AVS Forum banner
1 - 19 of 19 Posts

·
Registered
Joined
·
222 Posts
Discussion Starter · #1 ·
Hello,


I have a sony kd30-xs955. It has a aperture grill screen type.

Lately I have been playing with the overscan and geometry through the service menu. I have a question in regards how overscan relates to the resolution and the aperture gill. It the above screen type what happens when resolution is change to lower, lets say from max 1080I or 720P to for example 480P. Does it mean that less points are painted in the aperture grill? So for example, every two or something like that? Also if the tvs are overscanning and we make some modifications to make the overscan smaller, does it mean that we actually have to loose some resolution because there are no points in the aperture grill? When the 1:1 mapping happens? With the overscan or with no overscan? What is the true resolution for such a screen? I feel like those screens must be doing the overscan to obtain the max resolution and 1:1 pixel mapping?

What do you think?


Bart
 

·
Registered
Joined
·
614 Posts
480i/p will have fewer pixels per scan line (I think 720 across) and fewer scan lines. Overscan causes a loss of resolution because pixels are scanned beyond the screen edge, to hide the very edges of the image where there often is stray lines/dots or cropped edges. Reducing overscan fits the image within the borders better. The phosphor grill on the screen limits your set to less than 1400 pixels across (480 to 1080 vertical.) My 34HS510 has about 800 across. I believe the screen just displays the best it can which is quite well. A new plasma or LCD TV can give every single pixel on stationary image.
 

·
Registered
Joined
·
222 Posts
Discussion Starter · #3 ·
So if they company advertises that their native res tv is 1080I (1720x1080) and we assume 5% overscan { 5% x (1720 x 1080 ) } the true (no resolution) by visible area would be approximately 1634 x 1026. Now, what I am curios about is what happens when we lower the overscan to lets say 3%. The aperture grill has RGB stripes where beams can paint colors:
http://upload.wikimedia.org/wikipedi...le_closeup.jpg

The beams have to paint colors by pointing at the stripe in the grill, and they beam have to be a specific size. So I guess changing the resolution and changing the overscan are two separate things:

- when we change the resolution we change the number of of horizontal and vertical lines that we draw on the screen. So if we have 1080 source signal and we change the resolution to something lower then 1080 there will be a translation between 1080I and the new output resolution. I guess the video card is responsible for that? In that case we will loose some quality so this would not be a good way to fix the overscan which is the thing that a lot of people do. We will loose 1:1 pixel mapping. I believe that even with the overscan the mapping is 1:1 but some visible pixels are hidden due to the overscan.


- if we change the overscan by changing the HSIZE and VSIZE we either change the beam size which does not make any sense because the aperture grill has a specific size so the beam must heat the exact point in the grill or must change the relationship between the source resolution and the point or points on the screen.(how many points are used by one pixel) In case of native resolution 3 stripes on the grill are used for one pixel? If we lower the resolution more 3stripes segments will be used to represent one pixel and they must be incremented by 8. Am I right? It would actually brake the 1:1 mapping? And now the question is when is that 1:1 mapping. I guess tv was designed to overscan and at specific overscan percentage the 1:1 mapping happens?


I am slowly getting to the point when fixing the overscan is a bad thing?

What do you think?


Bart


-
 

·
Registered
Joined
·
1,493 Posts
My thoughts:


The crt scans in the analog domain. It scans 1080 interlaced horizontal lines, vertically but the horizontal scanning is an analog scan, so individual pixels of the horizontal lines don't really exist at that point.


Overscan or position adjustments in the horizontal plane have no impact on the actual horizontal beam placement relative to the aperture grill. It simply moves the start/stop point of the line around.


These adjustments in the vertical plane changes the spacing between the 1080 lines. Since the aperture grill is made up of many vertical wires, no lines are lost, except for the ones that fall on the grill support wires, which is why you can see the shadows created by the wires and the ones pushed off screen by overscan.


I do agree that overscan creates a reduction in vertical resolution (lost lines) and horizontal resolution (lost "pixels").


All input resolutions are actually scanned to the crt at 1080i and maybe 540p. The DRC and MID circuits perform the required conversion processing.
 

·
Registered
Joined
·
222 Posts
Discussion Starter · #5 ·

Quote:
Originally Posted by raouliii /forum/post/15485718


My thoughts:


The crt scans in the analog domain. It scans 1080 interlaced horizontal lines, vertically but the horizontal scanning is an analog scan, so individual pixels of the horizontal lines don't really exist at that point.

