"digital" TV's such as DLP, LCD, plasma, ... have a fixed number of triplet (red green blue) rectangular pixels on the display, usually 1920x1080 these days on HDTVs. During each "refresh"" cycle, each bar of color can only have a fixed intensity. (see image below). Non-native resolutions, such as standard def 740x480 with .9 width to height dot aspect ratio have to be digitally upconverted for a hi-def digtal HDTV. OLED tv's use one LED for each dot of color.
A single tube CRT also has a fixed number of triplet (red green blue) spots of phopspors. The spots can be rectagular and side by side (triniton type) similar to LCD panels, or round and arranged in alternating triangle patterns using some type of mask for each dot of color. The phosphors themselves have a much finer grain than the dot size, so a partial sweep of a beam only lights up a portion of each dot, allowing for image boundaries to occur within a dot, which allows CRTs to handle multiple resolutions (hi-def and standard def) without requiring digital up or down conversion. Because the phosphor grains respond independently to the electron beam, CRT monitors (used for computers) ignore mask boundaries and handle a wide range of resolutions without issue. (see image below).
A 3 tube CRT projector uses 3 tubes, red, green, blue, and no mask on any of the tubes. The image is controlled by the electron beam (beam width and sweep rate) in each tube. I think 2006 was the last year CRT rear projection HDTV's were made. They generally have two modes of beam rates, one for standard def 480p images, the other for 1080i images, and each mode needs to have the convergence of the images adjusted on a regular basis (2 to 4 times a year) to keep the images converged.
LCD panel image:
OLED panel image, since the OLED's have diffrent brighness levels, the sizes are smallest for green, largest for blue:
Shadow mask CRT image, note the outer dots are partially lit up corresponding to the edges of the arrow: