From an intuitive-divulgative point of view, you may think of an hologram as a plane or part of one that records the intensity and the direction distributions of light once passed through it. That's what the relative phase recording produces at the end.

When you look through a window you see part of whatever is in the other side, and you get 3D perception due to parallax: each eye gets different information (a 2D distribution of intensity), or if you move you get a different point of view. Then you say you see 3D objects. A possible model for that window may be not "glass+external world" but "plane whose points emit light in certain directions": if you trace a line on the glass in the points where you see some contour and you move, you see that contour through other points of the glass.

That directions distribution recording is achieved through, if you want, local diffraction gratings, that allows you to encode that 3D (2D$\otimes$directions) information in a plane (in surface holograms at least. In volume ones you have local Bragg gratings). The third parameter is the spatial frequency of the grating.

To reconstruct the scene you need to have part of the exact field it was recorded with, and that makes the "relative" in relative phase recover its meaning. That's why you usually chose a scene where there is an object (set of points "emitting" light) plus a collimated beam pointing to the plate. This allows to have an easy to reproduce part of the scene: the collimated beam. With it you can then reconstruct the complete scene, with the plate locally redirecting light to the exact directions, in a certain recording-reconstruction symmetry flavour. The hologram plate is the window through which you see the scene recorded.

I believe if you understand diffracting gratings theory you can make your own idea of holography by just thinking locally and with point/plane sources.

From an intuitive-divulgative point of view, you may think of an hologram as a plane or part of one that records the intensity and the direction distributions of light once passed through it. That's what the relative phase recording produces at the end.

When you look through a window you see part of whatever is in the other side, and you get 3D perception due to parallax: each eye gets different information (a 2D distribution of intensity), or if you move you get a different point of view. Then you say you see 3D objects. A possible model for that window may be not "glass+external world" but "plane whose points emit light in certain directions": if you trace a line on the glass in the points where you see some contour and you move, you see that contour through other points of the glass.

That directions distribution recording is achieved through, if you want, local diffraction gratings, that allows you to encode that 3D (2D$\otimes$directions) information in a plane (in surface holograms at least. In volume ones you have local Bragg gratings). The third parameter is the spatial frequency of the grating.

To reconstruct the scene you need to have part of the exact field it was recorded with, and that makes the "relative" in relative phase recover its meaning. That's why you usually chose a scene where there is an object (set of points "emitting" light) plus a collimated beam pointing to the plate. This allows to have an easy to reproduce part of the scene: the collimated beam. With it you can then reconstruct the complete scene, with the plate locally redirecting light to the exact directions, in a certain recording-reconstruction symmetry flavour. The hologram plate is the window through which you see the scene recorded.

I believe if you understand diffracting gratings theory you can make your own idea of holography by just thinking locally and with point/plane sources.

As for what @wetsavannaanimal-aka-rod-vance said about colour holography (I have no enough reputation to make comments) I can recomend to visit Yves Gentet site to at least see the videos and GIFs.