Calibration FAQ ( under development) - AVS Forum
Forum Jump: 
Reply
 
Thread Tools
post #1 of 57 Old 09-28-2005, 10:28 AM - Thread Starter
AVS Special Member
 
Tryg's Avatar
 
Join Date: Aug 2000
Location: Washington
Posts: 9,743
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 3 Post(s)
Liked: 11
Display Calibration FAQ sticky

Thanks to Chris Wiggles and any other volunteers: we appreciate your efforts to put together a FAQ for this forum

Chris Wiggles Go-to Guide for Setting Source Options Link
Tryg is offline  
Sponsored Links
Advertisement
 
post #2 of 57 Old 10-10-2005, 02:42 PM
 
ChrisWiggles's Avatar
 
Join Date: Nov 2002
Location: Seattle
Posts: 20,730
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 15
say! I will definitely help anyone who is interested with this, I am just quite busy at the moment, so I can't take lead to write the whole thing as I did with my previous guide. Maybe I'll have some free time in coming weeks, but it is unlikely. Further, I am not that well versed in tricks for the many kinds of displays out there, nor grayscale particularly. But definitely let me know what's going on and I'll lend help however i can! I also have a computer get wiped out, which had all my links and stuff to old threads, which would help a lot with this, unfortunately :-/

-chris
ChrisWiggles is offline  
post #3 of 57 Old 10-17-2005, 12:10 AM
 
ChrisWiggles's Avatar
 
Join Date: Nov 2002
Location: Seattle
Posts: 20,730
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 15
Quote:


Periodically, I hear people struggle with setting contrast. In particular, people get hung up on looking for blooming, avoiding geometry distortion, or thinking that the AVIA disc is only for CRT displays. Here's the five minute summary of the problem of setting contrast (white level) for newbies. It's hard to get it all because there are so many permuations of displays and what to look for. Hence, the on disc instructions could only cover the one - most common case in detail. Of course, a lot of forum members have non CRT based displays or have ones that differ in behavior. In general, here are some goals.

If you have a glowing phosphor based display like a CRT or plasma, one task is to make sure you don't have contrast set so high that you are rapidly burning the phosphor. You other task is to set contrast such that response to the signal is "linear." In other words, you keep contrast low enough that neither damage to the display or to picture quality result.

If you have a lamp based display like an LCD or DLP, you don't have to worry about avoiding damage by having too high a contrast, but you still have to avoid huring image quality.

On older CRT's an unsafely high contrast setting would cause minor defocusing of the electron beam aka blooming. This definitely was a bad thing so we tried to teach people to look for and AVOID ever letting the display run that high. Newer CRT displays don't bloom as much even when contrast is run up high enough to accelerate wear. So, you don't always see blooming and that is a bit confusing if someone new to video adjustments gets caught up in looking for blooming when actually the second part of the advice is the most important - Keep the contrast as low as possible to prolong phosphor life. In some modern CRT RPTV's you won't see obvious blooming but that does NOT mean it is good to keep contrast cranked up all the way.

Geometry distortion due to power supply problems are less common now. I wouldn't really worry about that on today's CRT's. When it's present and due to things like scan velocity modulation, newbies often can't get rid of it even if contrast is turned down too low.

On lamp based displays, having contrast doesn't shorten display life but causes clipping of hilites or shifts in the color of white as one primary color or another runs out of dynamic range. Clipping of hilites means that things that are bright but not quite white become indisinguishable from white. Shifts in the color of white means the color of white changes. On DLP's that shift is often blue-green.

A pro calibrator knows to set contrast to get proper light output, keep within safe limits, and obtain good grayscale tracking. Teaching the consumer to do that is hard because....

1. The consumer usually has no means of objectively measuring video display light output. Only a few advanced enthusiasts go so far as buying a colorimetry system like the OpticOne. I could have said on the disc to adjust contrast until white measures something like 12 to 16 footlamberts on a direct view or front projection and around 8 to 10 FL for RPTV CRT. Those aren't targets written in stone. Some people intentinally target higher, but they still do so keeping an eye on what is safe for the equipment. Most Avia users don't have a means of measuring how bright is 12 FL. Avia PRO users, yes, but not AVIA users.

Without a metering system, the best I could say is that the target brightness of white on screen is very roughly how bright a 4 white LED Lightwave 2000 flashlight lights up a white sheet of paper at 28 inches. That is fraught with uncertaintly and isn't that standardizable. People don't have that flashlight and even if they did, the flashlights vary in light output unit to unit and the battery freshness affects things. No matter what I have thought of in common objects as a possible reference, lots of factors ensure that what you see end up with for light from that object won't match what I am getting.

