A computer components & hardware forum. HardwareBanter

If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below.

Go Back   Home » HardwareBanter forum » Video Cards » Nvidia Videocards
Site Map Home Register Authors List Search Today's Posts Mark Forums Read Web Partners

Detonator Image Quality Test: Part II



 
 
Thread Tools Display Modes
  #1  
Old September 16th 03, 04:17 PM
magnulus
external usenet poster
 
Posts: n/a
Default Detonator Image Quality Test: Part II

OK, responding to criticisms of my previous test, I decided to do a more
thorough test using only the GeForce 5900 on one machine.

First, the system setup:

EVGA GeForce FX 5900 128MB
Athlon XP 2400
KT 400 motherboard
Windows XP

Det 45's and Det 51's, quality settings, 4xAA, maximum anisotropic
filtering, blend mimpmapping

Testing tools used: Aquamark 3 Professional (I paid 10 bucks for it), Paint
Shop Pro 7, and FRAPS

Benchmarks:
1280X1024 (native resolution of LCD)
Driver:
FPS
Det 45's
18.2
Det 51's
22.2

Image Quality Tests:

OK, first I watched both benchmarks 3 times on each driver. I couldn't tell
much difference, really. If I had to make a quick decision, I'd say the
45's looked somewhat better, more vibrant colors.

Next, I used FRAPS to capture a series of frames detailing the main subject
sub in the early part of the benchmark. I was able to capture two nearly
identical frames: 609 for the 45's, and 608 for the 51's. Upon close
inspection with Paint Shop Pro, I could clearly spot the differences between
the two drivers and images. Overall, the current official 45 drivers have a
much better dynamic range- colors are overall more vivid, especially yellows
and greens, but also the lighting is slightly better for the current,
official drivers. I saw no color banding in either set of drivers .

It's a tough call, but I think the newer 51 drivers perhaps sacrifice too
much image quality. The difference is slight, but noticeable on close
inspection. And the shame of it, they are sacrificing this image quality
for only about 4 extra frames-per-second. Overall my subjective experience
was that both drivers ran the game about the same in terms of performance,
neither one was particularly faster than the other. After all, we are
talking about framerates that are only a little better than a slideshow in
either case.

Conclusion: NVidia needs to find a way to increase driver performance
without sacrificing too much image quality. Failing that, NVidia should use
a slider setting so that users might be able to choose their own trade-offs
between performance and image quality.






  #2  
Old September 16th 03, 04:34 PM
magnulus
external usenet poster
 
Posts: n/a
Default

Oh, and I should also add... I just used Paint Shop Pro to count the
colors in the images, and the Det 51's are displaying 2000 fewer colors.

"magnulus" wrote in message
...
OK, responding to criticisms of my previous test, I decided to do a more
thorough test using only the GeForce 5900 on one machine.

First, the system setup:

EVGA GeForce FX 5900 128MB
Athlon XP 2400
KT 400 motherboard
Windows XP

Det 45's and Det 51's, quality settings, 4xAA, maximum anisotropic
filtering, blend mimpmapping

Testing tools used: Aquamark 3 Professional (I paid 10 bucks for it),

Paint
Shop Pro 7, and FRAPS

Benchmarks:
1280X1024 (native resolution of LCD)
Driver:
FPS
Det 45's
18.2
Det 51's
22.2

Image Quality Tests:

OK, first I watched both benchmarks 3 times on each driver. I couldn't

tell
much difference, really. If I had to make a quick decision, I'd say the
45's looked somewhat better, more vibrant colors.

Next, I used FRAPS to capture a series of frames detailing the main

subject
sub in the early part of the benchmark. I was able to capture two nearly
identical frames: 609 for the 45's, and 608 for the 51's. Upon close
inspection with Paint Shop Pro, I could clearly spot the differences

between
the two drivers and images. Overall, the current official 45 drivers have

a
much better dynamic range- colors are overall more vivid, especially

yellows
and greens, but also the lighting is slightly better for the current,
official drivers. I saw no color banding in either set of drivers .

It's a tough call, but I think the newer 51 drivers perhaps sacrifice

too
much image quality. The difference is slight, but noticeable on close
inspection. And the shame of it, they are sacrificing this image quality
for only about 4 extra frames-per-second. Overall my subjective

experience
was that both drivers ran the game about the same in terms of performance,
neither one was particularly faster than the other. After all, we are
talking about framerates that are only a little better than a slideshow in
either case.

