[lug] Color tweaks

D. Stimits stimits at idcomm.com
Wed Feb 21 17:09:02 MST 2001

John Starkey wrote:
> > > > The monitor peaks at 53.7K and 85Hz and the only mode line I can find
> > > > right now is 55.84k and 72Hz. Is there a way to adjust this mode line to
> > >
> > > Sorry.. that's 55.84 and 85. The modeline below is correct.
> > >
> > > # 800x600  @  85Hz,  55.84 kHz  hsync
> > > Modeline  "800x600"  60.75   800  864  928 1088    600 616  621  657
> > > -HSync  - VSync
> > >
> > > John
> > >
> >
> > Have you tried xvidtune? If you find a likeable modeline while using it,
> > you can copy it to your XF86Config file.
> Yeah. You had recommended it before and it helped here but what I think
> confuses me is, the above Modeline tells it to use "800x600" then it gives it a
> dot clock then a bunch of possible resolutions and finally "-HSync" and
> "-VSync" (in this case). But some mode lines have "+hsync" and "+vsync".  And
> some have neither. So.... questions:

There is a modeline FAQ or HOWTO out there somewhere, which I haven't
looked at in a VERY long time (though I see below you have found it).
I'm not sure what each timing is, but think of them in general as a
specification of things like compensation for lag (e.g., an artifact of
inductance), periods in which things must be blanked, and correction for
general distortions. Timing issues in general to fine tune your
particular monitor/video card setup. Anything closer in description I
couldn't comment on, but there IS a FAQ out there somewhere, I remember
seeing it a long time ago. You might find more help and information at:

> 1) What are they? What does +hsync and -HSync tell X?  (And why are the caps
> different, is there a "-hsync")?

I don't know if maybe cap's are ignored, I suspect they are. sync pulses
are something like a marker on a stop watch, they are used to mark
specific points in time when some event is relevant. If each scan did
not resync, the picture would begin to drift or distort over time. It is
a means of periodic recalibration of a timing.

> 2) The HOWTO tries to explain a way to trade off between sync and dot clock to
> optimize X and the monitor. I've never tried calibrating monitors before. I
> have a very very fundamental clue(?), I've calibrated audio equipment. But not
> video. I'm worried about venturing too far out. Is this a common trick?

Up until you reach monitor limitations, there is a tendency to gain a
better picture as greater bandwidth is supported. Bandwidth is a product
of horizontal scan rate, vertical scan rate, color depth, and
resolution. Increasing any one item decreases how much total bandwidth
is available for the others. The dot clock, that faster it is able to
run, I think increases the possible resolution for a given scan rate.
However, as a dot clock approaches its limits, its signal might lose its
nice square form, and start distorting or rounding, which degrades the
quality of what you see. So you would consider your maximum resolution
based not just on whether or not you can reach a particular resolution,
but also on whether it starts losing quality...a better dot clock means
you get get closer to the max bandwidth before it distorts. Take this
whole paragraph with a grain of salt, I haven't investigated too closely
what I think is probably correct.

> 3) Sync pulses, positioning them.... This just determines where the new line
> will start and end right?


> 4) -probeonly tells me my max dot clock is 135MHz and I'm only using 60.75 in
> the above example why won't it allow me to bring it up any higher? And
> could/would I benefit from it. I've gotten my 800x600 up to 51MHz and it
> appears it's speeding things up, but 51 is it's max. This is the amount of data
> being thrown at the card right?

Total bandwidth can't be exceeded. Bringing up the dot clock means
bringing down one of the other components that make up bandwidth. See
what happens to max dot clock if you drop color depth, as an experiment.

> Thanks,
> John

Remember, I'm the first one to say I might be totally full of bogus info
on the subject, it's a collection of things I vaguely remember.

D. Stimits, stimits at idcomm.com

More information about the LUG mailing list