Unfortunately, while many lenses do an incredible job of manipulating light, the laws of physics have their limits. By combining several lenses, and by using aspherical elements, engineers can minimize these aberrations, but there are trade-offs between image quality, cost, weight, zoom range, and so on – and in some conditions, it is tough to avoid geometric distortions.

DxO’s engineers have established a unique means of correcting distortions so that they can be removed with efficiency and precision. The result is our vast library of DxO Optics Modules.

As you can see, rather than using generic profiles that may not accurately address geometric distortions at all focal lengths and distances, DxO’s precision measurements are always perfectly suited to the specific lens, focal length, and focus distance.

Therefore, you get higher quality and greater field of view compared to using other software, and often better results than the JPEGs generated by your camera itself.

DiffractiveOptics Canon

However, there is a downside to addressing geometric distortion in software. After the distortion is characterized, the photo is warped to compensate for it, but this means the picture itself will no longer be rectangular. For example, images adjusted for pincushion distortion will have convex edges; and those corrected for barrel distortion will have concave edges.

HOLOEYE

The sample measures 7,5 cm x 2,5 cm. Use a laser pointer to see five different patterns. There are two star signs, the moon, a company logo, and a globe that rotates as the laser is moved along the microstructured surface.

Today we added a clever teaching tool to the shop that can help designers, engineers, and students to design better plastic components.   This moulded plastic piece includes features which illustrate a number...

DiffractiveIOL

A basic pinhole camera (which has no lens) produces no geometric distortion at all. However, as anyone who’s used one knows, it captures very little light by the nature of its design. Real cameras use lenses to capture more light and obtain sharper images.

Image

Diffractivelens

The degree of geometric distortion in a lens is evaluated using a reference image containing a dot plot laid out on a very precise grid.

Before exporting an image, it needs to be cropped in order to make it rectangular again, effectively shaving off the curved sides. This means that the original field of view becomes smaller, which is a significant problem when using wide-angle lenses – after all, maximizing field of view is why a wide-angle lens is chosen in the first place. Zoom lenses at their wide end are particularly affected by distortion, and after improvement, can require the harshest cropping.

What’s more, if the photographer’s creative intentions require it, DxO’s software allows the full cylindrical projection of the lens to be used without cropping, meaning that all pixels are available for manual trimming or editing. Our method therefore grants the most freedom possible in the process.

Modern lenses with their many elements can also exhibit more complex distortion than these three well-known types, and therefore resolving the issues requires higher-order distortion models to be created and applied.

Modern lenses are amazing, but they have their limits, and many cannot capture the geometry of a scene with absolute precision. Instead, they can introduce various types of distortions. These distortions are determined by the type of lens as well as how it is used.

Diffractive opticalElements

Images improved by DxO Optics Modules often require less cropping after geometric distortion has been addressed. That means a greater field of view is possible so you get more of the lens you’ve paid for. What’s more, depending on the intentions of the photographer, the total projection of the lens can be used. In this example using a fisheye lens, it’s possible to adjust the image to produce a cylindrical projection, giving outstanding width and the maximum field of view. Every available pixel appears in the final image, instead of being cropped as would happen with other software.

Image

Holoor

DiffractiveOptics

Diffractive optical elements are used for beam shapers, beam splitters, line generators, and diffusers. These components are used in machine vision systems, 3D sensors, bar code scanners, and measurement solutions. The Microsoft Kinect controller uses a projected dot pattern of infrared light generated with a DOE.

Also, it’s not unusual for compact lenses that go from wide to telephoto to demonstrate different types of distortion as you zoom in. So, you might see some barrel distortion at wide focal lengths which then shifts to pincushion distortion at the longer ones.

All this means that, with DxO’s method, many images require less cropping after adjustment and so less of the captured view is lost.

With complex distortions requiring equally complex correction, this is not something that can be adjusted manually by the user; for truly accurate results, they must be calibrated using special equipment. That’s where DxO comes in.

First let’s recap on what geometric distortion is, identify the common types observed in many lenses, explain why they happen, and how they’re corrected.

Diffractivevs refractive optics

Distortion – like the ‘barrel’ effect shown here – can make straight lines in an image appear curved, especially toward the edges of the frame.

Standard elements are available for many applications, but the microstructure – and thereby the light pattern – can be tailored to create virtually any pattern desired. The sample is made from polycarbonate, but you can get glass elements for high-energy applications such as laser cutting and drilling. DOEs are used mostly with laser light, but partially coherent light from LEDs or other light sources can also be manipulated.

• our calibration has very dense coverage of all focal lengths and focus distances, providing excellent corrections up to the image borders.

Distortion depends firstly on the focal length (which can be extensive if it’s a zoom lens) and secondly on the focus distance. To create each lens’s model, DxO’s experts test both by photographing a reference image that features dots laid out on a highly precise grid that is perfectly flat and parallel to the camera sensor. To achieve this, many calibrations need to be made, particularly for large-range zoom lenses — a lengthy and demanding process.

However, once the extent of distortion is precisely calculated for each position in the field of view, the calibration data can be compiled into a correction file called a DxO Optics Module.

This optical component has a microstructured surface that diffracts light into a tailored pattern. The light is redistributed with very little loss.

A QWERTY pattern is not available as a standard pattern, but it can be made. The information card that you get with the sample includes the contact information for the manufacturer.

Most lenses produce some geometric distortion, though in certain models this is negligible to the point of being invisible under regular observation. An ‘ideal’ lens would produce no distortion at all, and would translate any straight lines in the scene into identically straight lines in the image – a situation called ‘perfect perspective projection.’ Any deviation from this is classed as a geometric distortion.

The quality of these corrections can convey benefits well beyond the simple straightening of lines. When you apply one of DxO’s Optics Modules to an image, any lens distortion that’s been measured by our laboratory technicians will be automatically corrected. And while most image editing software can do this, DxO software goes further. Thanks to the accuracy of our measurements, photographers can even access a greater field of view after cropping.

Yes, it is possible to have custom patterns produced but the tooling cost starts at €5000, so it is only feasible for large volumes or high-value applications.

Image

With the hard work done, it takes only seconds for photographers to download the specific DxO Optics Module required and apply corrections to their images. One of these Modules exists for almost every camera and lens combination produced in the last 20 years, and the required module is automatically identified from a photo’s metadata.