1. A number of people have contacted me to ask how much lens correction there is on the new Leica SL. Or, more accurately, how much lens correction there is on the new SL lens, the VARIO-ELMARIT 1:2.8-4.0/24-90mm.

    Below are the comparisons for an image shot at 24mm - the uncorrected image is from AccuRaw. (The original image is courtesy of PhotographyBlog).

    I'm sure you won't need me to tell you which is which(!).






    For those interested in the gory details, here is the content of the correction:
    Opcode: WarpRectilinear, minVersion = 1.3.0.0, flags = 0
    Planes: 3
      Optical center:
        h = 0.500000
        v = 0.500000
      Plane 0:
        Radial params:     0.999557, -0.133087, 0.061558, -0.010640
        Tangential params: 0.000000, 0.000000
      Plane 1:
        Radial params:     0.999276, -0.132477, 0.060577, -0.010234
        Tangential params: 0.000000, 0.000000
      Plane 2:
        Radial params:     0.999683, -0.131579, 0.058723, -0.009416
        Tangential params: 0.000000, 0.000000
    5

    View comments

  2. (Edited 23 October to include information on how M lenses are handled.)

    As usual when new Leica cameras come out, I took a quick look inside a DNG from one of Leica's new Leica SL (Typ 601) cameras, using one of the new Leica L mount lenses:
    1. The camera is using what appears to be production software - version 1.10.
    2. The camera name shows as "LEICA SL (Typ 601)"
    3. The image data is 14-bit. There is no compression used in the DNG I looked at. 
    4. The DNG version is 1.4, with a "backward version" of 1.3. There is a reason for this - DNG 1.3 allows for opcodes, which Leica use for lens correction.
    5. In the DNG I looked at, which was shot with a "VARIO-ELMARIT 1:2.8-4.0/24-90mm ASPH. OIS" lens, lens correction is done by a single "WarpRectilinear" operation in the DNG. 
    6. In addition to the lens correction op code, there is also a "FixBadPixelsConstant" opcode, whose function is exactly as the name states. This is the same as in the Leica Q.
    Something to note is that the sensor appears built on identical technology to the Leica Q sensor. The color matrixes, black levels, etc are all identical.

    Thanks to Jonathan Slack, I now also have an example shot with some coded M lenses, including the WATE (the Tri-Elmar-M 16-18-21), as well as Leica 28 and 35 mm fixed focal length lenses. The answers are interesting:

    With the M lenses installed, the WarpRectilinear opcode still exists in the DNG file, as does the FixBadPixelsConstant opcode. However, there are no additional opcodes.

    The WarpRectilinear code  effectively does nothing, as the parameters are set to zero, so M lenses are not corrected for geometry, which is what you would expect.

    But the absence of additional opcodes implies that whatever corrections for vignetting/"red edge"/"Italian flag syndrome" are being made by modifying the actual raw data, as is the case on M-series cameras. That might seem surprising, as if you're using DNG 1.3 style opcodes, why not just use opcodes to correct, rather than messing with the raw data? However, this doesn't really surprise me. Without going into the technicalities of what the opcodes in the DNG specification can and can't do, correcting for color aberrations as complex as those produced by M series lenses by using opcodes was always going to be challenging. Possible, but not easy. So I would think that Leica have simply chosen to transfer the existing raw data modification algorithm from the M-series cameras to the new camera. This makes good sense - Leica users that have been around a while will recall that even on the M-series cameras, getting the corrections optimized took a few iterations. Changing to opcode correction would effectively reset that to zero.
    2

    View comments

Popular Posts
Blog Archive
About Me
About Me
My Photo
Author of AccuRaw, PhotoRaw, CornerFix, pcdMagic, pcdtojpeg, dcpTool, WinDat Opener and occasional photographer....
Loading