Conclusion: NVidia needs to find a way to increase driver performance
without sacrificing too much image quality. Failing that, NVidia should

use
a slider setting so that users might be able to choose their own

trade-offs
between performance and image quality.








  #3  
Old September 16th 03, 04:39 PM
Harry Muscle
external usenet poster
 
Posts: n/a
Default

"magnulus" wrote in message
...
Oh, and I should also add... I just used Paint Shop Pro to count the
colors in the , and the Det 51's are displaying 2000 fewer colors.


Would it be possible that the color difference is because you are comparing
two different frames. They're pretty close, but they're still not the same
frames?

Harry




-----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
-----== Over 100,000 Newsgroups - 19 Different Servers! =-----
  #4  
Old September 16th 03, 05:02 PM
magnulus
external usenet poster
 
Posts: n/a
Default


"Harry Muscle" wrote in message
...
"magnulus" wrote in message
...
Oh, and I should also add... I just used Paint Shop Pro to count the
colors in the , and the Det 51's are displaying 2000 fewer colors.


Would it be possible that the color difference is because you are

comparing
two different frames. They're pretty close, but they're still not the

same
frames?

Harry


Possible, although the images otherwise look identical to me

just check out this
website:http://www.gamersdepot.com/hardware/..._nvidia/dx9_re
l50/001.htm

The ATI Halo picture has 17000 more colors than the NVidia one (which is
running Detonator 50). I used Paint Shop Pro 7 to analyze the pictures.



  #5  
Old September 16th 03, 10:42 PM
magnulus
external usenet poster
 
Posts: n/a
Default


"Lenny" wrote in message
...
The ATI Halo picture has 17000 more colors than the NVidia one (which

is
running Detonator 50). I used Paint Shop Pro 7 to analyze the pictures.


When you do this color counting stuff, just make sure you're not using

jpeg
images as your base comparison, because that would skew the results. Maybe
not to the extent of adding or removing as much as 17.000 shades, but jpeg
IS a very heavy-handed compression scheme.

Just FYI...


Yes, but they are both JPG. That would only be valid if one were an
uncompressed bitmap and the other a jpeg. I would assume JPEG compression
removes an even amount of all colors, right? You can look at both images
and tell that one of them is alot darker than the other, with more washed
out colors.

I also did a comparison of 3 different Aquamark jpg images that were
greyscaled, one was ATI, one was GeForce at full precision (32-bit), another
one was at 16 bit. And the GeForce-32-bit actually had the most shades of
grey (about 4500), more than the ATI, but the 16-bit was the worst. There
were several hundred more shades in the NVidia-32-bit image than the ATI
image, and about 1000 more in the 32-bit vs. the 16-bit.

Working with uncompressed images would be the best, but if JPEG's all you
have, that's all you have.

Here's another case... two uncompressed bitmaps I took from Aquamark 3:

Det 45 frame 609 has 39134 colors
Det 51 frame 609 has 36926 colors

That's a difference of 2208 colors. Det 51 is displaying about 5 percent
fewer colors. If you do a subtraction function of the two images, you find
most of the color loss is concentrated on the submarine hull, almost
exclusively (terrain lighting is mostly unaffected). Rather than being a
bright greenish-yellow, it's more a dark, muddy green.


  #6  
Old September 17th 03, 12:23 AM
ho alexandre
external usenet poster
 
Posts: n/a
Default

magnulus wrote:
website:http://www.gamersdepot.com/hardware/..._nvidia/dx9_re
l50/001.htm


ROTFLOL
do you realize your are comparing two JPEG images (meaning they are
compressed with degradation) ?


--
XandreX
/I'm that kind of people your parents warned you about/

  #7  
Old September 17th 03, 12:33 AM
magnulus
external usenet poster
 
Posts: n/a
Default


"Harry Muscle" wrote in message
...
"magnulus" wrote in message
...
Oh, and I should also add... I just used Paint Shop Pro to count the
colors in the , and the Det 51's are displaying 2000 fewer colors.


Would it be possible that the color difference is because you are

comparing
two different frames. They're pretty close, but they're still not the

same
frames?

