Image softness is both a function of the camera sensor and more importantly of the lens used. The image of a single point can vary in size. The larger the spot (the more pixels it covers), the blurrier the image looks.
The result is detail loss, and lack of microcontrast.
Why does it matter?
1. Detail loss
As illustrated in this image, the lens has 'scrambled' some of the details, which are no longer visible, though they are still present in the image.
Photo: Philippe Tarbouriech
2. Lack of microcontrast
Blur also affects sharp edges, making the transitions from dark to bright less abrupt than they should be. To the human visual system, the impression of sharpness is greatest if there are steep transitions between areas of different brightness.
3. Even sharpness responds to postprocessing better
Filters applied for creative image enhancement or to tune images for a particular output will perform much better on images with uniform sharpness. Unfortunately, as explained below, the level of blur created by the lens can vary a great deal depending on the position of the focal point in the image field. 4. Image interpretation bias
Limiting depth of field helps direct the viewer to the focused area. Similarly, when the sharpness of the lens is higher in the center, a viewer tend to lose interest in an object located near the edges.
How complex is it to correct?
1. Capture settings
Image softness is a function of the lens and the camera, and varies with capture settings such as aperture, focal length, focusing distance. It is common to refer to the “soft spot” of a lens, which paradoxically refers in fact to that range of settings where it produces the sharpest images.
Photo: Philippe Tarbouriech
2. Softness variation accross image field
Sharpness is one characteristic that can vary a great deal across the image frame, generally being highest in the center. For each lens (and for every possible setting), the de-blurring needs to account for the sharpness fall off.
Image corner detail from a Canon EOS 1Ds -
Canon EF 16-35 mm f/2.8L USM:
softness increases significantly towards edge of image
Blur variation across the image field
for Canon ES 50 mm f/1.4 USM on Canon 10D, sharpen off
3. Astigmatism
Not only can the level of softness vary across the image field, but the direction of the blur itself can vary.
At the center of the image, the degree of softness is usually the same in both vertical and horizontal directions. But towards the image edges, softness may be worse in one particular direction, a phenomenon known as astigmatism. This type of softness cannot be corrected with standard tools on the market.
Arrow indicates general direction of softness: it is worse from bottom right to top left than from top right to bottom left.
4. Color dependency
Since the blur spot will vary from one image plane to another, the degree of sharpness correction applied needs to be adapted for each color channel. Applying the same correction to all color channels can lead to color shifts, as illustrated in the picture below.
Detail from original image
Detail from image deblurred without color control: introduces a color shift
Detail from image deblurred using DxO Optics Pro
5. Noise dependency
Noise is the worst enemy of sharpening filters, as increasing image sharpness leads to increased noise.
Furthermore, noise and noise perception vary as a function of both gray level and local image content: lowlight areas will exhibit more noise than highlights, and noise is much more visible in uniform areas than in areas of detail.
Hence an image sharpening algorithm needs to adapt the level of deblurring locally so as to keep noise below the level of perceptibility.
Photo: DxO Labs
DxO Optics Pro correction of image softness
As previously explained, image sharpness is an extremely complex defect to correct.
Common tools, and even more advanced ones, offer solutions that process the entire image field uniformly. If time is not an issue, setting the 3 unsharp masking parameters for each individual image is typically a good way to address lens softness. But this only works in cases where the blur is the same throughout the field, similar on each color channel, there is no astigmatism, and noise is not an issue. This will be true for only a small proportion of images.
Photo: Guillaume Gaiotti
DxO Optics Pro's fully automatic correction of image softness accounts for all the factors described below:
