Can an HTPC resolve clean 540 lines from DVD... - AVS Forum
Forum Jump: 
 
Thread Tools
post #1 of 16 Old 05-02-2001, 07:21 AM - Thread Starter
Advanced Member
 
George Kouzev's Avatar
 
Join Date: Jan 1999
Location: Ottawa, Ontario, Canada
Posts: 561
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
I built a Radeon LE based HTPC last weekend (7.1, 4.1, xx93 drivers) by following the excellent recommendation on this forum and I am sort of already thinking about another solution for DVD... I find the timing with Li On's post very interesting.

Here is what I observed. In the Avia DVD there is a Standard Video Test->Widescreen->Resolution pattern that exhibits a small circle at the bottom right corner with alternating white and black lines at 540 lines of resolution (6.75MHz I think). I switched the HTPC to many different (custom included) resolutions (turned the BLUE and RED guns off to avoid the need for reconvergence) but never got these lines to be uniform. There was always some sort of smearing at certain horizontal intervals along this section.

A first generation DVD player Toshiba 5109, although not ideal in every way, did really shine in comparison. The lines were rock solid, sharp and perfectly uniform. That meant this player could squeeze every single bit of real resolution the DVD format is capable of.

I am not sure if people have tried this particular test on their HTPC, but I would be very interested to see if there is some experience and perhaps a solution for this (a magic resolution settings?)...

GK
George Kouzev is offline  
Sponsored Links
Advertisement
 
post #2 of 16 Old 05-02-2001, 08:03 AM
Advanced Member
 
Dave W's Avatar
 
Join Date: May 2000
Location: Syracuse, NY, USA
Posts: 579
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
The HTPC is pulling the digital data off the DVD, and computers are certainly capable of showing much higher resolution than DVD (720x480), so my guess is the scaling in the video card is slightly blurring the lines. If you can get your video card into 720x480 mode, you might see everything sharpen up, since no scaling would be necessary. You could also try 1440x960 if possible, since it's an integral multiple of the native resolution.

This does bring up an interesting topic that hasn't been much discussed on this forum that I've seen. Even though DVD frames are 720x480 pixels, sampling theory dictates that the resultant picture can contain no more than 360x240 actual pixels of resolution, for the same reason that a CD can contain no frequency higher than a 22 KHz. If the data contains higher frequency information, aliasing will occur, which is exacerbated by scaling. The AVIA disc has content that goes all the way to 720 horizontal pixels, and I believe this is why artifacts are seen on a HTPC. If a DVD contains no more than 360x240 pixels of actual resolution, scaling artifacts will be minimal at any resolution above 720x480, provided quality scaling algorithms are used, since the display frequency is more than twice the original signal frequency.

When DVD reviews refer to "edge enhancement" being seen in the picture, I think what they are seeing is data that was sampled at too high of a frequency for DVD to properly handle.

Does anyone else have any insight?

Dave

Dave W is offline  
post #3 of 16 Old 05-02-2001, 07:27 PM
Advanced Member
 
Dave W's Avatar
 
Join Date: May 2000
Location: Syracuse, NY, USA
Posts: 579
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Anyone have any other comments? http://www.avsforum.com/ubb/smile.gif

Dave
Dave W is offline  
post #4 of 16 Old 05-02-2001, 11:58 PM
AVS Special Member
 
clehner's Avatar
 
Join Date: Feb 2000
Location: Berlin, Germany
Posts: 1,214
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
<font face="Verdana, Arial" size="2">Originally posted by Dave W:
The HTPC is pulling the digital data off the DVD, ...

When DVD reviews refer to "edge enhancement" being seen in the picture, I think what they are seeing is data that was sampled at too high of a frequency for DVD to properly handle.

Does anyone else have any insight?
</font>
Hi Dave,
I am not sure about your last point, but I do have a lot of questions http://www.avsforum.com/ubb/wink.gif

I think edge enhancement is often mixed up with aliasing artefacts. When there are just too many lines to be displayed for the limited resolution the most ugly artefacts are created (I think "Me Myself and Irene" was a good example for that). The Kell-factor tells us that a system of resolution N in a certain dimension can only resolve N*0.6(?) lines in that specific dimension.
So, if the frequency (put in analog terms) is to high the lines will be blurred/jaggy/ugly, right? So, well mastered DVDs avoid any frequencies/resolutions that cannot be handled by the specific limitations of the system.