What do you mean when you say that pixels do not exist? There has to be some kind of translation from digital to analog domain?

Quote:
Originally Posted by raouliii /forum/post/15485718


Overscan or position adjustments in the horizontal plane have no impact on the actual horizontal beam placement relative to the aperture grill. It simply moves the start/stop point of the line around.

If it moves the start/stop line and we shrink it, do we loose any information or it just all the digital pixels but represented in shorter width? This would mean that aperture grill have more resolution then input signal? My thinking is that if we shrink it, we must loose something?


Also in aperture grill the horizontal space between same color strips would be a limit of possible horizontal resolution, I guess? And I though they (RGB strips) could be treated as 1pixel?

Vertically there are no spaces so the resolution would depend on the size of the signal coming from the gun?

Is it possible that the tvs are designed so horizontal res has less limits then vertical? Thats why shrinking is possible while playing with the horizontal size?


These adjustments in the vertical plane changes the spacing between the 1080 lines. Since the aperture grill is made up of many vertical wires, no lines are lost, except for the ones that fall on the grill support wires, which is why you can see the shadows created by the wires and the ones pushed off screen by overscan.


bart
 

·
Registered
Joined
·
1,493 Posts

Quote:
Originally Posted by blwegrzyn /forum/post/15488125


What do you mean when you say that pixels do not exist? There has to be some kind of translation from digital to analog domain?......

Again, my understanding and limited ability to explain it. Internal to the set, analog inputs are converted to digital information for processing. The HDMI/DVI input is converted to analog and then reconverted to digital, oddly enough. Processing is done on the digital information and then converted back to an analog signal. At this point, the signal is not individually addressable pixels, it is a somewhat continuous stream of voltage levels that are scanned across the crt by the guns. There are individual lines but no individual pixels.

Quote:
Originally Posted by blwegrzyn /forum/post/15488125


.....If it moves the start/stop line and we shrink it, do we loose any information or it just all the digital pixels but represented in shorter width? This would mean that aperture grill have more resolution then input signal? My thinking is that if we shrink it, we must loose something?......

When the line is shortened, the same information is scanned, just in a shorter time. The number of grill wires fixes the number of phosphors that can be energized.
Quote:
Originally Posted by blwegrzyn /forum/post/15488125


.... And I thought they (RGB strips) could be treated as 1pixel?.....

Kind of, but they have no direct relation to the pixels as they existed on the HDMI input.
Quote:
Originally Posted by blwegrzyn /forum/post/15488125


....Vertically there are no spaces so the resolution would depend on the size of the signal coming from the gun?.....

Yes, the lines can be squeezed together or separated apart.
Quote:
Originally Posted by blwegrzyn /forum/post/15488125


....Is it possible that the tvs are designed so horizontal res has less limits then vertical? Thats why shrinking is possible while playing with the horizontal size?.....

The limits are the aperture grill. That's why the super fine pitch tubes are considered the best of Sony's sets. A tighter aperture grill spacing creates a finer detail of the analog scan line.


Hope this helped.
 

·
Registered
Joined
·
222 Posts
Discussion Starter · #7 ·
thanks for the great info,

are you aware of any links or books that explain that process in detail?
 

·
Registered
Joined
·
222 Posts
Discussion Starter · #9 ·

Quote:
Originally Posted by raouliii /forum/post/15488985


Yes, the lines can be squeezed together or separated apart.

One more thing, I want to make sure I understand the squeezing:

Lets say i have 10 lines and i squeeze them to 5.

Does it mean that during that process the analog signal was added together.

For example line 1 +2 = 1 line squeezed ?


Now , the horizontal squeezing: you say:
Quote:
Originally Posted by raouliii /forum/post/15488985


When the line is shortened, the same information is scanned, just in a shorter time. The number of grill wires fixes the number of phosphors that can be energized.

How come same information scanned but it shorter time results in shortened line? Is the process similar to my idea of adding lines?


thx
 

·
Registered
Joined
·
1,493 Posts

Quote:
Originally Posted by blwegrzyn /forum/post/15491353


One more thing, I want to make sure I understand the squeezing:

Lets say i have 10 lines and i squeeze them to 5.

Does it mean that during that process the analog signal was added together.

For example line 1 +2 = 1 line squeezed ?.......