2. The measured light output level needed to deliver the desired light intensity to the viewer varies with display system. A pro calibrator knows that RPTV's have a high gain screen and need a lower direct measured light output to attain the same visible light level as a direct view. They also know that front projection systems also have screen gain that affects how measured light intensity is seen by the viewer. All those additional situations would take much more teaching than is practical on a consumer disc. If we tried to teach it on disc, almost no one would understand. People get confused with just the simple basic instructions on the Avia disc - and that was about as simplistic as I could get it.


So this leaves us with the final portion of my advice for setting contrast - keep it as low as consistent with a good image in a fairly dimly lit room in the evening. If you have contrast high enough to compete with sunlight or bright room lights, you are probably too high. Many people are used to really high light levels from a TV. Indeed, my eyes hurt the last time I saw an RPTV at someone's home because the light output was too high. On a home theater display, the light output when set correctly and viewed with DARK ADAPTED eyes will just begin to feel too bright when the scene abruptly switches from an indoor shot to a bright outdoor shot. That's somewhere around white being about 12 to 18 FL when I measure it against my eyeball "too bright tolerance.

By keeping contrast lower, you are more likely to be in the physically safe and image preserving portion of the display's dynamic range. I wish there were a standard object or light out there that I could say "make white on your screen look as bright as object x." I haven't found anything readily available to consumers that has constant enough of a light output and close enough to D65 in color to recommend as a standard. As I mentioned above, even LED flashlights won't work. They change with battery condition. Fluorescent tubes won't do either because they vary in brightness with age. I once tried to describe the brightness of white in terms of candle flame brightness cast upon a sheet of paper. That wasn't too accurate and I doubt a single soul took it seriously.

So bottom line - Don't perseverate on looking for blooming or geometry distortions on the test patterns. Instead concentrate on keeping contrast down LOW. Not so low that viewing the picture in a darkened room still results in a dingy image, but as low as still consistent with a bright enough picture in a darkened room. That gives you the best you can do until you have a colorimeter or a pro can come by and get you set up truely right.


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO


AVIA Calibration Tip #1 - Needle Pulses + Steps Video Test Pattern

Needle Pulses + Steps Test Pattern

At first glance, one mistakes this pattern for a traditional Needle Pulse pattern, but there are important changes in the pattern which make it more useful. Most obvious is the addition of a vertical line on the right side of the pattern. The traditional pattern only has a line on t he left side of the pattern. By adding the right side line, AVIA makes it possible to look for closely for high voltage supply problems. Often geometry distortion with high voltage system overload appears worse and earlier along the right edge. Without the right side line, traditional Needle Pulse patterns don't allow you to visualize the change. You might be fooled into thinking a display has perfect geometry stability when in fact it is distorting the image.

Another obvious addition are the gray scale steps in the upper half of the pattern. On CRT displays one looks for both geometry distortion AND blooming in order to find the maximal usable white level setting. This used to require two separate test patterns. By adding the gray scale steps, AVIA makes it possible to look for both changes with a single pattern. No more switching back and forth between patterns.

If you also observe the upper and lower halves of the Needle Pulses pattern closely, you'll notice some bars moving back and forth. The "Black Bars" in the upper half of the pattern are useful for checking the black level (brightness) setting of your display. The "white bars" in the lower half of the pattern are for testing white clipping level of LCD projectors. This means a single pattern allows for testing of three limits of white level and a quick check of black level.

For those who prefer a more traditional pattern, AVIA also includes the Needle Pulses pattern which omits the gray steps but includes the black bars and white bars. I can't really imagine why one would want to omit the test for blooming, but AVIA provides you the choice.


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:17 PM
Post (#5 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



AVIA Calibration Tip #2 - Blue Bars, Red Bars, Green Bars

Color bars are useful for checking the proper function of color decoders in a display. With NTSC displays one can vary the saturation (amount) and hue (phase relationship) of the display. Ideally, the display is adjusted to recreate the same colors as encoded in the signal.

The traditional SMPTE color bar includes not only bars but small patches of colors which are in reverse blue order just below the color bars. By examining the amount of blue (using a filter or better yet by using the display's blue only mode) one can tell if the saturation and hue are correctly adjusted. This works because the gray portions of the pattern are encoded to be an intensity of gray which has exactly the same amount of blue in the blue portions of the pattern. Since gray contains no color content, turning the saturation control up and down does not affect its blue content. This allows the gray to act as a reference point against which to compare the amount of blue is present. Calibrating saturation can thus be done by adjusting blue intensity to match the fixed amount of blue in gray. Hue is observed by comparing the amount of blue in the magenta and cyan portions of the pattern. When hue is correctly adjusted, the blue intensities of magenta and cyan are identical.

