Plasma TV as the primary computer monitor

[Enforcer]

How can I test this? I certainly don't see any lack of color information or visual anomalies playing games on my PC via DVI->HDMI adapter @ 1080p on my Panny 42"

Here is a photo of 2 RED lines:

Look at the top of the image : only 1 (red) subpixel should be ON in each row, but 6 are glowing from 4 different pixel (4 red+green+blue).

Could any of you address my question about usability?

I would say the result will be pretty bad. Burn-in will be main problem, poor image quality (on panny) will be the second.

 

[Sxotty]

Here is a photo of 2 RED lines:

Look at the top of the image : only 1 (red) subpixel should be ON in each row, but 6 are glowing from 4 different pixel (4 red+green+blue).

I would say the result will be pretty bad. Burn-in will be main problem, poor image quality (on panny) will be the second.

Maybe it is the opposite. Maybe it is b/c they are trying to stop burn in and that is the result.

 

[kyleb]

It doesnt make any sense for me either, but thats what i see right now.

What I"m wondering is, how did you come to your conclusion that the vertical line not appearing red is the result of coverting to YUV with 4:2:2 compression? You stated this as if it was a fact, but haven't provided any evidence to prove as much.

I'm also curious as to the merit of your suggestion that all G10 and G11 Panasonic plasmas suffer from the fault you illsutated. I gather that you are testing your own display, but have you done or seen testing of all models?

Can you provide any sources to back your claims?

Maybe it is the opposite. Maybe it is b/c they are trying to stop burn in and that is the result.

What means of attempting to stop burn-in are you suggesting would result in discoloring the line in question?

 

[Enforcer]

What I"m wondering is, how did you come to your conclusion that the vertical line not appearing red is the result of coverting to YUV with 4:2:2 compression? You stated this as if it was a fact, but haven't provided any evidence to prove as much.

It's quite simple actually. I think i cropped image too much...
Did you see first 4 top rows? They are different from the others:
..._ r _ _ R _ B R G _R...
others are:
..._ _ _ _ R _ B R G _R...
This 2 different patterns repeat every 2 columns, making some nasty "block" effect, making line thinner (7 subpixels) or thicker (10 subpixels).

I assumed this "aliasing" is an interference of 2 "pixel frequencies", 1920 pixel/line and 960 pixel/line (well, the terminology is questionable).

Ok, why R (V in YUV) is shifted? Why not to use the same V value for both pixels? One possible explanation is that the device was designed for MPEG2 playback, where U and V sampling position was 0.25, instead of 0.5 (center) in MPEG4 and H.264.

* "B R G" in image is the Y component (the brightness), which is correctly sampled in full resolution

I'm also curious as to the merit of your suggestion that all G10 and G11 Panasonic plasmas suffer from the fault you illsutated. I gather that you are testing your own display, but have you done or seen testing of all models?

I tested TH-50PZ700E (G10), TH-R42PV70 (G10) and TH-R42PV80 (G11).
Last 2 arent fullhd (1024x768) and cannot run in pixel to pixel mode, but they have the same problem (vertical red lines are GREY), and the same "block" pattern.

 

[kyleb]

You obviously have a lot of speculation going on in what you previously had stated as if it were fact. I agree with you that having red lines come out grey is a problem, but have yet to see how you came to belive the issue is the result of converting to YUV with 4:2:2 compression, or that the issue is inherent to all G10 and G11 Panasonic plasmas.

 

[Enforcer]

You obviously have a lot of speculation going on in what you previously had stated as if it were fact. I agree with you that having red lines come out grey is a problem, but have yet to see how you came to belive the issue is the result of converting to YUV with 4:2:2 compression, or that the issue is inherent to all G10 and G10 Panasonic plasmas.

I agree, my exaple with red line was not good enough.
Here is another try:
Image was made from red( 255,0,0 RGB) and green (0,130,0 RGB) vertical lines.
Both have about the same Y in YUV space.
Diagonal white line was made for reference.

2x:

4x:


As you can see, "grid" completely disappeared in first image, which proves half (960x1080) resolution for U and V.

 

[kyleb]

Assuming I understand correctly what the actual image is your using for testing and considering your explantion for why it is turning out wrong, shouldn't striping show correctly in your x2 image rather than showing the same effect as the first?

 

[Enforcer]

Assuming I understand correctly what the actual image is your using for testing and considering your explantion for why it is turning out wrong, shouldn't striping show correctly in your x2 image rather than showing the same effect as the first?

Yes, unfortunately the situation is even worse than "just YUV 4:2:2 compression".
Some kind of low-pass filer is applied to downsampled to 960x1080 U and V, probably to hide aliasing.

 

[Sxotty]

I am glad you noticed this enforcer. It is a curious issue.

 

[kyleb]

Yes, unfortunately the situation is even worse than "just YUV 4:2:2 compression".
Some kind of low-pass filer is applied to downsampled to 960x1080 U and V, probably to hide aliasing.

