Need sample RAW image taken with a lens fully supported by LensFun

Carmelo_DrRaw · January 14, 2017, 3:15pm

I would like to complete the LensFun integration in photoflow, adding vignetting and CA correction.

However, none of my lenses is fully supported by LensFun, hence i have no RAW sample to check the results.

Does anyone happen to have a wide-angle lens, with visible distortion/vignetting/CA, and which is fully supported by LensFun? If yes, would it be possible to have a sample RAW file that I could use to test the lens corrections (ideally some architectural shot with straight lines or something equivalent)?

Thanks in advance!

james · January 14, 2017, 4:50pm

I haven’t AFMA this lens yet, so I hope focus doesn’t matter
IMG_6998.CR2 (20.0 MB)
Realised under CC BY-NC-SA

Edit: Sigma 18-35 F1.8 DC HSM A

Carmelo_DrRaw · January 14, 2017, 5:26pm

@james Thanks! Looks fine… only TCA is almost invisible on this lens, it must be too good

paperdigits · January 14, 2017, 5:42pm

I’m sure the kit Yi lens has horrible distortion, but I haven’t done the lenafun calculations yet

Carmelo_DrRaw · January 14, 2017, 7:28pm

In fact, I am more interested in strong TCA and vignetting… I have a cheap kit lens with strong distortion, but lens fun has no tca and vignetting parameters for it…

james · January 14, 2017, 7:36pm

My subject matter probably wasn’t great to entice it either. I can try and bring it out with some high contrast edges. I’ll see what I can get tomorrow, but I think it is very well controlled on this lens. I have the ef-s 17-85 with very strong CA, but unfortunately TCA is not supported by lensfun for this model.

assaft · January 15, 2017, 11:44am

Some files here with the same lens (Panasonic 20mm / F1.7) on an Olympus E-M10.

LensCorrection.zip (68.0 MB)

The first image is for verifying vignetting and distortion correction, the others are for CA (see the air conditioner area on the right hand side; I focused at different distances to get purple fringing in the near out-of-focus area and green fringing in the distance).

The lens has a field of view of a 40mm lens on a full frame camera, so it’s not wide, but the distortion is still quite evident. It is a lens that is meant to be corrected by software.

In the ZIP file there are also developments of each RAW file using the Olympus software, and RawTherapee. The Olympus software corrects only distortion. In RT is used to Neutral profile.

james · January 15, 2017, 12:10pm

@Carmelo_DrRaw According to lenstip the worst case for my lens is @24mm f/8 but even with these settings it is very mild, so I don’t think it will help for TCA testing. Let me know if you would like the sample anyway.

assaft · January 15, 2017, 4:10pm

I’m adding a pair of shots in this zip file:

Vignetting.zip (27.8 MB)

They should be helpful for checking that Vignetting is corrected properly. The F4 shot can be used as a reference for how the F1.7 shot should look like after correction.

t3mujin · January 15, 2017, 4:34pm

Let me know if this can help:
Nikkor 16-35mm f4, no vignettting correction but wide and extreme lines
Lisboa-201310659.NEF (35.2 MB)

Fuji 18-55mm, not as wide but shot at widest length, full correction support
CiudadRodrigo_201634457.zip (22.8 MB)

Carmelo_DrRaw · January 15, 2017, 5:51pm

Dear all,

thanks a lot for all the samples, they have been already quite useful for fixing part of the PhF code…

I think I have enough test images for the moment… in case I need something more specific, I will drop a message here again.

Thanks!!

Eneen · January 17, 2017, 10:01pm

@Carmelo_DrRaw: maybe it will be useful for you: you can convert Adobe profiles from DNG converter to lensfun using that script:

github.com

lensfun/lensfun/blob/master/apps/lensfun-convert-lcp

#!/usr/bin/env python3
# -*- coding: utf-8 -*-

"""This program reads Adobe LCP files and converts their content to one Lensfun
XML file.  It was tested against the LCP files shipped with “Adobe DNG
Converter 9.0” although it should work with other sets of LCP files, too.

This program assumes that one LCP file contains the data of exactly one lens.

It writes a single XML output file, by default into the personal DB directory.
While this directory has highest priority, note that other files in this
directory – albeit highly unlikely – might contain entries that override data
in that XML file.
"""

import os, argparse, sys, re, copy, glob, multiprocessing
from xml.etree import ElementTree


parser = argparse.ArgumentParser(description="Convert LCP files to a Lensfun XML file.")

This file has been truncated. show original

Carmelo_DrRaw · January 18, 2017, 8:02am

Thanks Alex! In fact, I discovered this a few days ago, and this was the main motivation for updating and completing the LensFun support in photoflow.

Some updates:

photoflow is now bundling the git HEAD version of LensFun, which includes support for the Adobe lens corrections model. It will therefore be possible to convert Adobe lens profiles to the LensFun format, and include them in the custom user database
I have added full support for vignetting/aberrations/distortions corrections, and introduced this corrections directly into the RAW processing module.

There are two main reasons for including the lensfun tool inside the RAW processing module:

the vignetting correction requires RGB data in linear encoding, so it is best applied on the linear RGB data in the camera colorspace, and before the conversion to the working colorspace
the RAW processing module also provides an auto CA correction, which can be used instead of the one provided by LensFun

Therefore, for the CA correction I have introduced an option that uses the LensFun-provided method with higher priority, and fall back to the automatic method if LensFun does not provide CA correction coefficients.

Here is how the new interface looks like:

Linux packages (Ubuntu + AppImage) will soon be available, and I am working to provide updated packages for OSX and Windows as well…

Any feedback will be very much appreciated.

Eneen · January 18, 2017, 3:03pm

Thank you @Carmelo_DrRaw!
I was hoping that lensfun could be added to G’MIC so all apps using it would benefit.
Of course vignetting correction would not work correctly there unless camera response curve is used.

I even wrote email to developers of krita, gmic, gimplensfun but unfortunately get no response.