Display Scaling- Analog vs Digital (CRT vs LCD)

[mdzapeer]

I was wondering about this, but why does scaling on CRT (mutli-sync) look so much better than digital scaling on fixed pixel panels. I also remember reading that the higher the resolution on fixed panel display the better the scaling would look.

Can someone explain to me how this works? It possible to have a modern display technology not based on fixed pixels like a CRT?

[spacediver]

This is an interesting question, and I've explored it a bit over here:

http://www.curtpalme.com/forum/viewtopic.php?t=35579

[mdzapeer]

CRTs also have their 'best' or native resolution, but they look much better when it displays its non-native resolution than it does on LCDs. From your thread it seems there is some analog scaling going on from the guns to the screen.

Another interesting point as well. A video camera recording or photo(ex. 1080p) of a CRT displaying an image at a given resolution. When that recording or image is displayed on fixed pixel panel at its native recording/capture (1080p in our example), looks much better than if the image was scaled by the display itself.

[spacediver]

CRTs also have their 'best' or native resolution, but they look much better when it displays its non-native resolution than it does on LCDs. From your thread it seems there is some analog scaling going on from the guns to the screen.

From what I understand, the guns are just receiving a resolution appropriate signal, and the phosphor is like a canvas that can paint the signal. In the context of an aperture grille layout, there are limits on the horizontal rendering of an image, because the triad of phosphor stripes are not infinitely thin. So suppose there are 1920 sets of triads across the width. There are only a certain number of horizontal resolutions that will fit well. 1920 is obviously one of them. But 1919 would not work as well as it wouldn't fill the display perfectly (you'd have one triad left over on the right or left edge of the screen). If you look at the supported horizontal resolutions of a given CRT (e.g. for my FW900s: 2304, 1920, 1600, 720, 640), they seem to be related to each other for the most part by pretty well rounded numbers (1920/1600 = 1.2; 1920/640 = 3). I wonder whether this relates to the idea of using the triads in a well rounded manner. One complication with the FW900 is that it has a variable dot pitch across the width (it's larger towards the edges). From what I understand, this means that the width of the triads increase as you go towards the edges of the screen. I'm not sure how this plays into optimizing different resolutions.

Another interesting point as well. A video camera recording or photo(ex. 1080p) of a CRT displaying an image at a given resolution. When that recording or image is displayed on fixed pixel panel at its native recording/capture (1080p in our example), looks much better if the image was scaled by the display itself.

That certainly seems to prove that CRT "scaling" is superior to fixed pixel scaling

p.s. I have edited your post, as I think you meant to say "looks much better than if the image was scaled..."

[flood]

a crt never attempts to hit an individual dot of phosphor (for shadow masks) or line of phosphor (aperture grill). the electron gun just scans whatever is sent to it, and really has no idea which phosphor it is hitting, other than that the red gun will for sure hit some red phosphor and likewise for green/blue.

it's entirely possible with software to emulate crt-like resampling on an lcd which has identical subpixel geometry as the crt. basically nearest neighbor/box filtering with a slight gaussian blur.
the reason lcd's appear blurry for low-res is because the method usually used is bilinear resampling.

[blargg]

the reason lcd's appear blurry for low-res is because the method usually used is bilinear resampling.

And then to add insult to injury, do some edge-enhancement post-processing that just makes it look like a pig with lipstick.