It should be possible to calculate the ptlens profile that best
approximates the Adobe information. I might be able to do that, but no
guarantees as to when.
I'm not sure why acm would not be supported if you've compiled the software
from git, unless someone added it to the docs but never implemented it.
I'll check this out and find out why it looks like should work but doesn't.
Thanks,
Graeme
On Tue, 23 Nov 2021, 20:54 Gunar Kachel, <gunar.kac...@gmail.com> wrote:
> Many thanks for your feedback!
>
> I was hoping for a straight forward implementation in lensfun using the
> distortion model "acm".
> However, the compiled lensfun version 0.3.95 (from git) doesn't seem to be
> working with darktable ("Bad attribute value `model=acm' for element
> <distortion>!").
>
> Unfortunatelly I'm a noob when it come to tranformation calculation (to
> distortion models which are supported in lensfun 0.3.2).
> I've the feeling to be at a dead end for now...
>
> Gunar
>
>
> Am So., 21. Nov. 2021 um 23:46 Uhr schrieb Graeme Leese <gnl...@gmail.com
> >:
>
>> Not sure if it helps, but the parameters for the 3 planes are so similar
>>> so that I simply would take the mean values of each parameter and go with
>>> it.
>>>
>>
>> I agree that this is definitely the place to start.
>>
>> The vendor of the lens also had to fit the model to the lens - the
>>> deviations in each of the 3 planes are compatible with typical small
>>> fitting errors. Parameters are equal to the 3rd or 4th significant digit
>>> which is very precise.
>>>
>>
>> You may well be right, but remember that the full width of the image is
>> 1.0, while a chromatic aberration is typically on the order of 1-2px, so I
>> would expect to find very similar values here. For example, the first
>> random lens that I've looked up in the lensfun db has tca parameters around
>> 10^-4.
>>
>> Thanks,
>> Graeme
>>
>>
>>
>>> On 21 Nov 2021, at 19:18, Graeme Leese <gnl...@gmail.com> wrote:
>>>
>>>
>>> Hi,
>>>
>>> I know basically nothing about DNG, but Googling suggests that yes, the
>>> planes are probably referring the different color channels. There's a tag
>>> CFAPlaneColor that sets which order the colors map to the planes or else
>>> they're RGB.
>>>
>>> Chromatic aberration is essentially when the different colours are
>>> distorted differently, so my first guess is that you should calculate the
>>> distortion from the average of the planes. Then the difference between the
>>> planes should be encoded in the TCA parameters. Alternatively, since the
>>> TCA parameters are given for R and B, perhaps it would work better to take
>>> the distortion only from the G plane, I'm not sure.
>>>
>>> I haven't tried to work out how the TCA parameters would be derived. If
>>> the method looks like it's working then I can probably have a look and see
>>> if the calculation is simple enough, or else it should be simple enough to
>>> write a script to find the values that match the warp from the dng file.
>>>
>>> I don't think lensfun encodes the optical centre from the warp (so it
>>> would have to be set to 0.5, 0.5). I'm not sure how much that would affect
>>> the accuracy of the data. It's also worth noting that the DNG file uses
>>> different coordinates for pixels (0,0 in the corner) than lensfun (0,0 in
>>> the centre).
>>>
>>> Hope that helps. I'm interested to know how you get on with converting
>>> the information.
>>>
>>> Thanks,
>>> Graeme
>>>
>>>
>>>
>>> On Wed, 17 Nov 2021, 13:05 Gunar Kachel, <gunar.kac...@gmail.com> wrote:
>>>
>>>> Hi, my lens is not yet supported by lensfun. I therefore searched for a
>>>> method to to get the right distortion and chromatic aberration profile to
>>>> be used in daktable. (And I was not very successful applying the "manual"
>>>> method using Hugin.)
>>>>
>>>> However, as this is a modern lens the profiles are already integrated
>>>> in the RAW file and I therefore followed this procedure:
>>>>
>>>> 1. convert the RAW file (NEF) to DNG using Adobe DNG Converter
>>>> 2. extract OpCoodeList3 parameters from DNG file using
>>>> dng-validate.exe (at the bottom)
>>>>
>>>>
>>>> I assume I can implement these values using the lensfun distortion
>>>> profile model "acm".
>>>> However I'm puzzled that values are provided in several "planes".
>>>> Are the planes referring to different color channels?
>>>> How to implement the planes in lensfun?
>>>> How can I cover chromatic aberration using these information?
>>>>
>>>> Thanks in advance for the insight!
>>>>
>>>> Sample output from dng-validate.exe -v:
>>>> >>>>>>>>>>>>>>>>
>>>> Opcode: WarpRectilinear, minVersion = 1.3.0.0, flags = 0
>>>> Planes: 3
>>>> Optical center:
>>>> h = 0.501405
>>>> v = 0.504913
>>>> Plane 0:
>>>> Radial params: 1.025498, -0.124349, 0.058447, -0.003755
>>>> Tangential params: 0.000000, 0.000000
>>>> Plane 1:
>>>> Radial params: 1.025181, -0.124368, 0.058438, -0.003759
>>>> Tangential params: 0.000000, 0.000000
>>>> Plane 2:
>>>> Radial params: 1.025277, -0.124355, 0.058413, -0.003751
>>>> Tangential params: 0.000000, 0.000000
>>>> <<<<<<<<<<<<<<<<<
>>>> _______________________________________________
>>>> Lensfun-users mailing list
>>>> Lensfun-users@lists.sourceforge.net
>>>> https://lists.sourceforge.net/lists/listinfo/lensfun-users
>>>>
>>> _______________________________________________
>>> Lensfun-users mailing list
>>> Lensfun-users@lists.sourceforge.net
>>> https://lists.sourceforge.net/lists/listinfo/lensfun-users
>>>
>>>
_______________________________________________
Lensfun-users mailing list
Lensfun-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/lensfun-users