Regards
Christoph

Regards
Christoph
*************
Contact via Public Profile
clehner is offline  
post #5 of 16 Old 05-03-2001, 03:22 AM
AVS Special Member
 
Frode's Avatar
 
Join Date: Mar 2001
Location: Asker, Norway
Posts: 1,723
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
<font face="Verdana, Arial" size="2">Originally posted by Dave W:

This does bring up an interesting topic that hasn't been much discussed on this forum that I've seen. Even though DVD frames are 720x480 pixels, sampling theory dictates that the resultant picture can contain no more than 360x240 actual pixels of resolution, for the same reason that a CD can contain no frequency higher than a 22 KHz. If the data contains higher frequency information, aliasing will occur, which is exacerbated by scaling. The AVIA disc has content that goes all the way to 720 horizontal pixels, and I believe this is why artifacts are seen on a HTPC. If a DVD contains no more than 360x240 pixels of actual resolution, scaling artifacts will be minimal at any resolution above 720x480, provided quality scaling algorithms are used, since the display frequency is more than twice the original signal frequency.
</font>
Umm..can you explain this to me? Why would you need 4 pixels to display one pixel of resolution? This makes no sense to me. I simply cannot understand how analog waveform sampling theory is related to DVD encoding and playback. When it comes to pixels bits are bits. I can see that color resolution might be affected in the output conversion to analog - but not the actual picture resolution itself.


------------------
/frode

/frode
Frode is offline  
post #6 of 16 Old 05-03-2001, 07:01 AM - Thread Starter
Advanced Member
 
George Kouzev's Avatar
 
Join Date: Jan 1999
Location: Ottawa, Ontario, Canada
Posts: 561
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
After staying a bit late last night I got the answer of my very own question, so:

YES, an HTPC could reproduce clean 540 lines of resolution (DVD limit) for me but ONLY at 720x540p! At that rate, what comes in goes out, no processing artifacts.

Yes, this is a 4x3 ratio (so is my screen) but that is actually great since for anamorphic disks I will be squeezing 540 lines in the 16x9 ratio, whereas in a 720x480p (16x9) resolution there will be 480 lines only (more visible scan lines).

With AVIA I could not test 720px480 as the test gets boxed in the middle, bad... Perhaps the multiples (like 1440x1080p) would have worked, will try it tonight.

So, I may keep my HTPC after all.

Dave,

You have a point here. The resolution statement for DVD should be "up to 540 lines". The clarity of the tiniest details would depend on whether the sample covered them exactly (a black pixel would be a black pixel) or they were in the border of two adjacent samples (a black pixel would translate to two gray pixels).

This is why 3MPix digital cameras that have over 2000 horizontal pixels are said to have horizontal resolution limit of a little over 1000 lines (across the whole width).

GK
George Kouzev is offline  
post #7 of 16 Old 05-03-2001, 11:07 AM
Advanced Member
 
Dave W's Avatar
 
Join Date: May 2000
Location: Syracuse, NY, USA
Posts: 579
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
I find this subject very interesting http://www.avsforum.com/ubb/smile.gif.

It took me a while to realize that pixels have a different meaning in the context of video than they do in the context of computer data. In the context of video, pixels ARE the actual samples. In fact, I think video is a beautiful demonstration of sampling theory. Display a DVD in its native resolution (720x480) with no scaling. Walk up to the screen and you can see the pixels (samples), but as you slowly step back the pixels eventually disappear, and your left looking at a beatiful picture. You're probably near the effective Nyquist frequency of your eyes at that point. I don't know if there's any analogy in audio. I suppose slowing down the audio to bring the high frequency samples down into audible range is similar.

Unfortunately, DVD resolution is pretty low, so if we were stuck at the position where we're seeing the resolution at the Nyquist frequency of our eyes, we'd be left looking at a relatively small picture. That's why we all want scalers to create more interpolated pixels. You don't get more actual resolution, but at least your not staring at the samples anymore. We're trying to create a picture that looks like it was created from sampling at a higher rate so that we can make it bigger without seeing the pixels (samples).

