lensfunpy is an easy-to-use Python wrapper for the lensfun library.
How to find cameras and lenses:
import lensfunpy
cam_maker = 'NIKON CORPORATION'
cam_model = 'NIKON D3S'
lens_maker = 'Nikon'
lens_model = 'Nikkor 28mm f/2.8D AF'
db = lensfunpy.Database()
cam = db.find_cameras(cam_maker, cam_model)[0]
lens = db.find_lenses(cam, lens_maker, lens_model)[0]
print(cam)
# Camera(Maker: NIKON CORPORATION; Model: NIKON D3S; Variant: ;
# Mount: Nikon F AF; Crop Factor: 1.0; Score: 0)
print(lens)
# Lens(Maker: Nikon; Model: Nikkor 28mm f/2.8D AF; Type: RECTILINEAR;
# Focal: 28.0-28.0; Aperture: 2.79999995232-2.79999995232;
# Crop factor: 1.0; Score: 110) How to correct lens distortion:
import cv2 # OpenCV library
focal_length = 28.0
aperture = 1.4
distance = 10
image_path = '/path/to/image.tiff'
undistorted_image_path = '/path/to/image_undist.tiff'
im = cv2.imread(image_path)
height, width = im.shape[0], im.shape[1]
mod = lensfunpy.Modifier(lens, cam.crop_factor, width, height)
mod.initialize(focal_length, aperture, distance)
undist_coords = mod.apply_geometry_distortion()
im_undistorted = cv2.remap(im, undist_coords, None, cv2.INTER_LANCZOS4)
cv2.imwrite(undistorted_image_path, im_undistorted)It is also possible to apply the correction via SciPy instead of OpenCV. The lensfunpy.util module contains convenience functions for RGB images which handle both OpenCV and SciPy.
How to correct lens vignetting:
import lensfunpy
import imageio
db = lensfun.Database()
cam = db.find_cameras('NIKON CORPORATION', 'NIKON D3S')[0]
lens = db.find_lenses(cam, 'Nikon', 'Nikkor AF 20mm f/2.8D')[0]
img = imageio.imread('/path/to/image.tiff')
focal_length = 20
aperture = 4
distance = 10
width = img.shape[1]
height = img.shape[0]
mod = lensfunpy.Modifier(lens, cam.crop_factor, width, height)
mod.initialize(focal_length, aperture, distance)
did_apply = mod.apply_color_modification(img)
if did_apply:
imageio.imwrite('corrected.tiff', img)
else:
print('vignetting not corrected, calibration data missing?')Install lensfunpy by running: `sh pip install lensfunpy`
64-bit binary wheels are provided for Linux, macOS, and Windows.
If you have the need to use a specific lensfun version or you can't use the provided binary wheels then follow the steps in this section to build lensfunpy from source.
First, install the lensfun library on your system.
On Ubuntu, you can get (an outdated) version with:
Or install the latest developer version from the Git repository:
After that, install lensfunpy using:
On Linux, if you get the error "ImportError: liblensfun.so.0: cannot open shared object file: No such file or directory" when trying to use lensfunpy, then do the following:
The lensfun library is installed in /usr/local/lib when compiled from source, and apparently this folder is not searched for libraries by default in some Linux distributions. Note that on some systems the installation path may be slightly different, such as /usr/local/lib/x86_64-linux-gnu or /usr/local/lib64.
These instructions are experimental and support is not provided for them. Typically, there should be no need to build manually since wheels are hosted on PyPI.
You need to have Visual Studio installed to build lensfunpy.
In a PowerShell window:
The above will download all build dependencies (including a Python installation) and is fully configured through the four environment variables. Set USE_CONDA = '0' to build within an existing Python environment.
lensfunpy depends on NumPy. The minimum supported NumPy version depends on your Python version:
| Python | numpy |
|---|---|
| 3.7 | >= 1.14 |
| 3.8 | >= 1.17 |
| 3.9 | >= 1.19 |
| 3.10 | >= 1.21 |
| 3.11 | >= 1.23 |