The Holding Tank

Home Page Photography

Photography RSS 2.0 Feed

| Photos & Themes RSS | Hardware | Technical | Links | Books and Magazines  | Downloads |

Technical

| Flash Tests | RAW Converter Tests | Depth of Field, Image Quality and Sensor Size |

| MTF | Resolution | Cropping and Printing | Image Stabilization and Magnification |

| Lens Image Simulation | Anti-Alias Filter and Sharpening | | Tilt and Shift Lenses |

Lens Distortion Correction and Image Detail | Digital Darkroom Workflow | Optical Coatings |

Lens Distortion Correction and Image Detail

  Introduction
Pattern Method
Pattern Results
Pattern Conclusions
ISO SRF Analysis
Crops Compared

ISO Test With an Ultra Wide
ISO SRF Conclusions

Introduction

Photoshop ACR and some other programmes provide features to correct for various lens distortions such as linear distortion and chromatic aberration. Correction of barrel or pincushion distortion involves stretching or compressing the image at the centre of the frame. Of necessity these algorithms must interpolate tonal values between pixels to achieve this. So the aim of this study is to establish the effect of such correction on fine image detail.

Pattern Method

The methodology was to computer generate a set of test images of 3500 X 2300 pixels as 16-bit RGB TIF files, with a checkerboard pattern of 1, 2, 5, 10, 20 and 50 pixel pitch of alternating 8% and 92% grey.  An example of the input test pattern with a 2 pixel grid is downloadable here, so readers can experiment themselves.

Each pattern was subjected to barrel and pincushion correction using Photoshop CS2 to the degree required for the popular EF 17-40mm f4L lens at the focal lengths 17, 20 and 40mm.

The degree of correction was determined from linear distortion test images for full frame from William Castleman’s lens test. The degree of Photoshop lens correction for barrel and pincushion distortion is independent of image size. These are relatively small amounts for distortion. An downsampled example of the 20 pixel full frame grid after 17mm distortion correction is shown below.

To determine the effect of the correction a 100 X 100 pixel crop was taken from the centre, edge and corner of each image and saved as a JPG file, these are shown below in the Results section.

Pattern Results

You may need to view the images at 400-800% to clearly see the effect, ignore the JPG artefacts and concentrate on the areas of mid grey where the correction algorithm has had to interpolate the pixel luminance.

Image Centre

Image Edge

Image Corner

Pattern Conclusions

The first surprise is that even for small amounts of correction the whole of the image is affected, not just the edge and corners. In fact the image centre can be as bad or worse than the image periphery. The table below relates the pixel spacing to the equivalent line pairs per millimetre of detail resolution.

The most significant image detail is in the 5-10 lp/mm range with fine detail around 30 lp/mm. This suggests that the 5 and 10 pixel spacing test patterns will show the effect for main image contrast and the 1 and 2 pixel spacing the effect on fine image detail. A 400% example from the 17mm 2 pixel squares  centre of image is shown below

Even at a minimal adjustment of 1 the distortion correction causes considerable softening of the 10 pixel spacing detail. For finer detail the contrast is almost completely destroyed over about 50% of the area at periodic intervals, the periodicy depending on the degree of correction and location in the image. In fact the interpolation is causing a form of aliasing because of the regular test pattern.

So does sharpening help? Yes to an extent. Below is the same 400% crop after applying USM of 300% 0.7 0.0.

The image size used here was chosen for convenience and represents a 8Mp image. It seem likely that as the number of megapixels increases the loss of detail caused by using this correction will reduce. However, it should still be significant even on the 16Mp 1Ds MkII.

ISO SRF Analysis

ISO measurements were made of the very sharp EF 50mm f1.4 at f1.4 and f5.6. The +4 distortion correction was applied to a copy of the test image and remeasured. This shows approximately a 10-20% loss of MTF 50% frequency due to the distortion correction, even at f1.4 where the image is understandably slightly soft in the corners there was a noticeable effect. 

However using the chromatic aberration correction features does not have any detrimental effect.

The following are links to comparative detail and crops at the two fstops examined:

Uncorrected f1.4        Uncorrected f5.6

Corrected f1.4        Corrected f5.6

Centre and Corner crops side by side are here.

The distortion correction summary results are below.  

Crops Compared

Centre

Corner

ISO Test With an Ultra Wide

Here the ISO test was repeated with a EF-S 10-22mm at 10mm f8. A photoshop correction of +3 seemed appropriate.

The resulting MTF 50% across the lens and centre 100% crops are shown side by side in the table below.

ISO SRF Conclusions

The SRF analysis shows quite a dramatic loss of sharpness. It should be noted that some of this may be systematic due to introducing a small curvature in the ISO measurement edge, although this will be very small over the few pixels of the measurement rectangle; at worst the analysis provides an upper estimate for the sharpness loss.

However examination of the 100% crops in the centre and corner shows a definite softening in the stopped down case. This would be difficult to discern in the wide open case, however experience shows that the SRF analysis is much more sensitive than the visual examination.

Further checks with an actual ultra wide where the barrel correction should be providing an improvement of the ISO line straightness shows a much lower loss of sharpness. About 5% in the centre and 10% in the corners. This is a much lower level and not easy to spot in the 100% crops.

Would this show in the real world photo? Probably, but for me the conclusion is do not routinely apply distortion correction unless the photo needs it, ie for seascapes or architectural shots where straight lines abound.

Last Updated 05/01/2009

All Content © 2005-09 Lester Wareham All Rights Reserved     

All material is supplied  as is and without warranty, use at your own risk.  

All opinions stated are the authors own.

All quoted information remains the copyright of the respective authors.

Contact: (anti spam remove underscores)

Note this email gets spammed a lot so if you want to get your message through please prefix "Holding Tank" in the title

f_ w_ a_ r_ e_ 9_ 0_ @_l_i_n_e_o_n_e_._n_e_t