degrees. If the camera then produced barrel distortion by warping the image inward, the corners would have been exposed, make a further crop necessary to produce the final image. Since it was shot with a rectilinear lens at 188 on crop, it would have had an (original) field of view equivalent to about 28 mm. It is probably best to characterize this as a severe crop from the field of view that a short-focal-length fisheye would give. I tend to agree that there is slight inward curvature even of the vertical lines, though they do not appear to have the increasingly greater effect toward the edges that a fish would have (Gerry's impressive and imaginative presentation notwithstanding he applied the already-curved horizontal foreground to the verticals). That gives the image the impression of a much greater fisheye effect than it actually has. This is a bit of a misleading image to apply the effect to, since the entire lower half of the pictures is made up of a circular plaza that would exhibit similar curvature even in a fully rectilinear projection. The verticals on the buildings in the background are just as curved as the horizontal lines. I suspect the camera was applying a cylindrical warping algorithm to the rectilinear image projected by the lens, which produces some of the curves that suggest a fisheye, without truly reproducing the fisheye effect. This is typical of what you would get if you shot a panorama and used a cylindrical projection the most commonly used. Although it isn't the same type of subject, compare with the fisheye posted by BobSC. In a fisheye there would be barrel distortion in both axes the vertical lines would curve as well, more so the further from the center line. The fact that the horizontals above and below the horizon are curved, but the verticals (even those away from the center line) are essentially still straight up and down suggests that this is more akin to a cylindrical projection than to a true fisheye. Can a knowledgeable person tell that this picture below uses an in-camera effect and was not taken with a fisheye lens? What's different? (Note that I added the vignette - it's not a product of the in-camera mode.) Stereo rectification for fisheye camera model.I've never used a fisheye lens or explored fisheye photography but I tried the "fisheye effect" recently on my SL2. Termination criteria for the iterative optimization algorithm. fisheye::CALIB_FIX_K1., fisheye::CALIB_FIX_K4 Selected distortion coefficients are set to zeros and stay zero.fisheye::CALIB_FIX_SKEW Skew coefficient (alpha) is set to zero and stay zero.fisheye::CALIB_CHECK_COND The functions will check validity of condition number.fisheye::CALIB_RECOMPUTE_EXTRINSIC Extrinsic will be recomputed after each iteration of intrinsic optimization.Otherwise, (cx, cy) is initially set to the image center (imageSize is used), and focal distances are computed in a least-squares fashion. fisheye::CALIB_USE_INTRINSIC_GUESS K1, K2 contains valid initial values of fx, fy, cx, cy that are optimized further.fisheye::CALIB_FIX_INTRINSIC Fix K1, K2? and D1, D2? so that only R, T matrices are estimated.Output translation vector between the coordinate systems of the cameras.ĭifferent flags that may be zero or a combination of the following values:
Output rotation matrix between the 1st and the 2nd camera coordinate systems. Size of the image used only to initialize camera intrinsic matrix. Input/output lens distortion coefficients for the second camera. Input/output second camera intrinsic matrix. Input/output vector of distortion coefficients \(\distcoeffsfisheye\) of 4 elements.
If any of fisheye::CALIB_USE_INTRINSIC_GUESS, fisheye::CALIB_FIX_INTRINSIC are specified, some or all of the matrix components must be initialized. The methods in this namespace use a so-called fisheye camera model.