Unfortunately, some people find it difficult to accurately tell when the blue intensities are equal. The bars and patches are of unequal size and color separation artifacts can make the color transition zones blurred or of uneven darkness. For these reasons, AVIA's "Blue Bars" add flashing patches within the color bars to aid in finely discerning when the intensities are equal. Human vision is very sensitive to flashing. AVIA takes advantage of this by having users adjust saturation and hue to minimize visible flashing in blue. This allows higher accuracy than comparison of static bars and patches. For those who still prefer more traditional static comparisons, AVIA also provides traditional split color bar patterns.

Several other features are built into the color bars. You may not have noticed that the transitions between bars and patches is closer to the center of the screen than SMPTE bars. This moves the critical comparison area of the pattern further from on screen displays which often appear at the bottom of the screen when televisions are adjusted. The white reference rectangle at the bottom of the pattern includes animated white bars for detecting white clipping. The lower right black portion of the pattern has animated black bars for checking black level. These are positioned where the PLUGE pattern is on SMPTE bars, but animation avoids the optical illusions of aligned rectangle edges that sometimes make it difficult to tell if a PLUGE rectangle is visible. Animation makes visibility obvious. Also the black bars do not rely on a blacker than black component for proper use. However, the black bars in this pattern are only to be used on displays which need a high APL during black level adjustment.

One can also use SMPTE bars in red or green only, but AVIA makes evaluation of green and red primary handling by providing "Red Bars" and "Green Bars." These are used in the same manner as the blue bars, except one views the red bars in red-only and the green bars are to be viewed in green-only. The patches which need to be compared are also moved to be positioned below the bar against which comparison needs to be made. This makes intensity comparisons easier than the wide separations that arise with a blue optimized color bar pattern. AVIA also adds its innovative flashing patches to the red and green bars to enhance viewer accuracy.

Why red and green bars? If the color decoder is perfect, adjusting to blue only accuracy would make the red and green color bar patterns also appear perfect. Unfortunately, it often isn't perfect. We'll visit that in AVIA Calibration Tip --- Color Decoder Check.


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:17 PM
Post (#6 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



AVIA Calibration Tip #3 - Sharpness Pattern

NTSC video carries most of its resolution in the luminance portion of the signal. Onto this is overlaid lower resolution color information to yield the final picture. By doing this, the designers of the NTSC system were able to provide an image which gave much of the perceived effect of having high resolution in both color and luminance but in a smaller amount of bandwidth.

Television displays provide a sharpness or peaking control whose behavior is much akin to the treble control of an audio receiver. The control should be used to compensate for the attenuation of high frequency video information that blurs images horizontally. Unfortunately, this control is often misused or not designed in a way that accomplishes this goal. Users often keep sharpness set too high and suffer a picture that appears sharper to the naïve eye, but is actually filled with extraneous image artifacts. Sharpness is perhaps the most difficult control to teach people to set properly.

There are often recommendations to simply turn sharpness all the way down, but that can be excessive. It's best to actually use a test pattern which points out how the sharpness control is altering the image. Then you can rationally determine optimum setting on your display. AVIA provides a dedicated Sharpness pattern which combines several tests of parameters important in determining optimal sharpness setting.

1. A horizontal frequency sweep occupies the top of AVIA's sharpness pattern. This is a constant amplitude sweep that goes from low video frequency to high video frequency. If the video bandwidth of your display is lower in a portion of the video bandwidth, then that section of the sweep appears darker. Ideally, the brightness of the sweep is constant throughout its range. As you adjust sharpness up and down, look to see if any portion of the sweep goes up and down in brightness. That is the set of video frequencies the sharpness control of your display affects. If your display's sharpness control is well designed, all you need do is adjust to make the sweep as evenly bright as possible.

2. Some people find it difficult to compare different sections of a sweep because the gradation of brightness is continuous from one section to another. The frequency bursts at the bottom of the chart provide discrete patches of video frequencies that can be compared. You'll probably find that the rightmost patch is slightly darker than the rest of the patches even on the best of displays. Just try to equalize the other patches.

3. There are black vertical lines, diagonal lines, and a circle in the center of the AVIA sharpness pattern. These are needed because many sharpness controls not only alter frequency response but add ringing artifacts. Ringing is overshoot and undershoot of the video signal at abrupt luminance transitions. You see this as false outlines next to the actual black lines. If setting sharpness at the point which equalizes video bandwidth also yields visible ringing, you should decrease sharpness to the point at which the false outlining is just barely visible. Otherwise, you will be adding artifacts rather sharpening actual image detail.

4. There are also vertical lines set against black and white backgrounds. These are also used to look for ringing, but because the luminance transitions are larger than going from a gray background to black, these serve as especially severe tests for ringing. These are primarily for testing display circuit design quality rather than for actually setting the sharpness control, because these are usually too severe a test for consumer grade displays. Setting sharpness low enough to avoid all outlining of black/white transitions will often yield an excessively blurred image on consumer grade displays.

