|
|
| Author |
Message |
| 20 new of 269 responses total. |
twinkie
|
|
response 250 of 269:
| 
Jun 12 13:28 UTC 2002 |
If you're going to be using Windows, I would strongly suggest 2mb as the bare
minimum, 4 or 8 if you're going to be doing any sort of work with graphics.
There are probably around a hundred companies that make video cards now, but
they're almost all based on a reference design by Nvidia, ATI, Matrox, S3,
or SiS.
Trident is gone, to my knowledge. S3 was gobbled up by Diamond MultiMedia,
which became part of SonicBlue. I didn't think they still made video chipsets
until yesterday, when my new laptop reported that it had S3 graphics.
Currently, a low-end video card would have 16 or 32mb, and support 1024x768
32-bit color, and DirectX 8 support. Most new systems ship with 64mb cards,
supporting 1600x1400 (I think).
|
keesan
|
|
response 251 of 269:
|
Jun 12 14:34 UTC 2002 |
I had not heard of 32 bit color. I can't distinguish 16 from 24 bit easily.
Seems like Kodak is aiming only at the new computer business with its photo
scanning resolution. We don't have monitors that look good at more than 800
so 1M RAM is adequate (gives us 64K).
So that explains why I could not find any S3 drivers or utilities - they are
now Diamond instead. I have not run across SonicBlue or Nvidia or Matrox or
SiS. Thanks for the info.
What is DirectX 8?
|
twinkie
|
|
response 252 of 269:
|
Jun 12 15:04 UTC 2002 |
32-bit color is "true" color. Its doubtful that you'd notice much difference
between 24-bit and 32-bit...but you'd notice a substantial difference between
16-bit and 32-bit color in high-quality photos.
The issue with RAM on your video card has less to do with what your monitor
can display, than it does with how fast the video card can deliver images to
your monitor. You would (or should) notice a difference in overall system
speed if you're going from a 1mb card to a 4mb card.
If you're in need of S3 drivers, go to https://ranger.s3graphics.com and
use" archive" as the login and password. You might want to download S3ID, which
is a DOS program that will tell you precisely which S3 video card you have.
(You can download it from http://www.s3graphics.com/S3id.exe)
DirectX is basically a "helper" that makes games render graphics with your
computer's CPU, rather than your video card's GPU. Though, it's doubtful that
any of the cards you have are DirectX compliant, except for an S3 ViRGE if
it has enough memory.
|
keesan
|
|
response 253 of 269:
|
Jun 12 19:32 UTC 2002 |
We don't have any games, anyway. I already have S3id.exe but whatvga2 tells
us something about the card (trio64 or vision964, for instance) as well as
testing resolutions and numbers of colors.
Win95 seems to offer 16 and 32 bit but not 24 bit color. None of our cards
do 32 but some do 24. Another reason to use DOS.
Is there some reason to look for different Win95 graphics drivers for the
various S3 cards? Win95 seemed to recognize our Trio32 and did not need a
driver for it. For DOS I use VESA drivers (and the two .vrs drivers for
WP51/DOS).
We will experiment with the 32 and 64 bit S3 cards and time them with 1M or
2M RAM on large images. Would the difference be noticeable in a slow Pentium?
|
twinkie
|
|
response 254 of 269:
|
Jun 12 19:58 UTC 2002 |
re: 253 -- If the card can do 24-bit color, and the company that made the card
included driver support for 24-bit color, Windows 95 will offer it as an
option. Sometimes, getting "unsupported" drivers will "unlock" oddball
resolutions for you. (24-bit is a bit uncommon)
Usually, getting the latest and greatest drivers will provide you with more
streamlined drivers that reduce system load, and often contain new features
(different resolutions, for instance)
|
keesan
|
|
response 255 of 269:
|
Jun 13 01:37 UTC 2002 |
I find 24 bit to be quite common in our 1994 cards - the older ones went only
to 16 or 8 bit. Perhaps by 1995 32 bit was more common. I know that the two
cards we put into Win95 computers did 24 bit (Trident and S3).