Harry


Update, I just captured two identical frames at 609, and the difference in
color depth is still there. Still about 5 percent of all the colors missing
in the NV_51.75 version, and to top it off most of the color loss is
concentrated on the submarine and some of the lighting effects.

OH, and these are NOT JPEGS, THEY ARE 32-BIT-BITMAPS!


  #8  
Old September 17th 03, 12:41 AM
ho alexandre
external usenet poster
 
Posts: n/a
Default

magnulus wrote:
I would assume JPEG compression
removes an even amount of all colors, right?


Of course not !!
Take a file filled with '1' and another one containing an english
message, and then zip both of them. Compression rate is not the same,
and so is degradation when you do a JPEG compression.

Moreover, you don't even know if the JPEG configuration was the same
when making both images.



Here's another case... two uncompressed bitmaps I took from Aquamark 3:


That comparison is far more interessting indeed.



BUT you took PSP because you could'nt see the difference with your own
eye (nobody blames you, you're not the six million dollar man you're
just a regular human). I guess it is logical to say you won't see the
difference while playing either.



--
XandreX
/I'm that kind of people your parents warned you about/

  #9  
Old September 17th 03, 04:53 AM
duke
external usenet poster
 
Posts: n/a
Default

all i know is thank you for the info
seems like what you are doing is some very advanced research

"magnulus" wrote in message
...

"Harry Muscle" wrote in message
...
"magnulus" wrote in message
...
Oh, and I should also add... I just used Paint Shop Pro to count the
colors in the , and the Det 51's are displaying 2000 fewer

colors.


Would it be possible that the color difference is because you are

comparing
two different frames. They're pretty close, but they're still not the

same
frames?

Harry


Update, I just captured two identical frames at 609, and the difference

in
color depth is still there. Still about 5 percent of all the colors

missing
in the NV_51.75 version, and to top it off most of the color loss is
concentrated on the submarine and some of the lighting effects.

OH, and these are NOT JPEGS, THEY ARE 32-BIT-BITMAPS!




  #10  
Old September 17th 03, 09:23 PM
Bratboy
external usenet poster
 
Posts: n/a
Default

"Lenny" wrote in message
...

I would assume JPEG compression
removes an even amount of all colors, right?


Well, assuming every pixel in the image was identical, that would be the
case yeah. Since the two driver versions do not produce the same output
that's not exactly the case.


well since he is comparing the very SAME frame on both drivers seeing a
reduction in colors on the newer version of drivers does seem to indicate,
at least to me, that there is some adjusting down of color display in the
new drivers to help up its general scores. And your right in the comment
that the 2 drivers are not producing the same output, which is what some are
concerned with hehe. Whether tweaking the drivers to down sample number of
colors to improve speed would fall into the "real optimizations" or another
undisclosed cheat I have no idea. Have seen a fair number of posts and
articles discussing that yes you do get a bit more performance out of 51's
but their also saying IQ has dropped again. If that is the case then it
would seem Nvidia is once again jerking ppl around about how good its stuff
is when in fact what their doing is reducing their quality to claim
performance. I loved all my nvidia cards Riva 128,lightspeed128 TnT2 GF3 but
when I got a fx5200 I wanted to puke. Sent that back got a 5600 ultra and at
same time recieved a 9800 as a gift so compared both and decided to put the
ATI in my main computer and the FX is in a box I rarely ever use. Maybe next
gen of chips Ill come back IF nvidia gets its act together and starts
sticking with industry standards so that the cards can be relyed on once
more.


 




Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
my new mobo o/c's great rockerrock Overclocking AMD Processors 9 June 30th 04 08:17 PM
Why poor quality video via ATI AIW / MMC TV??? Adam White Ati Videocards 2 April 20th 04 02:16 AM
ALLARD ON THE FUTURE OF XBOX - PART TWO Zackman Ati Videocards 2 April 13th 04 07:40 PM
64 benches Ed Light AMD x86-64 Processors 2 April 4th 04 08:16 PM
spec tech/Ram Max Dennis E Strausser Jr Overclocking 0 March 21st 04 03:09 AM


All times are GMT +1. The time now is 05:08 AM.


Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2022, Jelsoft Enterprises Ltd.
Copyright 2004-2022 HardwareBanter.
The comments are property of their posters.