5. A vertical frequency sweep occupies the left side of the pattern. This is used in conjunction with the horizontal lines of the pattern to set vertical aperture (vertical sharpness) of video processors which have this control. As with the usual sharpness control, adjust to equalize video bandwidth and avoid false outlining.

Once you have correctly set sharpness, it is very tempting to return to your previous excessively high setting. Don't. Instead, view the picture several days at the new, less artifact inducing setting. You will find that the old overly high setting yields an unnatural picture.

Properly adjusting the sharpness control is only part of getting maximal image detail from your display. Keeping white level below the point of blooming and controlling room lighting should also be done. Once all these are done, you might wish to measure your display's resolution using resolution patterns which I'll cover in another AVIA Calibration Tip.


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:18 PM
Post (#7 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



AVIA Calibration Tip #4 -- Color Decoder Check

NTSC video signals must be separated, decoded, and matrixed to form the final red, green, and blue signals which drive the display. Professional grade displays accurately decode the color signals and render colors correctly. However, consumer grade televisions often break the rules and have non-standard color decoding. This is most often seen as exaggerated reds (red push) and wreaks havoc when one attempts to adjust colors on a consumer display using just color bars.

Color bars are encoded such that the amount of red, green, or blue is 75% in each bar which contains the color. For instance the amount of blue is 75% in the gray, blue, cyan, and magenta portions of color bars. Similarly, the amount of red is 75% in the gray, red, yellow, and magenta portions. Because the amounts of each primary are identical in the various patches, one can compare the intensity of each color to learn how a decoder is functioning. AVIA also includes 50% & 100% color bars for testing of color circuitry linearity but we'll ignore those for now and concentrate on the more commonly used 75% variety of color bars.

75% Gray has zero color difference from gray so adjusting color saturation up and down doesn't alter its appearance. Hence, gray serves as the reference point against which the intensity of color saturation may be compared. Turning saturation up and down alters the intensity of the colored portions of color bars. View the blue portions of color bars in blue-only as you increase saturation. You'll notice that blue increases in intensity with increasing saturation. When saturation is correctly set, the intensity exactly matches that of gray. On a professional display with NTSC accurate color decoding, this same saturation setting also makes the red and green portions of the pattern match gray. Hue is adjusted by comparing portions of the pattern that contain two primaries such as cyan vs magenta.

AVIA has a Color Decoder Check pattern which lets you measure and compensate for non-standard color decoding. The pattern has a gray background against which you compare the brightness of red, green, and blue color patches. The patches range from +25 to -25%. If the color decoder is perfect, then the 0% patches of each color match the gray background (when viewed in only that color). If a display has red push, then a higher (darker encoded) red patch matches the gray background. You can read the percentage push by finding the patch which best matches the gray.

You may find other imbalances with the AVIA Color Decoder Check pattern, but red push is the most important to control. This is because red push is more objectionable to most viewers than under push or green push. A professional calibrator can sometimes correct the color decoder axes to achieve NTSC standard decoding, but for most sets that is not possible. You may want to check the accuracy of color decoding of a display prior to purchase since this problem is often not correctable. The only recourse is to hide (not correct) the error by decreasing saturation to make the measured red push 10 to 20%. This desaturates the overall picture but avoids making flesh tones too orange. Leave hue alone when making this compensation.

There are two other things to remember when considering non-standard color decoding. Don't confuse correcting the color decoder axes with resetting gray scale. This problem cannot be corrected by decreasing red drive because that would alter the underlying gray scale of the picture. The problem is with the way color DIFFERENCES from gray are being interpreted by the display's color decoder, not with the amount of red in the gray scale.

The second thing to remember is that this pattern and color bars are most accurate if one turns off the other two color guns of the display when examining each color. Color filters leak through a bit of the other colors and falsely make the gray background brighter than it really is. This tends to make your observations through filters about 5% lower than if color filtering were perfect. The difference is small, but if you want highest accuracy, turning off or capping the other two color guns is best.


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:18 PM
Post (#8 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



AVIA Calibration Tip #5 -- Resolution Patterns

Look in your local newspaper advertisements and you'll see amazing claims of 800 to higher lines of resolution. Unfortunately, you'll be hard pressed to find a salesman who understand how those numbers were measured or how they relate to the display's actual ability to display fine image detail. One problem is that there is no legal standard for how the number must be derived and reported. A manufacturer could theoretically bypass most of the display's electronics and simply measure the performance of the best amplifier section. In the final analysis, you need to either rely on magazine reviewers to measure true TVL or learn a little about the resolution patterns in AVIA and measure it yourself before you buy a set. The test patterns in AVIA can serve as a powerful means of separating the chaff from the wheat when you're shopping.

