Warp.

Go to the Lens Correction tutorials to see how it works.

Typical lens distortion causes a square object to appear either as barrel or pin cushion shaped. The most obvious way to tell whether your images are distorted is to look at the straight edges of objects. If lines that should be straight are actually curved, this is indicative of distortion. If there are no long straight edges in the scene, for instance a natural scene, then it is much more difficult to discern distortion even when it is significant. To compute lens distortion, you need an image with one or more curved lines that should be straight or a distortion map.

For line calibration, you can either use your source material if it has suitable lines or you can photograph a calibration grid. In either case, an open shape spline is drawn over a curved line that should otherwise be straight. This spline in conjunction with a camera model is then used to compute the distortion.

If you already have distortion maps, you can remove or work with distortion by importing your maps instead of line calibrating.

There are two parts to correcting distortion--1a) Select a distortion map or 1b) Create an open shape spline over one or more curved lines that should be straight and 2) Correct the distortion.

The image is automatically corrected based on the distortion map.

Uses a calibration clip, normally a grid, when identifying the curved lines that should be straight. If not hooked up, the main input is used instead.

Uses a distortion map to correct the distortion.

Creating at least one calibration line over a curved line that should be straight using the integrated Roto shape tools is the most important part of the process for 1-Parameter, 2 Parameter and Anamorphic calibration. This determines what lines should be straight in the clip.

Since open shapes are required for Lens Correction, it has its own shape tools similar to those in the Roto node. You can choose from B-Spline, Bézier, X-Spline or Magnetic Freehand shapes and have use of the Transform and Reshape tools as well.

After creating the calibration lines, a camera model must be selected.

Use only with RealViz Rz3 files.

Computes and applies the lens correction based on the open shapes and the selected camera model, either 1-Parameter, 2-Parameter or Anamorphic.

Chooses the filtering method when lens correction is applied. Bilinear is the default.

Considered a medium-quality method, it works by averaging the color of the pixel above, below, and to the right and left of each pixel.

The Triangle filter is not the highest quality, but fine for scaled images.

Quadratic is like triangle, but more blur with fewer artifacts. It offers a good compromise between speed and quality.

Cubic is the default filter in Photoshop. It produces better results with continuous tone images, but is slower than Quadratic. If the image contains fine details, the result may be blurrier than desired.

This produces good results with continuous tone images which are scaled down, producing sharp results with fine detailed images.

Gaussian lacks in sharpness, but is good with ringing and aliasing.

A good balance between sharpness and ringing, Mitchell is a good choice when scaling up.

Keeps small details when scaling down with good aliasing.

The DOD of the image is automatically updated based on the lens correction. When enabled, this ensures that image areas corrected beyond the image boundary are not lost when the image is later distorted.

Undistorts the image by removing lens distortion.

Distorts the image by adding lens distortion.

The distortion values for the current camera model are computed automatically using the Calibrate button. If you don’t get good results using the calibration procedure, or have known distortion parameters, you can adjust them manually.

Manipulate the distortion parameters and observe the effect on the image, choosing the parameters that straighten the curved lines as accurately as possible. This should at least deal with the worst effects of the distortion.

K1 is the radial distortion coefficient. If you double the value, you get twice as much distortion. 0 means no distortion, positive corrects for pincushion and negative corrects for barrel distortion. The 1-parameter model is a uniform distortion centered around the Center x/y position,

K1 and K2 are the radial distortion coefficients. K1 works the same as in the 1-parameter model and K2 is a fine tuning of the outer edge of the distortion field. By reversing the sign of K1 vs K2 you can model lenses with uneven, wavy, distortion characteristics.

The Anamorphic camera model uses the Cxx, Cyy, Cxy and Cyx parameters. Cxx and Cyy are a non-linear stretch in the respective direction and have no effect at the image center, increasing toward the edge. Cxy and Cyx are separate horizontal and vertical components of radial distortion. They work similar to the K1 control, but the values are not interchangeable between the models.

The image center is naturally set at the center of the image by default. The coordinates can be entered manually to match the center position of the lens by adjusting the Center X or Center Y parameters. Otherwise, the center is adjusted automatically when Calibrate is selected.

Sets the horizontal center.

Sets the vertical center.

Outputs the result of the Lens Correction node.

Outputs a 32 bit float map that undistorts. Only 1 frame needs to be rendered unless the distortion is animated.

Outputs a 32 bit float map that distorts. Only 1 frame needs to be rendered unless the distortion is animated.