[lug] Monitor/Card tweaking - was Re: Color tweaks

D. Stimits stimits at idcomm.com
Fri Feb 23 13:36:10 MST 2001


John Starkey wrote:
> 
> Thanks Dan and Tkil. After a lot of playing around I've determined I need to
> get a new monitor. Even the best color I can get still looks foggy. I swore
> that once or twice I had decent color but it's hard to judge without a ref
> point after all this. I did manage to get better color than Windoze :}

One thing that matters in addition to the monitor is the video card, and
whether you're using a video cable extension. What video card and
monitor do you have? Do you have an extension for the video cable?

> 
> I was thinking one other thing which I can't try at the moment would be to
> split the 60v circuits, I've got 4 computers plugged into the same 3 power
> strips on the same circuit. Even the iMac looks like it might be a bit weak at
> the moment.
> 
> Regardless, the info you guys posted was a really big help in understanding
> what I was doing. I was able to confirm that this thing sucks :}
> 
> Thanks again,
> 
> John
> 
> Tkil wrote:
> 
> > i remember thinking that ESR's discussion of monitor tweaking was very
> > helpful to me:
> >
> >    http://www.linuxdoc.org/HOWTO/XFree86-Video-Timings-HOWTO/
> >
> > the tradeoff i tend to make is for maximum horizontal resolution,
> > until i start losing vertical single-pixel lines.  a toy i put
> > together to help me test this, as well as color convergence, is at:
> >
> >    http://slinky.scrye.com/~tkil/xcalib.tar.gz
> >
> > it just draws a grid of white lines on a black background; i resize it
> > to fit the screen, and then look to see that everything looks square
> > (tests for aspect ratio) that there aren't any blue fringes (color
> > convergence problems) and whether the vertical lines are the same
> > apparent brightness and sharpness as the horizontal lines (tests to
> > see if i'm pushing one of the components past its limits).
> >
> > things to think about, to make sure you have good X:
> >
> > 1. video card.  for coding work, you want cards that have top-notch 2d
> >    display capabilities.  3d benchmarks (pixels or polys per second)
> >    don't matter at all; rather, it's the speed and quality of the DAC
> >    that matters when you're talking to most monitors.  i've been very
> >    happy with all the matrox cards i've ever had.
> >
> >    having said all that, most of the current crop of 3d-accelerator
> >    cards have more than adequate RAMDACs for most 2d work.
> >
> > 2. cable / interconnect.  if your monitor supports it, the use of BNC
> >    connectors (preferable 5-tail) is a good thing; i can notice the
> >    difference.  i can also notice the negative impact of using a
> >    keyboard / video / mouse (KVM) switch.
> >
> > 3. monitor.  there are very good monitors available for not much money
> >    these days.  even if you can't afford super-duper top of the line
> >    things (say, like a sony GDM-FW900 ... drool.), there are many
> >    reasonably priced ones that give a good tradeoff on usability.
> >
> > since the keyboard, monitor, and mouse are all things i use very much
> > continuously, it's worth a reasonable investment to me.
> >
> > having said that, here are my tricks for getting the most out of
> > marginal or out-of-date hardware.
> >
> > a. use a light-on-dark scheme for text whenever possible.  especially
> >    if you choose a color primary on black, this removes any issues
> >    with color convergence, and it drastically reduces any strobing
> >    from low refresh rates.
> >
> > b. consider non-standard resolutions, if your monitor supports them.
> >    i run my Sony Multiscan 20se (GDM-20SE1, same exact hardware as
> >    apple two-page color display, and as the HP monitor i have at work)
> >    at 1360x1024, since that's closer to 4x3 than 1280x1024 is.  given
> >    the constraints of the rest of the system, this gives me a vertical
> >    refresh of 78.08Hz (horizontal is 82.76, pixel clock at 145MHz).
> >    my mode line for this is:
> >
> >       Modeline "1360x1024"  145    1360 1381 1491 1752  1024 1027 1030 1060
> >
> > c. don't use KVM switches for machines you're spending lots of time
> >    at.  for an admin console on a couple of different machines, sure;
> >    for your primary development box, forget it.
> >
> > d. prefer sans-serif fonts.  serif fonts are easier to read *on
> >    paper*; i've found that switching to helvetica and lucida
> >    drastically improves the legibility of on-screen text.
> >
> > e. arrange your colors so that they don't clash with each other.  a
> >    snapshot of my current color scheme (fvwm2 and xemacs) can be found
> >    at:
> >
> >       http://slinky.scrye.com/~tkil/color-scheme.png
> >
> > f. when balancing resolution vs. refresh rate, pay attention to the
> >    image quality.  usually this will start to degrade as you get
> >    closer to the specified limits; i've had best results by staying
> >    within the bottom 90 to 85% of the specs at all levels (RAMDAC,
> >    monitor bandwidth, monitor hsync freq range, monitor vertical
> >    refresh freq range.)
> >
> > anyway.  my personal recommendation would be to buy a quality video
> > card (G200 PCI or AGP with 8MB of RAM is fine for development work,
> > and either one is available for less than 100 USD), buy a quality
> > monitor (i've had good luck even with 300 USD 19" CTX monitors from
> > Sam's club, and the Hitachi "elite" 19" monitors are very nice as
> > well.)  finally, if your monitor supports it, buy or scrounge a PC-VGA
> > (HD15 sub or whatever it's called) to 5x BNC cable.  they're about 60
> > USD new, but you might be able to find one lying around if you work
> > somewhere there used to be lots of unix workstations.
> >
> > hope this helps,
> > t.
> > _______________________________________________
> > Web Page:  http://lug.boulder.co.us
> > Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug
> 
> _______________________________________________
> Web Page:  http://lug.boulder.co.us
> Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug



More information about the LUG mailing list