Tweet
Geforce 3 Image quality
I offer the following for your perusal. I tested the following cards myself.
I know some of you may be contemplating buying a GF3 in the light of recent comments, that the image quality on GF3 cards is vastly improved.
Well, you can see what I found out below.
I set up G450 AGP 32Mb (bulk), ATI Radeon 64Mb DDR (bulk) AGP and Elsa
Gladic 920 AGP (retail) inside one computer (Iwill KK-266R based machine
with Enermax 450W PSU) one by one. Each with latest reference drivers
running under Win98.
I then asked three of my colleagues (who were not aware of which
card was inside the machine) to judge the image quality by writing
a relative value on a sheet of paper indicating the relative
performance of each card compared to others. Furthermore, they
were prompted to write down qualitative words to describe the
image they saw from each card.
They used a combination of bit map graphics, black text on a white
background (and vice versa), color/hue purity test pattern and
cross hatch & line convergence patterns to evalute the image quality.
On all of the systems, hw calibrated Mitsubishi Diamond Pro 2040u
was used as the display device. For connecting to the display a double
shielded 75 Ohm terminated UltraVGA cable with BNC connectors was used.
Cards were manually uninstalled and installed with complete removal
of drivers and registry entries along with accompanied other software.
Display settings were scaled to fill the whole viewable area at
both tested resolutions.
The jurors were allowed to change the their preferred color temp
from the default 6500K during the evaluation - they were not allowed
to touch other monitor controls other than color temp, brightness and
contrast (remember, the display is calibrated for maximal accuracy
when fed a perfect signal from a signal generator).
Everybody started the evaluation at 6500K and were
adviced to concentrate on that temp if unsure. Gamma was at 2.2.
They were presented with the cards in following order with
display setttings at 1600x1200@85Hz and 1280x1024@100Hz (in
that sequence):
A. G450
B. Elsa 920 GF3
C. Radeon 64 DDR
The jurors did not see each others markings, nor did they discuss
with each other about the quality or see each others evaluations
before the test ended.
The results are as follows:
1. G450 (2 highest ratings, 1 2nd highest rating)
Adjectives used: "sharp", "steady", "vibrant", "[color] pure", "contrasty",
"
2. Radeon 64 DDR% (1 highest rating, 2 2nd highest ratings)
- "sharp", "steady", "not as contrasty as A", "clearly better than
B", "slight lightness ghosting at 1600x1200 with black on white"
3. Elsa GF3 (3 3rd highest ratings)
- "unfocused", "color fringing", "soft", "color focus problems e.g. with
blue and yellow transition", "less contrasty than A", "wavy image"
After the test I put Radeon in one computer (with Hitachi 823F 21"),
G450 in one computer (Hitachi CM772 19"), and Elsa 920 in one computer
(Mitsubishi 2040u 22") and looked at them side by side. 2040u is
hw calibrated, others only in software.
At this point jurors were allowed to give additional comments
on the quality of the images (they still didn't know which machine
had which card). Monitors were different, but they were side by side.
People now had more trouble discerning between A and C (G450 and Radeon)
especially at 1280x1024@100Hz.
People commented on how much softer the B (Elsa) now looked (more so at
1600x1200). One person said that the image of B at 1280x1024@100Hz
wass fuzzier than the image of A or C on 1600x1200@85Hz, although
he thought that A and C were "pin-sharp" at 1280x1024@100Hz.
One juror said that looking at the slightly fuzzier image of B (Elsa)
was pleasant with graphics, but not when reading small text or details.
Two people also noted that A (G450, running on 19" Hitachi) seemed a bit
unsharp compared to the original test (when it was running on 2040u),
when the brightnes was increased and resolution was 1600x1200@85Hz.
In conclusion they all felt that if they had to pick one, they
would pick either A or C for their own displays. Only one
person said was absolutely sure about picking A (G450) over C (Radeon).
All of them though B was too soft to be used at higher resolutions
and when looking at small details.
What does this tell you - if anything?
I leave that for you to discuss. I won't participate in
the discussion anymore. I think I've contributed more useful data
than anybody else in this thread and if you don't like it, you can
do your own double blind studies. Best of luck in getting them published
(perhaps you can include them in your PhD studies).
I'm through with this and I know what card to put inside my computer
to satisfy my needs and my eyesight. This took half of my working day and
I need to catch up.
cheers,
Halyon
PS I'm still not claiming one card is better in all respects over
another one. I was only trying to find an answer to the question:
"Which card has the best image quality especially for
doing a lot of text work at high resolutions and high refresh
rates, while still not sacrificing 3D speed completely." I've
found my answer - I hope you'll find yours too.