|
twinkie
|
|
response 256 of 269:
|
Jun 13 05:50 UTC 2002 |
Aaah...it didn't hit me until just now...the difference in production cost
would have been significant enough in 1994 to make cards that did "almost
true" color @ 24-bit.
|
keesan
|
|
response 257 of 269:
|
Jun 13 13:27 UTC 2002 |
The Tseng PCI card also does 24-bit color, which seems to be referred to as
'true color' in my other programs.
|
twinkie
|
|
response 258 of 269:
|
Jun 13 14:38 UTC 2002 |
It's referred to as "true color", but it isn't. IIRC, the industry definition
for true color is anything over 16-bit (high color), but the "real" definition
is 32-bit.
|
keesan
|
|
response 259 of 269:
|
Jun 13 17:48 UTC 2002 |
True analog color would be more than that, would it not? Can you see any
difference between 16, 24 and 32 bit color? I read that some women have two
different genes for seeing red, and can therefore distinguish more colors than
most people. It is an X-linked gene so men are more often unable to see red.
|
mdw
|
|
response 260 of 269:
|
Jun 14 06:04 UTC 2002 |
For color reproduction, you care about the dynamic range, # of
intensities, and the three "primaries" used as references. The human
eye uses a logarithmic scale for dynamic range and is capable of
handling a very wide range -- wider than most man-made devices. On the
other hand, the human eye has serious limitations (high fidelity color
and detail in only a small area, blind spot, persistance issues) that
make it much worse than any man-made device; the human brain does a lot
of post-processing to correct for all of these errors.
Color photography is more limited than the human eye in terms of dynamic
range, and CRTs and LCD displays are even more limited. Given the
limited dynamic range, 256 intensities is sufficient; in most cases the
human eye can't spot the difference between neighboring intensities. If
displays could handle a wider dynamic range, a logarithm intensity scale
would be useful, but this is uncommon, if it is done at all. With any
color system that mixes 3 colors, there are colors "around the edge"
that cannot be reproduced in terms of those 3 colors. NTSC and PAL use
slightly difference reference colors, so there are certain shades of red
in PAL that cannot be accurately reproduced in NTSC.
Red is actually not the hardest color to reproduce. The eye is most
sensitive to small changes in wavelength in yellow. This is a big
problem when making yellow LEDs, because very small changes in the
chemistry can result in very minor shifts of the output wavelength,
which would not be at all noticeable in blue, and insignificant even in
red. But for yellow, the difference is enough to make it look
"different".
Keesan is correct in that some of eye pigments are encoded in the "x"
chromosome. Men, who only have 1 "x" chromosome, are more likely to
have various forms of color blindness. There are also "x" genes that
encode alternate pigments that "work", but respond slightly differently
to color. Since men only get one "x" chromosome, they are more likely
to see colors slightly differently. Women, who get 2 "x" chromosomes,
are likely to have less signficant differences in the way their eyes
encode color, so are likely to see colors more uniformly. I would have
thought that colors would be most likely to be different for yellow.
But, I suppose a red that was significantly shifted would work as Keesan
describes.
I had the impression that many 32-bit graphics cards actually still do
8/8/8 encoding, and leave an extra byte that is not used. The main
advantage would be the potential for faster/cheaper hardware (since
there would be no need to map 3 into an inherently binary addressing
scheme.) Some cards use the extra byte for "overlay" data -- this can be
used for cursors, text, or anything else that needs to be "overlaid"
over an image without having to do nearly as much bit twiddling in
software. Some graphics storage formats (such as PNG) use the extra
byte for an "alpha" channel, allowing for partial transparency (the
MacOS X GUI makes extensive use of transparency).
