I have an image which was taken from an airplane. I know the airplane position when the picture hasbeen taken (lat lon height, heading pitch yaw and camera fov). Suppose I would like Cesium to show the earth exactly like the image was taken. I can do it by:

Set Cesium container div to have same width/height as in the image.

Set camera.frustum.fov to appropriate values.

Use camera.setView with positionCartographic and angles.

The registration data of it are not accurate, thus the image has additional data produced by accurate registrator. The data ihaves a fixing formula that should be applied on the image. The formula is represented as an affine matrix that should be applied on the image (2d, 3x3 matrix in which the last line is 0 0 1).

To get this fix on Cesium, I should apply something similar to the affine transformation on Cesium camera. Applying standard operations on the camera like moveLeft, rotateLeft, etc. are not mathematically equivalent to affine transformation as explained here: http://stackoverflow.com/q/33547958/2463642 . Thus I need to find a way to apply the affine transformation after any world-to-screen calculation and before any screen-to-world calculation is done. This is a purely technical question in Cesium core rather than mathematical one. How should I do that? I'm afraid that such transformation is performed in many places in the code...