Now, that gets back to my original point that a DVD can contain no more than 360x240 pixels of actual resolution. Perhaps pixels is the wrong word. Lines of resolution might be better. Think about the following experiment:

Create a source photo that is composed of increasingly finer interleaved vertical black and white lines. Then use a 720x480 digital video camera to create a video panning across the image left to right. If you directly put that data onto a DVD, you'd see nasty aliasing artifacts, particularly as the finer lines went in and out of phase with the pixels on the imager of the digital camera. It would look terrible, and it would drive scaling algorithms crazy. But if you applied a filter to the data that dropped the actual horizontal resolution to 360 lines and made a new DVD, it would look smooth as it panned, but the finer lines would be blended into solid grey. That's the compromise that must be made when mastering a DVD. If you include too much high frequency content, there will be aliasing artifacts, and if you filter it all out, you get a soft picture with no fine detail. That's why they call it edge enhancement. It shows up wherever there's an abrupt change across just a couple pixels that hasn't been filtered out. It looks artificial because it shouldn't be there. A properly made DVD should be able to be displayed at any resolution above 720x480 without artifacts, provided quality scaling algorithms are used, because the scaling algorithm can make a good guess (or at least not a bad guess) what the new pixels should be. If two adjacent pixels are drastically different, the scaling algorithm doesn't know what the new pixels should be. It's the same idea as an oversampling filter in audio.

Dave


Dave W is offline  
post #8 of 16 Old 05-03-2001, 09:27 PM
AVS Special Member
 
Brian Hampton's Avatar
 
Join Date: Apr 2000
Posts: 6,846
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 129 Post(s)
Liked: 79
George,

I have looked at this pattern as well, particularly the 6.75Mhz lines. These lines are vertical. So...when they are smearing like you noticed, it's because the horizontal resolution is not 720. You found them clear at 720x540 only but I find them clear at 720x480 as well which is the actual resolution of DVD's. I don't know why they didn't include a similar circle for measuring and/or verifying the vertical resolution because this circle is easier to read than the wedge patterns for me.

I think you are right that at 720x540 the horizontal scaling is removed but to remove vertical scaling you need to use 720x480.

If I'm missing something here or I'm incorrect somehow please help me clarify this.

Thanks,

Brian
Brian Hampton is offline  
post #9 of 16 Old 05-04-2001, 06:50 AM
Advanced Member
 
daredevil23's Avatar
 
Join Date: Sep 2000
Location: Niagara Falls, ON, CA, Earth
Posts: 953
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
<font face="Verdana, Arial" size="2">Originally posted by Dave W:

Now, that gets back to my original point that a DVD can contain no more than 360x240 pixels of actual resolution. Perhaps pixels is the wrong word. Lines of resolution might be better.
Dave

</font>
Hard to explain it so just view this link, maybe all will be clear http://www.avsforum.com/ubb/wink.gif

http://www.cs.tut.fi/~leopold/Ld/Vid...ntalResolution

------------------
Married Men Live Longer Than Single Men, But Married Men Are More Willing To Die.

Married Men Live Longer Than Single Men, But Married Men Are More Willing To Die.
All Rights Reserved.
daredevil23 is offline  
post #10 of 16 Old 05-04-2001, 10:50 AM
AVS Special Member
 
Brian Hampton's Avatar
 
Join Date: Apr 2000
Posts: 6,846
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 129 Post(s)
Liked: 79
Daredevil,

Your link was interesting at least. But crt devices are not constrained to "square pixels" or 4:3 screens and most are capable of displaying the full 720x480 resolution offered by a dvd. In many cases the crt is capable of higher resolutions.

A HTPC can output the exact 720x480 information on a DVD. If a higher resolution is used like 800x600 the effects of upsampling can be seen in certain patterns like the Avia resolution patterns.

I have no experience with digital projectors but I realize the pixel structure is fixed and the aspect ratio of the screen is usually fixed as well. However, if the device has a higher number of pixels than the 720x480 offered by a dvd then a program like YxY could be used to map the DVD to the portion of the screen needed to display the full resolution properly.

Again, Please correct any misunderstandings I may have as I want to be sure of my thinking.