Different people probably mean different things by words such as "true
color", "grayscale", "direct color", etc. The most common definitions
I've found are that:
indexed - pixels encoded as 8 or 12 bit value; map one of 256 or 4096
color pallete entries which may each be 8/8/8.
true color -pixels encoded as 5/6/5, 6/6/4, or 8/8/8, which
then directly encode the output intensity.
direct color -8/8/8 rgb value is mapped to a r'g'b' value in hardware
to implement gamma correction (this allows certain differences
in color between different kinds of display devices to be
automatically corrected.) Generally, to do this right requires
adjusting the *individual* display's gamma correction using an
accurate (and typically very expensive) color standard.
So, curiously enough, true color is more direct yet less accurate than
direct color. Neither has anything to do with the # of bit/pixel.
16-bit true color display adapters were quite common at one time.
For more information, try these URLs:
http://www.nuclecu.unam.mx/~federico/docs/x-concepts/c0102.html
http://www.eclipsecorner.org/articles/SWT%20Color%20Model/swt-color-model.h
tm
http://www-personal.engin.umich.edu/~jpboyd/ eng403_glossaryreferences.pdf
http://www.gateway.com/s/VIDCARD/STB/v00804/Gloss3d.shtml
http://www.shutterbug.net/
http://www.refillink.nl/glossary.htm
http://www.forums.pctechguide.com/glossary/bycat.php?catSelected=6
http://whatis.techtarget.com/definition/0,,sid9_gci213224,00.html
|
oval
|
|
response 261 of 269:
|
Jun 14 18:27 UTC 2002 |
very interesting marcus. when i was working on print graphics it was a major
headache. i tend to use very subtle colors, and i use a powerbook, which has
a different look of color than the other monitor hooked up to my laptop, both
of which don't correctly display the colors i am actually expecting in the
final printout (CMYK). it's like painting with colors that are just slightly
off. this is hard when you've got say 2 colors that are really close in value,
yet subtly different hues. (the way that mustard color can go green next to
red etc)
|
mdw
|
|
response 262 of 269:
|
Jun 14 23:53 UTC 2002 |
Print graphics gets you into a whole bunch of new problems, like dot
spread and subtractive color. But yes, if you want accurate colors
between different monitors & printers, that's hard to do right. I don't
know how smart Windows is about all this. This seems like an area where
there would be the most interest in doing this right on an Apple
Macintosh, but I found surprisingly few references about that.
|
oval
|
|
response 263 of 269:
|
Jun 15 19:11 UTC 2002 |
color calibration is tricky, and probably more trouble than it's worth
considering that matching is really not possible anyway with 2 different
mediums. i think most have just accepted using pantone swatches as a
reference, and some know colors by their code and don't get bothered by the
fact it looks different on their monitor, and don't have time to. (those
swatches cost an arm and a leg).
|
gull
|
|
response 264 of 269:
|
Jun 17 13:38 UTC 2002 |
Re #258: I thought 24-bit was "true color" (meaning it could produce more
colors than the eye can distinguish between), and 32-bit was just 24-bit
unpacked for more efficient processing. I can see the difference between
16-bit and 24-bit color on some images, especially if I open up a lot of
windows, but not between 24-bit and 32-bit.
|
keesan
|
|
response 265 of 269:
|
Jun 17 14:18 UTC 2002 |
I see ads now for scanners that do true 42 or 48 bit color. ?
A (female) friend gave me a tour of their house, which they just repainted
in several shades of what look to me pretty much like white. There are pale
blue or violet tints to the walls, she says. It was important to her that
each room be a different 'color'.
The thing about X-linked genes, is that women are chimeras. In each cell one
of the X-chromosomes gets turned off, at random. So if you are a woman with
one each of the two different genes for detecting red (which I am guessing
are most sensitive at slightly different wavelengths) half of the cells in
your eye will have one gene functioning, and half the other. If you have the
colorblind gene and are F, half of the cells in your eye will see red and the
other half will be ignored by your brain.
Do the slight differences in white offered by the paint company affect
people's moods? I can tell cool from warm white but not much else.
|
slynne
|
|
response 266 of 269:
|
Jun 17 14:37 UTC 2002 |
well, when I worked at ASH I got to pick the color of the living room.