I'm still not seeing how you are conviced it is converting to chroma compressed YUV. How can you be sure it isn't simply low-pass filtering or perhaps some other image processing producing the results you have presented?

 

[Sxotty]

I'm still not seeing how you are conviced it is converting to chroma compressed YUV. How can you be sure it isn't simply low-pass filtering or perhaps some other image processing producing the results you have presented?

The important thing is it doesn't work to display images correctly.

Figuring out why is well and good for you folks, but people considering buying them just want to know whether it works properly, not why it doesn't.

 

[kyleb]

Surely you'd want to know why it is doing what it is if doing it "properly" instead would generally provide less visually pleasing results, while only looking improper in pathological cases such as the test patterns such as the ones shown here?

 

[AlphaWolf]

Surely you'd want to know why it is doing what it is if doing it "properly" instead would generally provide less visually pleasing results, while only looking improper in pathological cases such as the test patterns such as the ones shown here?

Does it look improper only in pathological cases or is it actually effectively reducing horizontal resolution.

 

[kyleb]

That is what I am wondering, and hence my intrested in understanding exactly what is happening rather than jumping to conclusions. However, my TH-50PH9UK doesn't demonstrate the same behavior as Enforcer has shown from his display, so I'm not rightly in a position to do any testing.

 

[Blazkowicz]

remember that with our human vision, we perceive color with much less precision than luma, so it's possible that you don't notice the halved horizontal color resolution at all in the majority of cases.

 

[kyleb]

Sure, assuming that is even what is actually happening there. Regardless, I made up an image with single pixel wide horizontal and vertcal red lines on a black background and they both come out red on my display, both when downsampled from a 1080p input and when running native.

 

[Mintmaster]

Surely you'd want to know why it is doing what it is if doing it "properly" instead would generally provide less visually pleasing results, while only looking improper in pathological cases such as the test patterns such as the ones shown here?

If it was just the first one, I'd agree, but seeing it on the 2x image is pretty bad, as that's not very pathological at all. Coloured text would look like crap.

Anyway, we know that this is only from the DVI output of a PC (using a DVI-HDMI adapter, I assume), and if his theory is right, the VGA input would behave the same way on the 85U models.

I don't have a HDMI video card, but how good does component output look?

Also, any more opinions on my usability question?

Regardless, I made up an image with single pixel wide horizontal and vertcal red lines on a black background and they both come out red on my display, both when downsampled from a 1080p input and when running native.

Maybe it's only the Asian/European models. Different electronics?

 

[Enforcer]

That is what I am wondering, and hence my intrested in understanding exactly what is happening rather than jumping to conclusions. However, my TH-50PH9UK doesn't demonstrate the same behavior as Enforcer has shown from his display, so I'm not rightly in a position to do any testing.

Its good to know G9 is not affected.
Now I am curious how and why is this (strange behavior) started.
Here is what i found:
http://www.pasctraining.panasonic.com/pass/elearn/files/Plasma/070424_10th_Gen_PDP.pdf

2006...
2007...
The 2007 42” plasma models have less boards than the previous generation. The PA, DT, DG, H, and HC boards used on the 2006 42” models are now incorporated within the new A board of the 2007 models.

NTSC, ATSC, and QAM television reception is performed by the A board. The unit also contains two video inputs, two HDMI inputs and two component inputs that are directly connected to the A board. The A board selects the video signal for display on the screen. It is responsible for converting all inputs into YUV data for video processing. The A board performs PIP (Picture in Picture) and picture control operations such as brightness, contrast, color, tint, etc. The PEAKS firmware of the unit also resides in this board. The RGB output of the A board passes through an LVDS (Low Voltage Differential Signaling) transmitter for conversion into serial data.

The HDMI receiver section of IC4510 converts the incoming HDMI signals to a YUV video signal.

Some facts from this document:
G10 share the same design.
New panasonics convert HDMI RGB to YUV and back to RGB.
The document does not approve or disapprove 4:2:2 compression.
The fault is localized: it resides in "A board".
There are some (very low) chance that it can be fixed by firmware upgrade. Actually at least the following conditions have to be met for that:
- (highly improbable) resolution loss has to be in IC8001 (peaks lite 2), not in IC4510
- should be enough bandwidth
- should be enough DRAM

 

[kyleb]

If it was just the first one, I'd agree, but seeing it on the 2x image is pretty bad, as that's not very pathological at all. Coloured text would look like crap.

That really depends on what is actually being done to produce those results. Having seen fairly fine colored text on current generation Panasonic plasmas which looked fine, I'm left to wonder what exactly is happening in Enforcer's examples.

Maybe it's only the Asian/European models. Different electronics?

Certainly possible.

 

[Bogotron]

Maybe it's only the Asian/European models. Different electronics?

Don't think so. I just tried this on my European 50PZ700.

Test pattern:

And resulting (cropped) output:

Now if you take into account camera artifacts and other settings that influence the image (color temp etc) that looks pretty identical to me. At least no downsampling of the chroma channels.

Edit: Forgot to add this is over HDMI from a G35 board.