-Brian
Brian Hampton is offline  
post #11 of 16 Old 05-04-2001, 11:41 AM
Advanced Member
 
Dave W's Avatar
 
Join Date: May 2000
Location: Syracuse, NY, USA
Posts: 579
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Brian-

Digital projectors have no way to show the native 720x480 pixel DVD image at the correct aspect ratio. The projector has square pixels, and the DVD does not. Non-anamporphic DVDs have pixels that are narrower than they are tall, and anamorphic DVDs have pixels that are wider than they are tall. However, a properly mastered DVD will scale very nicely up to higher resolutions for display on a digital projector. DVDs with excessive high frequency content will show aliasing and edge enhancement on CRT projectors as well as digital projectors. CRTs are particularly bad in the vertical domain, since they are painting discrete scan lines. Digital projectors are discrete in both dimensions, and the scaling algorithms used tend to make the artifacts worse.

Dave
Dave W is offline  
post #12 of 16 Old 05-04-2001, 12:08 PM
Senior Member
 
RussSchultz's Avatar
 
Join Date: Feb 2001
Location: Austin, TX
Posts: 217
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Until your display resolutions is several multiples of the source image size, digital aspect ratio correction will introduce (potentially) icky scaling artifacts.

Either pony up the money for a panamorphic lens, or live with tall skinny people until we get much higher resolution displays. http://www.avsforum.com/ubb/wink.gif


------------------
What?! Me worry?

What?! Me worry?
RussSchultz is offline  
post #13 of 16 Old 05-04-2001, 12:14 PM
AVS Special Member
 
Brian Hampton's Avatar
 
Join Date: Apr 2000
Posts: 6,846
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 129 Post(s)
Liked: 79
Thanks for the input,

I think I may have a better understanding now.

Good luck,

Brian
Brian Hampton is offline  
post #14 of 16 Old 05-04-2001, 08:27 PM
Member
 
gabe nell's Avatar
 
Join Date: May 2001
Location: Seattle, WA
Posts: 157
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Dave,

I'm sort of confused by your sampling theory.

As I recall from my signals class, the nyquist theorem basically says the highest frequency you can capture using ideal impulse-train sampling is 1/2 the sample frequency.

So if a DVD has an image 720x480, then the highest _frequency_ that can be captured is 360 "Hz." In quotes because hertz is s^-1 units.

I believe what you said the second time is more accurate -- 720 pixels corresponds to 360 lines, because I believe a "line" is a black then white line.

So 720 pixels represents 720 pixels, which I beleive would translate to half as many "lines."

Gabe
gabe nell is offline  
post #15 of 16 Old 05-05-2001, 04:39 PM
Advanced Member
 
Dave W's Avatar
 
Join Date: May 2000
Location: Syracuse, NY, USA
Posts: 579
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Gabe-

Thanks, that's exactly what I was trying to say http://www.avsforum.com/ubb/smile.gif. 720 pixels corresponds to 360 lines (270 lines when normalized to image height), since a "line" has to complete a cycle from black to white. So the AVIA test image that shows resolution up to 540 horizontal lines of resolution is a contrived image that contains content way above the nyquist frequency. It's useful to measure CRT bandwidth, but it seems to me it has limited usefulness on a digital projector. If you're using a SVGA or XGA digital projector, obviously the projector has bandwidth capabilities way beyond that of DVD. Looking at high frequency resolution charts just shows the inevitable aliasing and scaling artifacts. Those artifacts do not show any weakness on the part of the projector, the source image is at fault.

Dave


Dave
Dave W is offline  
post #16 of 16 Old 05-05-2001, 09:10 PM
Advanced Member
 
daredevil23's Avatar
 
Join Date: Sep 2000
Location: Niagara Falls, ON, CA, Earth
Posts: 953
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Stacey Spears of at Secrets of Home Theater and HIFI has an excellent set of articles about DVD and sampling, etc...

http://www.hometheaterhifi.com/volum...eo-9-2000.html

Stacey writes:
"The sampling rate of luma (the black and white, brightness, portion of the signal) is 13.5 MHz, which provides 720 active samples per line. The sampling rate of chroma (the color portion of the signal) is 6.75 MHz. The Nyquist theorem states that the sampling rate must be twice that of the highest frequency so that aliasing will not occur. This limits the maximum luma bandwidth to 6.75 MHz (half of 13.5)and the maximum chroma bandwidth to 3.375 MHz (half of 6.75)."

------------------
Married Men Live Longer Than Single Men, But Married Men Are More Willing To Die.

Married Men Live Longer Than Single Men, But Married Men Are More Willing To Die.
All Rights Reserved.
daredevil23 is offline  
Closed Thread Home Theater Computers

User Tag List

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

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

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off