I chose a soft white and then had them add some pink into it. Not
enough that anyone would walk into the room and say "OH MY! The walls
are pink" but I thought it did add a calming element to the room. I
liked to think so even though I would still get the "So and so is
swinging a branch from outside and threatening everyone." It also didnt
stop me from getting smacked in the head or my co-worker from getting
pepper spray in his eyes, etc.
|
oval
|
|
response 267 of 269:
|
Jun 17 18:14 UTC 2002 |
what also would make a difference would be the color of other things in the
room such as curtains, furniture, carpet - things that would be seen right
next to the walls at the same time could push the subtle tone to show up more
or less depending on the other colors in there. (plus lighting of course)
the pants i'm wearing look like kind of a raw sienna right now, and on the
orange-seated subways they look almost green, since i guess the orange pulls
out any blue in there.
personally i find that warm colors are more relaxing than cool. if my sanity
were questionable (heh) i'd be much calmer in, say, an olive green room with
white trim and a nice hardwood floor, than in sterile whitish, clinical,
airline motif with flourescent lighting..
|
slynne
|
|
response 268 of 269:
|
Jun 17 19:02 UTC 2002 |
Yeah, this room had blue furniture and a blue area rug (not picked out
by me), dark wood trim and dark hardwood floors. It wasnt a bad room. I
thought it was somewhat calming even though I didnt like the furniture
much (except for the cherry desk that I got for a song from property
disposition)
|
keesan
|
|
response 269 of 269:
|
Jun 21 15:06 UTC 2002 |
Just checked out several video chips for possible use on our latest computers.
Trident 8900, Tseng Et4000, Cirrus Logic GD5428, and S3 (various).
The first three come with their own DOS utilities, which let you choose
monitor type and set mode. S3 does not. But all four have special drivers
(graphics and text) available for WP51, and there is a generic vmode.exe
program which will set modes.
First of all the Trident cards claim to be VESA 1.2 (or 2.0) compatible but
they are not very compatible. Most of ours went only to 800 resolution in
VESA mode (to 1280 in Trident mode). One went to 1024. The other cards seem
to be fully compatible to at least 1280 resolution.
The Cirrus and S3 cards have in addition to the usual text modes (80x25, 28,
43 and 50) also 132x25 and 132x43 (so you can read more on the screen at
once). You can switch to these modes and then use Kermit, Lynx, List, etc.
(Not all programs will let you do this).
Trident 8900 also has 80x30 and 80x60 and 132x30 and 132x60 text mode.
Tseng has 80x28 and 80x60 and 132x28 (not 60?) and also 100x40 which is really
helpful if you want to type a text that will be printed in 12 cpi instead of
10 cpi. I am told it also displays true underlines in mode mono, unlike S3.
Trident should therefore be fine for use in Windows (uses its own Trident
drivers) but not so helpful for DOS text applications.
Cirrus Logic and S3 are VESA compatible so okay for DOS graphics applications,
but not quite as nice for text as Tseng (no 100 column mode). S3 also does
not come with DOS utilities, or I cannot find them.
WP supplies a large file of drivers for the various cards but omitted Cirrus,
which supplies its own. It took me about five years to find this file, and
four to find the Cirrus drivers. The disadvantages of using 10 year old
software.
The cards which I tested were PCI except for Cirrus (which seems not to make
PCI video cards as it went out of the video business).
The S3 Virge does not work with the WP S3 graphics drivers. There is now a
WP/DOS vesa driver for graphics, and that works.
The various graphics modes also display text, but don't seem to be usable in
programs. ALong with no cursor at the DOS prompt, you get no display when
using LIST (blank screen). 128x64 and 100x37 and 96x64 are listed but not
usable.
Is there any perceptible difference between the different brands of the latest
video cards other than speed or resolution? (Are they all VESA compatible?).
|