No. Using your somewhat unrealistic example: The squeeze would simply result in the 10 lines occupying the vertical space that previously held 5. There would still be 10 scan lines. The grill doesn't restrict how the lines can be scanned across the tube (the spacing of the lines). This is the same type of squeeze available on most 4:3 HDTV tube. With an SD input, 480i was upconverted to 1080i and scanned on the entire tube. With an HD 1080i input, the raster is squeezed down to a 16:9 ratio that still has 1080 lines but they occupy a space that only had about 800 lines with upconverted SD. The top and bottom bars are not scanned at all.

Quote:
Originally Posted by blwegrzyn /forum/post/15491353


......Now , the horizontal squeezing: you say:
Quote:
Originally Posted by raouliii /forum/post/15488985


When the line is shortened, the same information is scanned, just in a shorter time. The number of grill wires fixes the number of phosphors that can be energized.

How come same information scanned but it shorter time results in shortened line? Is the process similar to my idea of adding lines?


thx

I'll change that quote a little: When the line is shortened, the same information is scanned, in the same amount of time, just in a shorter distance. The speed at which the gun scans across the tube changes.
 

·
Registered
Joined
·
222 Posts
Discussion Starter · #11 ·

Quote:
Originally Posted by raouliii /forum/post/15491797


I'll change that quote a little: When the line is shortened, the same information is scanned, in the same amount of time, just in a shorter distance. The speed at which the gun scans across the tube changes.

Does that mean that scanning is the process of translating from digital to analog signal?
 

·
Registered
Joined
·
1,493 Posts

Quote:
Originally Posted by blwegrzyn /forum/post/15495743


Does that mean that scanning is the process of translating from digital to analog signal?

CXA2170/IC2801 on the D board is the Y/C/CRT Drive processor that does the last D/A conversion. Its RGB output is sent to the CH/CX boards which drive the CRT.
 

·
Registered
Joined
·
10,501 Posts

Quote:
Originally Posted by blwegrzyn /forum/post/0


what happens when resolution is change to lower, lets say from max 1080I or 720P to for example 480P.


Also if the tvs are overscanning and we make some modifications to make the overscan smaller, does it mean that we actually have to loose some resolution because there are no points in the aperture grill? When the 1:1 mapping happens? With the overscan or with no overscan? What is the true resolution for such a screen? I feel like those screens must be doing the overscan to obtain the max resolution and 1:1 pixel mapping?

What do you think?

As others have already indicated, CRTs don't really offer 1:1 mapping the same way as "fixed pixel" displays. Generally speaking, the Sony HD tubes I've looked at support three different display modes: 480p, 960i, and 1080i. And all of the standard NTSC and ATSC resolutions have to be fit into at least one of those modes. Here's the way my 34XBR800 displays the different resolutions (and I suspect many of the other Sonys work similarly)...

NTSC/ATSC SIGNAL/RESOLUTION:HOW IT'S DISPLAYED:
480iEither as 480p (DRC Progressive or DRC Cinemotion), or as 960i (DRC Interlaced)
480p480p
720pConverted to 1080i
1080i1080i

480p and 1080i signals can be displayed essentially "as is" at their native resolutions. However 480i and 720p have to be converted to one of the other display modes supported by the TV. 720p is converted to 1080i by the TV's digital processing. And 480i is either deinterlaced to 480p, or effectively line-doubled to 960i, depending on how the Advanced Video Options (aka DRC) are configured in the User Menus.


I recommend avoiding 480p for regular viewing though because the progressive display mode tends to be harder on the eyes, and because the motion quality and vertical resolution are generally better in the interlaced display modes. This means avoiding 480p inputs in general, and using whatever interlaced DRC modes your TV has for 480i. If you have a progressive DVD player for example, then I'd suggest setting the Component output to the interlaced option, rather than progressive. (Or you could switch to an HDMI upscaling DVD player that has good 1080i output.)


You may see a bit more jaggies on the 480i signals with the interlaced 960i display mode, but most Sony users I've encountered here seem to feel that's an acceptable tradeoff for the other benefits you get with that mode. If your TV has other advanced DRC upconversion options for 480i than the ones listed above, then you may want to give those a try as well. But I generally recommend avoiding the DRC modes which display progressively.
 

·
Registered
Joined
·
10,501 Posts
As to your other question: Is it possible to lose resolution by making adjustments to overscan in the service menu?


The answer is YES... if you use some of the MID geometry controls to digitally downsize the picture, especially in a vertical direction. You can easily see this loss in resolution with DVE's vertical multiburst test pattern. This applies to all* signals/resolutions being put through the TV btw, including 480p and 1080i. Because they all pass through the digital DRC/MID circuits at some point. This is one of several reasons why I try to avoid tinkering with the MID geometry controls.