TVL (TV Lines) resolution is expressed as the number of vertical black and white lines (counting both black and white) which can be resolved across a width of screen equal to the height of the screen. This differs from computer graphics in which horizontal resolution is measured across the entire screen rather than just a width equal to the height. For a 1.33:1 ratio screen (4:3), this means that TVL resolution is 3/4 the number of discernible lines across the entire screen width. On a 1.78:1 ratio screen (16:9), the TVL resolution is 9/16 of the number of discernible lines across the entire screen.

Exercise: You are writing advertising and press copy for a new projection television. What resolution number would you state, if your engineers tell you the following?

1. Lenses in the optical system have a maximal theoretical resolution of 1100 lines.
2. The final video amplifier stage (bypassing most of the electronics) can resolve 800 lines across the entire screen.
3. The overall electronics & optics resolves 600 lines across the entire screen width
4. The TVL resolution is 450

Remember, there aren't any legal requirements on how you report the resolution. Would you say that your display only has 450 lines of resolution?

So now we see how "lines of resolution" can mean different things if one doesn't specify how they are being expressed and measured. For actually viewing images, the overall TVL resolution is the most meaningful.

Some people are confused because they also know that NTSC video has a fixed number of scan lines of which about 480 are visible on screen. If the number of scan lines are fixed, then how can lines of resolution vary from set to set? The answer is that TVL resolution is measured along the HORIZONTAL direction, not the vertical direction. It's a totally different performance parameter. As long as the deflection and synchronization circuitry is working correctly, you'll always have the correct number of scan lines (ignoring overscan). The TVL resolution, on the other hand, is critically affected by the quality of the video circuitry and how high frequency video information is preserved. If video bandwidth is limited, then fine image details blur or are completely lost.

I should also mention that TVL resolution refers to luminance resolution in NTSC video. This is because NTSC video carries most of the fine picture information in the luminance channel and very low resolution information for color transitions.

Now it's time to examine AVIA's resolution patterns. There are three resolution patterns supplied in AVIA. They are the WSE Resolution, 100 TVL Resolution, and 200 TVL Resolution patterns. Along with AVIA's multiburst and sweep patterns they are used to measure how well a display resolves fine detail.

The resolution patterns are similar so we'll cover them together.

The pattern has a gray background and markers which indicate percent overscan. The innermost rectangle represent 5% overscan from each edge. In the very center of the pattern is a zone plate pattern consisting of concentric circles of varying spacing. Four resolution wedges are also in the inner portion of the pattern. The outer four circles serve as corner geometry checks and also lines which are spaced at frequencies of special importance.

The Zone Plate portion of the pattern is a test of color separator performance. If you play this pattern through the composite input of a display, the varying angles and spacing of the concentric circles severely test the ability of the display's color separator (often a comb filter) to cleanly separate color and luminance portions of a composite signal. Perfect separation would produce concentric circles without any coloration or disappearance of sections of the concentric circles. You can see how perfect separation appears if you view this pattern via S-video or component connections. Most comb filters produce some cross color (rainbowing). The better the comb filter, the less cross color is present. Some displays cheat and include a notch filter which removes information near 3.58 MHz. This can avoid cross color but dramatically limits image detail.

The following assumes a COMPOSITE connection so you can test the color separator.

The 3.0, 3.58, 4.18, and 6.75 MHz corner circles all contain vertical lines spaced at that frequency. 3.58 MHz is the NTSC color carrier frequency and serves as a very severe test of color separation. 4.18 MHz is near the maximal broadcast resolution, and 6.75 MHz is the maximal resolution possible on DVD video. Many consumer displays don?t have sufficient resolution to display the finely spaced vertical lines in the 6.75 MHz circle. Front projection systems usually have enough TVL resolution to show the individual fine vertical lines.

The 3.0, 3.58, and 4.18 MHz circles also have their lower halves filled with diagonal lines at the same horizontal frequency. This is another test of comb filter type. There are several designs of comb filters available and they differ in their ability to avoid cross color. Fairly simple comb filters (1 or 2 line) can avoid cross color on vertical lines (even at the worse case 3.58 MHz), but cannot avoid cross color on diagonal lines. It takes a high grade "3D" comb filter, which uses information from more than one video frame, to avoid cross color in diagonal lines. This is why I included diagonal lines in the lower halves of the corner circles. BTW, the 6.75 MHz circle does not have diagonal lines in the lower in its lower half.

Vertical and horizontal wedges of lines (actually sinusoidals) which converge closer and closer together are used to measure resolution. Use either of the vertically oriented wedges to measure horizontal TVL. Find the point in the wedge at which the lines cease to be separate and blend together. Then read the scale markers to find your display's actual TVL resolution.