I know some of you may be contemplating buying a GF3 in the light of recent comments, that the image quality on GF3 cards is vastly improved.
Well, you can see what I found out below.
I set up G450 AGP 32Mb (bulk), ATI Radeon 64Mb DDR (bulk) AGP and Elsa
Gladic 920 AGP (retail) inside one computer (Iwill KK-266R based machine
with Enermax 450W PSU) one by one. Each with latest reference drivers
running under Win98.
I then asked three of my colleagues (who were not aware of which
card was inside the machine) to judge the image quality by writing
a relative value on a sheet of paper indicating the relative
performance of each card compared to others. Furthermore, they
were prompted to write down qualitative words to describe the
image they saw from each card.
They used a combination of bit map graphics, black text on a white
background (and vice versa), color/hue purity test pattern and
cross hatch & line convergence patterns to evalute the image quality.
On all of the systems, hw calibrated Mitsubishi Diamond Pro 2040u
was used as the display device. For connecting to the display a double
shielded 75 Ohm terminated UltraVGA cable with BNC connectors was used.
Cards were manually uninstalled and installed with complete removal
of drivers and registry entries along with accompanied other software.
Display settings were scaled to fill the whole viewable area at
both tested resolutions.
The jurors were allowed to change the their preferred color temp
from the default 6500K during the evaluation - they were not allowed
to touch other monitor controls other than color temp, brightness and
contrast (remember, the display is calibrated for maximal accuracy
when fed a perfect signal from a signal generator).
Everybody started the evaluation at 6500K and were
adviced to concentrate on that temp if unsure. Gamma was at 2.2.
They were presented with the cards in following order with
display setttings at 1600x1200@85Hz and 1280x1024@100Hz (in
that sequence):
A. G450
B. Elsa 920 GF3
C. Radeon 64 DDR
The jurors did not see each others markings, nor did they discuss
with each other about the quality or see each others evaluations
before the test ended.
The results are as follows:
1. G450 (2 highest ratings, 1 2nd highest rating)
Adjectives used: "sharp", "steady", "vibrant", "[color] pure", "contrasty",
"
2. Radeon 64 DDR% (1 highest rating, 2 2nd highest ratings)
- "sharp", "steady", "not as contrasty as A", "clearly better than
B", "slight lightness ghosting at 1600x1200 with black on white"
3. Elsa GF3 (3 3rd highest ratings)
- "unfocused", "color fringing", "soft", "color focus problems e.g. with
blue and yellow transition", "less contrasty than A", "wavy image"
After the test I put Radeon in one computer (with Hitachi 823F 21"),
G450 in one computer (Hitachi CM772 19"), and Elsa 920 in one computer
(Mitsubishi 2040u 22") and looked at them side by side. 2040u is
hw calibrated, others only in software.
At this point jurors were allowed to give additional comments
on the quality of the images (they still didn't know which machine
had which card). Monitors were different, but they were side by side.
People now had more trouble discerning between A and C (G450 and Radeon)
especially at 1280x1024@100Hz.
People commented on how much softer the B (Elsa) now looked (more so at
1600x1200). One person said that the image of B at 1280x1024@100Hz
wass fuzzier than the image of A or C on 1600x1200@85Hz, although
he thought that A and C were "pin-sharp" at 1280x1024@100Hz.
One juror said that looking at the slightly fuzzier image of B (Elsa)
was pleasant with graphics, but not when reading small text or details.
Two people also noted that A (G450, running on 19" Hitachi) seemed a bit
unsharp compared to the original test (when it was running on 2040u),
when the brightnes was increased and resolution was 1600x1200@85Hz.
In conclusion they all felt that if they had to pick one, they
would pick either A or C for their own displays. Only one
person said was absolutely sure about picking A (G450) over C (Radeon).
All of them though B was too soft to be used at higher resolutions
and when looking at small details.
What does this tell you - if anything?
I leave that for you to discuss. I won't participate in
the discussion anymore. I think I've contributed more useful data
than anybody else in this thread and if you don't like it, you can
do your own double blind studies. Best of luck in getting them published
(perhaps you can include them in your PhD studies).
I'm through with this and I know what card to put inside my computer
to satisfy my needs and my eyesight. This took half of my working day and
I need to catch up.
cheers,
Halyon
PS I'm still not claiming one card is better in all respects over
another one. I was only trying to find an answer to the question:
"Which card has the best image quality especially for
doing a lot of text work at high resolutions and high refresh
rates, while still not sacrificing 3D speed completely." I've
found my answer - I hope you'll find yours too.
Is this a copy of a post from another forum?
Comment