< Previous | Contents | Manuals Home | Boris FX | Next >
The Reference Image
The reference image has two independent gradients: a horizontal gradient in the red channel from left to right, and a vertical gradient in the green channel from bottom to top.
When the image is distorted, you get something like this (with a box around it to make the boundary apparent):
Software can use this image to reproduce the same distortion on other images.
Each pixel in the output image is selected from the input image by using an X coordinate determined from the red channel, and a Y coordinate determined from the green channel. The ability to do this can often be found in existing compositing nodes.
TECHNICAL DETAILS (for consistency with other developers). The map's colors are defined so that red=0.00 is the left edge of column 0 of the source, red=1.00 is the right edge of column w-1 and similarly green=0.00 is the bottom edge of row 0, and green=1.00 is top of row h-1. To pull from the lower-left pixel of an HD image, red=0.5/1920, green=0.5/1080. This definition is required for consistency and to permit the use of a map at one resolution with other images. In some apps the red and green channels must be slightly adjusted; Fusion can be used for a quick undistort/redistort test of the difference this makes.
You can make your own reference image if you like using Photoshop... or simply by opening a shot in SynthEyes, not setting up any lens distortion, and writing the distortion map.
©2024 Boris FX, Inc. — UNOFFICIAL — Converted from original PDF.