You could use a similar pattern (SMPTE Resolution) in Video Essentials. However, that pattern is older and ends at 500 TVL so it cannot test if your display is fully resolving DVD's maximal detail because the pattern never attains the limit. AVIA's resolution patterns reach the full DVD video limit or 540 TVL (6.75 MHz) and allow you to see if your display is capable of showing the full degree of DVD resolution. On a related note, one should not use the VE multiburst or sweep patterns to check DVD player frequency response because both those VE patterns fall off rapidly at high frequencies. In contrast, the AVIA multiburst and sweeps maintain nearly full amplitude out to their limits because they were synthesized directly in the digital domain. They also contain built-in markers for -3 and -6 dB signal attenuation. This is important for setting up video processor and doing equipment reviews, but I digress.

The horizontal wedges are used to measure vertical resolution. Unless you display has markedly poor vertical synch timing, you should see the full 480 scan lines resolved. Some interlace flicker is likely visible in the horizontal wedges.

One final experiment is worth doing while viewing the resolution pattern. Watch the vertical wedges while you turn your display's sharpness control up and down. You'll see the line spacing (frequency) at which the control has its greatest effect go up and down in brightness. When sharpness is adjusted to flatten video frequency response, the vertical wedges have even brightness from top to bottom. However, you still may need to compromise video frequency response to avoid ringing artifacts.

Now for the differences between the three AVIA wedge patterns:

The 100 and 200 TVL patterns differ only in the starting points of their resolution wedges. You can use either, but the 200 TVL pattern was supplied to allow finer measurement of TVL resolution in the range of most consumer displays.

The WSE (widescreen enhanced) resolution pattern is used like the 100 and 200 TVL patterns, but the display must be in 16:9 enhanced mode AND the DVD player must be set to 16:9 shaped video display.



I'll leave you with two study questions: If you can solve these, you understand how TVL resolution is defined. The answers are below, but do try to solve these yourself.

1. Given that the pixel map in NTSC DVD video is 720 x 480, why is the TVL limit 540 on a 1.33:1 (4:3) screen?

2. Is the TVL limit the same on a 1.78:1 (16:9) screen?


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:19 PM
Post (#9 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



AVIA Calibration Tip #6 -- Subwoofer Test Tones and Bass Management

AVIA users have occasionally wondered why nothing comes out of their subwoofers during AVIA's subwoofer setup test tones unless their speaker sizes are set to "small." The test signals are working as they should, but AVIA is probably the first test material you've encountered which starkly reveals the effects of bass management. The subwoofer test tones in AVIA do not DUPLICATE the low bass information from the five main channels (front left, center, front right, surround right, surround left) into the LFE track. We could have duplicated the low bass information into the LFE track. This would have yielded subwoofer output whether one set speaker size to "large" or "small," but would also completely obscure how bass management is actually working. With AVIA, what you get at your subwoofer is only what the bass management system in the receiver is doing, not what we put in the LFE channel. Duplicating the low bass information into the LFE channel would make more systems APPEAR to function as users expected, but wouldn't clue them into what is actually happening in their system's bass management.

You may well discover that setting main speaker sizes to "large" completely stops a receiver's bass management from routing low frequency information in the main channels to the subwoofer. This means a lot of people are finding out that the bass management in their receivers isn't quite doing what they believed. Actually getting low bass information from the main channels to the subwoofer often requires one to set the speakers sizes to "small." Most users don't expect this to be true in their equipment and are shocked to find nothing coming out of their subwoofer when speaker size is set to "large." ----- > If your system behaves this way with the AVIA subwoofer test tones, this is also what it does to low bass in the main channels of a movie sound track. < ----- You may not have realized that you have been relying on only your main speakers to reproduce low bass in the main channels and the subwoofer has ONLY been producing low bass from the LFE channel. AVIA's subwoofer setup tones reveal this facet of bass management.

Many receiver and speaker systems are actually more appropriately set to "small" size despite their having "full range" speakers. This allows the subwoofer to receive and reproduce low bass from all the channels rather than only the LFE. The "large" setting implies that the speakers are capable of reproducing the pounding LF effects one usually gets from a subwoofer in a home theater setup. At least try it both ways to see which yields better results in your system.

Sometimes bass management limitations make it objectionable to set speakers to "small." Perhaps one has full range main speakers, a high crossover frequency in the receiver, and a sub which does not integrate well in the low-mid bass. The resultant gap in coverage is difficult to fix. A more flexible audio processor or some creative feeding to the sub of bass from main speakers (set to large) + bass from the LFE track may be needed. Two channel listening of music also comes to mind as a special situation.

AVIA's subwoofer setup tones reveal how your equipment's bass management system actually deals with low bass. This gives you a chance to understand and optimize bass management on your equipment.


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:20 PM
Post (#10 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



Black and White Level Adjustment

Let's first examine what the black level (brightness) and white level (contrast) controls actually do inside the set and the performance limitations that set the bounds within which you should keep the controls.

