From: jpc@tauon.ph.unimelb.edu.au (John Costella) Subject: Re: TECH: CRT's vs LCD's -- comparison? Date: Tue, 27 Apr 93 5:00:48 EET [moi] > > Slight problem here Diego if you add the intensities together. If you > > add colour to white then you will get a less-saturated colour. > > > > Diego, your idea is repaired (in principle) if you used an LCD display > > as a *filter* of a CRT image. > [diego] > I can't understand why, if you use an LCD to filter the CRT image this > would solve the problem you highlighted... My original idea was > something like using an half silvered mirror: > > (LCD) > --- > | > |/ | > (eye) * ------------/-------| (CRT) > / | > (mirror) > > (what an horrible eye!) No you're still missing the point Diego. A half-silvered mirror transmits half of the intensity and reflects half (in either direction). The total e-m field is the SUM of the two (so you are *not* in fact filtering; rather, you are superimposing)---so for incoherent sources you get 1 1 I_eye = - I_lcd + - I_crt 2 2 (half for each because the other half gets bounced downwards and lost), where I'm using I for `intensity'. Take the hypothetical example again of the letter 'A', which the CRT can display in full resolution (but in white), but for which the LCD has a single pixel for the whole letter (but in colour). If you want a red 'A', and display a red pixel on the LCD and a white 'A' on the CRT, you will end up with a half-intensity red square with a pink 'A' superimposed on it! You can add up the intensities like this to convince yourself: LCD CRT TOTAL R G B R G B R G B lit red by LCD but not by CRT: 1.0 0.0 0.0 0.0 0.0 0.0 0.5 0.0 0.0 lit red by LCD and lit by CRT: 1.0 0.0 0.0 1.0 1.0 1.0 1.0 0.5 0.5 where total = 0.5 * lcd + 0.5 * crt for each of r, g and b---and (0.5, 0.0, 0.0) is half-intensity red and (1.0, 0.5, 0.5) is pink. The problem is your half-silvered mirror: this ADDS the two signals together. You really want to MULTIPLY the two signals together---and to do this, you need to pass the CRT white light *through* the LCD pixels, in such a way that only the selected frequencies get through. Assuming it acts like a linear filter, the above table then gives you LCD CRT TOTAL R G B R G B R G B lit red by LCD but not by CRT: 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 lit red by LCD and lit by CRT: 1.0 0.0 0.0 1.0 1.0 1.0 1.0 0.0 0.0 where now total = lcd * crt for each r, g and b. So now you get a fully-saturated red 'A' at the resolution of the CRT, on a black background, even though the colour information is only lower resolution, which is what your idea was aiming at. If you try it according to your diagram, you'll just get junk (pink on dark-red; worse in the general case). Simulate the two formulas above on your regular graphics display with a few images, and you'll see why the former is lousy and the latter not-too-bad. This idea is just like what you have with colour text-based displays: you store colour values for each letter only, not for each pixel, which is why I used a letter 'A' above as an example. It's a nice idea though, and I wouldn't want to restrict it to LCD-CRT pairs. If you don't like CRTs on your head you might adapt the idea to two LCDs: one active, high-resolution, monochrome one, and another passive, low-resolution, colour one, mounted in front. The stuff they use for colour overhead-transparency-machine graphics units would be ideal for the latter, in small form. John ---------------------------------------------------------------------------- John P. Costella School of Physics, The University of Melbourne jpc@tauon.ph.unimelb.edu.au Tel: +61 3543-7795; Fax: +61 3347-4783 ----------------------------------------------------------------------------