1 |
Apparently, though unproven, at 22:27 on Wednesday 08 September 2010, Paul |
2 |
Hartman did opine thusly: |
3 |
|
4 |
> On Wed, Sep 8, 2010 at 3:06 PM, Alan McKinnon <alan.mckinnon@×××××.com> |
5 |
wrote: |
6 |
> > Then there's non-square pixels. Without funky voodoo graphics algorithms, |
7 |
> > my screen displays circles as ovals. |
8 |
> |
9 |
> That problem should not exist on LCD if you're using the screen's |
10 |
> native resolution. For example, the most common case of this in CRT |
11 |
> days was 1280x1024 resolution which is not a proper 4:3 aspect ratio |
12 |
> (it is 1.25 rather than 1.33). In other to make a circle look like a |
13 |
> circle you'd need to use 1280x960 instead and adjust the monitor to |
14 |
> make the picture fill the screen, or your programs would need to be |
15 |
> aware of the pixel shape and adjust accordingly (those funky voodoo |
16 |
> graphics algorithms). |
17 |
> |
18 |
> With LCD monitors, the 1280x1080 panels are actually a small bit |
19 |
> taller than a standard 4:3 panel, so a circle should look like a |
20 |
> circle without having to do anything special. (However, if someone |
21 |
> uses any other resolution their circle will be oblong). |
22 |
> |
23 |
> I was a die-hard CRT guy but I've found LCD with at least 100 dpi to |
24 |
> be acceptable compared to the CRTs I've had in the past. And in the |
25 |
> case of my S-IPS monitor I think it is really superior to any CRT I've |
26 |
> ever used. (My monitor with TN panel, however, is pretty bad.) |
27 |
|
28 |
I spent 10 years fixing TVs of every imaginable model and type from the best |
29 |
to the worst, and all the improvements in between. As a result I'm finely |
30 |
tuned to departures from the ideal with any display device. probably finely |
31 |
tuned to a fault :-) |
32 |
|
33 |
I can see pixels refreshing on all flat panels, even the best of the best LED |
34 |
models from Samsung. I finally understood why when I found out how that "Xms |
35 |
refresh time" spec is actually measured. I can see non-square pixels by |
36 |
looking at thin but wide arcs, even more so when an oblong pixel is in a |
37 |
square grid. |
38 |
|
39 |
LCDs are easiest driven in terms of pixels - it maps to video memory. If they |
40 |
are no-square, one has to know the horiz and vert dpi and apply a fudge factor |
41 |
to make the image proportional. If the pixels are on a square grid, then one |
42 |
does not fudge the image. All very horribly complex and frankly, more detail |
43 |
than I can really be bothered with. |
44 |
|
45 |
Which all goes to say that I have an unusual frame of reference, one that is |
46 |
*not* universally applicable :-) |
47 |
|
48 |
Your point about poor CRTs is taken. A lousy CRT is unwatchable but a lousy |
49 |
LCD is tolerable. The finest CRTs though still outshine even the best LCD |
50 |
(again, imho only) |
51 |
|
52 |
|
53 |
-- |
54 |
alan dot mckinnon at gmail dot com |