Within the television, there are video amplifier stages. Somewhere near the end of the video signal processing chain, just before the signal reaches the final amplifiers, the black level and white level controls operate. At the amplifier stage where these two controls operate, a small video signal undergoes amplification before being passed to the next circuitry stage. The higher the signal, the higher the associated beam current and hence more light output. (I'm ignoring the inverted polarity of the signal to simplify this discussion.)
Essentially, the black level control sets the baseline (bias) level of the amplifier stage. Think of it as setting what the lowest point (black) upon which the rest of video signal rides. The white level control sets how much amplification is performed on the signal. A higher white level setting means that video signal excursions are larger.

The combination of black level and white level controls allows you to control the amplitude and baseline of the video signal. If the display were capable of unlimited light output as video signal amplitude increases, one would only have to worry about light output and the blackness of black when setting these controls. Real life displays have limitations and exceeding these limits can damage or shorten a display's life.

I'll concentrate on CRT displays for now. The video signal is processed and eventually delivered to the red, green, and blue electron guns. A higher signal makes the gun emit more electrons and produce a brighter spot on the screen. CRT's are limited in how much beam current can be safely used. If too high an output is attempted, the phosphor at the front of the screen can be physically damaged by the electron beam. It isn't practical for you to measure the beam current so we use the proxy of beam defocusing to estimate when too much beam current is being used. As beam current increases, it becomes more and more difficult to confine it to a sharply focused beam. For most CRT's the point at which beam focus worsens is below the point at which phosphor damage can occur. By staying below this point of "blooming," you avoid immediate phosphor damage.

There are other limitations of usable beam current. Heat is produced by the current flow. Direct view CRT's use a metal mask to help direct the red, green, and blue beams to the appropriate color phosphors. A high beam current can heat up the metal mask and warp it. This is seen as a sudden shift in color of the screen and could indicate the danger of permanent mask warping. Obviously you want to keep your controls below this point.

Projection CRT's are driven at much higher beam currents, so high that liquid cooling of the phosphors is virtually required. This means that projection sets are even closer to the physical limits of the phosphors and particular attention must be paid to never overdriving the tubes.

Over time, phosphors age and solarize. The more light emission they are forced to produce, particularly if near maximal output, the faster their light output drops. This is why bright, fixed images can permanently burn themselves into a screen. By limiting beam current to reasonable levels, you slow this process and reduce the risk to your screen.

The electrons are emitted by a heated filament, the cathode, of a CRT display. The cathode is coated with special rare earth elements to improve electron emission. Over time, this coating loses its effectiveness and you notice this as a blurring of the electron beam even at low light output. The aging process causes a larger portion (other than the tip) to become involved with emitting electrons. Since the beam spot is essentially an image of the active region of the cathode, this increase in active region appears as enlarged (blurred) electron beam spot size. Higher beam currents accelerate this aging, but not to the same degree that it damages the phosphor.

Now we've gone over some reasons to keep white level down in order to protect the display. There are also imaging quality reasons. We've already mentioned the defocusing of the beam when current is too high. This blurs the image and reduces resolution. Also, running too high causes the relationship of input signal to output light to be altered. If this runs outside the "linear" range of the CRT, the relative brightness of signal levels from black to white are distorted. You perceive this as something being unrealistic or wrong with the contrast of a picture.

The high voltage supply used to produce the electron beam is often derived from the horizontal deflection circuitry of the television. If beam current demands are too high, the demand can drag down the horizontal deflection circuit. You see this as a horizontal geometry distortion. Hence the visible bending of the left and right vertical lines on either side of a Needle Pulses pattern in AVIA. Although not something that will damage a set, this type of geometric distortion degrades the image. By keeping white level down, you also avoid this problem. Some displays are designed such that this effect doesn't occur. In these displays, you never see the vertical lines bend, however you will still see blooming.

In short, white level should be set to avoid increasing the risk of permanent damage to the display, slow aging of the phosphors & cathode, and improve image resolution, gamma response, and geometric accuracy.

Black level, which determines the baseline upon which the video signal rides sets the appearance of black, not how much overall light is output by the display. This needs to be set at the lighting condition which is to be used for viewing because the correct setting varies with ambient light. Too high a black level washes out the picture. Too low a black level causes shadow details to be clipped and displayed as black.

Most consumer displays complicate setting of black level because they do not hold black level constant as overall picture level changes. That is, black is displayed differently depending on how bright the rest of the image is. This is also called imperfect DC restoration or clamping. The solution in this case is to bias the display with a moderate picture level image while setting black level. This allows you to arrive at a compromise level which works for most images.

Test patterns traditionally used a blacker-than black signal to help indicate when black level was correctly set. After all, you can't actually make the display any blacker than black so if "black" on the display is brighter than it should be the even darker (signal wise) BTB signal would appear be visible as a dark feature. When the BTB and black just appear identical, the display is correctly set to make black appear black.