(*The one exception to this rule are TVs which have the "1080i scrolling bar fix" applied. In those particular cases, 1080i should bypass the DRC/MID circuits, so they should be uneffected by any of the MID geometry controls.)


If you stick primarily with the horizontal and vertical controls in the 2170D-1 and 2170D-2 groups for adjusting overscan instead, then you shouldn't run into this problem, because those controls simple move, squeeze and stretch the video raster in different directions without reducing the picture's overall resolution.


For horizontal sizing and positioning, I use 2170D-2/HSIZ and HPOS. For vertical adjustments, you can use either 2170D-1/VSIZ and VPOS, or 2170D-1/ASPT and SCRL.


VSIZ and VPOS are global controls. ASPT and SCRL are signal and zoom/stretch mode sensitive, meaning they have different settings depending on what type of signal and zoom/stretch mode you're using.


So if you're just trying to reduce the vertical overscan a little for all of the different resolutions and display modes on the TV, it's probably gonna be easiest to use 2170D-1/VSIZ and VPOS. I use ASPT and SCRL alot myself though because these controls do not seem to be as closely intertwined with the other vertical geometry adjustments as VSIZ and VPOS. So adjusting ASPT and SCRL seems less likely to produce unintended warping or other distortions in the picture.


The only resolution I use on my TV is 1080i, which makes all of the geometry adjustments alot simpler (since I don't really care what happens to the other resolutions).


Although I don't necessarily recommend this to others, I try to keep the picture on my TV as close to 0% overscan as possible, or even slightly underscanned, because I like how much sharper the image looks with the raster reduced in size on my 34" screen. And I like to see what's goin on near the edges of the screen just as much as the middle. It takes alot of work to get all the other geometry controls configured just right to make that work though. (A ruler can definitely help with that.) And if you're using a smaller 30" tube, then you may not need/want the picture much smaller or sharper than it already is.


For the record, I've never been able to get all of a 1080i image to completely display on the screen. I can of course reduce the raster so that the picture is smaller than the height/width of the screen. But regardless of how I configure the controls on the TV, there's always a small amount of the picture missing either on the top and/or bottom and sides of the 1080i raster.
 

·
Registered
Joined
·
222 Posts
Discussion Starter · #16 ·
ADU,


hmm, thats interesting, I did not know that the tv was converting 720P to 1080I. Would this mean that although the signal from the computer is sent as 720P it is still displayed interlaced but the tv is showing 720P mode?

This kind of does not make sense because I can see the difference in number of vertical lines between two?


What is you recommended procedure to set geometry , overscan and convergence? What would be the order? Which settings first , which last?


thx
 

·
Registered
Joined
·
1,757 Posts

Quote:
Originally Posted by blwegrzyn /forum/post/15519069


ADU,


hmm, thats interesting, I did not know that the tv was converting 720P to 1080I. Would this mean that although the signal from the computer is sent as 720P it is still displayed interlaced but the tv is showing 720P mode?

This kind of does not make sense because I can see the difference in number of vertical lines between two?


What is you recommended procedure to set geometry , overscan and convergence? What would be the order? Which settings first , which last?


thx

When you press "display" on the remote, the information bar is telling you the resoultion of the input source, never the output. The TV will always output 480 or 1080.


As far as adjustments, I would very first recommend reading the "Sony Service Codes" thread. Spend a few hours reading if you want to do this right. You need to at least understand the difference between universal vs resolution/input specific modifications. The universal settings MUST be done first, or you chase your tail. I would then start with overscan, leave it overscanned just slightly to allow for aperture grille temperature expansion, then geometry, last convergence.
 

·
Registered
Joined
·
222 Posts
Discussion Starter · #18 ·

Quote:
Originally Posted by ADU /forum/post/15518470


For the record, I've never been able to get the full resolution of a 1080i image to display on the screen. I can of course reduce the raster so that the picture is smaller than the height/width of the screen. But regardless of how I configure the controls on the TV, there's always a small amount of the picture missing either on the top and/or bottom and sides of the 1080i raster.

I also noticed that I can never get the full screen. I noticed that the screen is being cut off a little at the top when I send a test pattern. What is the cause of that? This would mean that the tv is not passing the vertical resolution test , right?
 
1 - 19 of 19 Posts
Top