Using lens correction data

#1

Hi,

has anyone used the lens correction data that is supplied for most modern DSLR camera lenses to be used in the LensDistortion node in Nuke?
I imagine, as the data comes directly from the makers, it would be superior when compared to any visually based undistortion method.
Has anyone tinkered with that idea? I know, the data is not readily available as a file (or is it?), but only accessible in an indirect way through whatever raw processing utility you are using (like the Eos Utility software or Adobe Lightroom). One would have to find a way of extracting that information and then convert/apply it to what the LensDistortion node needs.
But still, it sounds tempting to me :slight_smile:

#2

dont know it its fits but the lens correction data from adobe products is stored in XML files

e.g.: “C:\ProgramData\Adobe\CameraRaw\LensProfiles”

in the XML you will find a section:

<stCamera:ResidualMeanError>0.000176</stCamera:ResidualMeanError>
<stCamera:ResidualStandardDeviation>0.000321</stCamera:ResidualStandardDeviation>
<stCamera:RadialDistortParam1>-0.036734</stCamera:RadialDistortParam1>
<stCamera:RadialDistortParam2>0.004499</stCamera:RadialDistortParam2>
<stCamera:RadialDistortParam3>-0.000356</stCamera:RadialDistortParam3>

maybe this fits the 2 distortion parameters nuke wants

#3

Interesting find!
These may very well be the values we need.
I dug around a litte in those files and saw that not only are values given for different focal lengths, but also for different focus distances and apertures for the same focal length!
I didn’t know that distortion also is influenced by these values.

Also, i found this disturbing image:
http://toothwalker.org/optics/distortion/vw05112.jpg

Which shows that distortion can be different depending on the objects distance from the lens. Yikes!

#4

The manufacturer values don’t match up to the values in the node. Even if they did the manufacturers have tolerance levels depending the quality of the lens. So there is an allowable error threshold. You should always shoot grids. Even supposed “matched” lenses for stereo still don’t 100% match.

#5

So you’re saying the manufacturers values are in different units or something?
But if you have no other means of getting the distortion, i guess those would be better than nothing, wouldn’t they? If one could convert them, that is.
I am aware of tolerances when the lenses actually get produced, i am just wondering how much influence that has. I have to make some tests with the eos that i have at home. Just don’t have access to it right now.