Many DVD players and some video processors don't pass the BTB signal so this method of detecting when black level is correctly set doesn't always work. Also, the traditional patterns unfortunately aligned edges of pattern features with each other leading to optical illusions which confused viewers whether or not the BTB bar is visible. For these reasons AVIA uses "Black Bar" patterns to indicate correct black level. These are a pair of animated bars which move back and forth on screen. The motion makes it easy to see if a bar is visible and helps avoid the ambiguity of optical illusions of aligned, fixed lines. One bar is very near black, the other slightly brighter. When black level is correct, the darker black bar is just very barely visible. If black level is too low, one or both bars disappear. This allows one to find correct black level whether or not equipment passes blacker than black.

The situation is different on LCD's because the usual phenomena of geometry distortion and electron beam defocusing (blooming) don't occur with LCD's. Another limiting factor comes into play. LCD control circuitry in LCD projectors have a relatively abrupt point above which video signals become displayed as white. We call this "white clipping" on AVIA. electron guns.

Basically, if you set white level too high on an LCD projector you will find that near white details turn into white instead of something that is darker than white. Highlight details are hidden and the image looks solarized. Finding the white level setting which avoids white clipping is the key to maximizing LCD projector light output without degrading image quality.

As an aside, digital domain video displays can also exhibit a similar clipping effect when the video signal cannot be represented within the bit range of the system. You can sometimes see this on computer monitors displaying video from a DVD-ROM.

AVIA has new moving "white bars" in its main pattern for adjusting white level, the Needle Pulses + Log Steps pattern. You'll find a pair of near white bars which move back and forth. If you set white level too high on a LCD projector you'll see one or both of these white bars become white ( and disappear since you can't see white on a white background). The maximal white level setting is found by adjusting your LCD projector to just below the point at which the rightmost (brighter) white bar becomes white. Once that point is found, you know the max usable white setting for your LCD projector.

The Needle Pulses + Log Steps pattern in AVIA serves as a unified tool for both CRT and LCD display white level adjustment by combining tests for geometry distortion, blooming, gray scale linearity, and white level clipping into a single pattern.

Once the limits for white level are found for your display. The next step is to drop down to a white level setting which is below the max and still produces a white which appears white rather than gray.

Black level is set the same way with AVIA on both LCD and CRT projectors.


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:21 PM
Post (#11 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



Black Level Clamping (aka DC Restoration)

Why does the appearance of black vary with different scenes?

The appearance of "black" often varies depending on the remainder of the image on consumer grade video displays. Another more technical term for how stable black remains is "DC Restoration."

Let's review how the video signal is represented as an electronic signal. (Yes, I'm ignoring phase inversions in the amplifier stages for this discussion because it would confuse the central issue) The image information is an AC signal whose voltage level corresponds to how bright the image should be at that moment of the image scanning process. In American NTSC video, the luminance runs from 7.5 (black) to 100 IRE (maximal white). Synchronization pulses are negative to -40 IRE. Notice that black is represented by the voltage level being at 7.5 IRE rather than 0 IRE. In Japanese NTSC, black is represented by 0 IRE. The reasons and ramifications of this difference would be an entire other topic.

As the signal passes through the video system, it requires several stages of amplification. Each stage is coupled to the next usually via a capacitor. The capacitor passes AC signals but not the DC offset from relative to ground to the next stage. This means that the zero level of the signal is not actually transferred from stage to stage. You get a signal which varies up and down in voltage but trying to tell by looking at the final signal, you can't tell where black should be because you don't have an absolute reference of where on the waveform is black. The appearance of black to drift up and down with the average signal amplitude.

The above would be entirely unacceptable so the video circuits are actually designed with various means of "clamping" the black level at a known level. This process of "DC Restoration" is usually imperfect in consumer sets. Hence, the appearance of black darkens and lightens depending on the average picture level. This means usually means that black is a bit lighter during dark scenes than bright scenes. Professional level monitors and projectors include circuits which produce nearly perfect black level clamping.

It has been argued whether or not perfect DC restoration is desirable. Some viewers like black to be slightly lighter during darker scenes so they can better see shadow details in difficult night scenes. Others believe this compromises the intent of the scene.

Most of use have displays which have imperfect DC restoration. How should one deal with the calibration of a signal who appearance varies? The solution used in both AVIA and VE is to use calibrate black while displaying a test pattern whose average picture level is moderate. For instance AVIA uses Black Bars with a Half Gray pattern for setting black level. This means that black level is calibrated to a compromise which satisfies most viewing situations.

It is sometimes desirable to calibrate black level intentionally with a higher or lower APL test pattern. For instance, you may have a projector which has nearly perfect black level clamping but has some light scatter. A high APL image on such a display would tend to obscure the appearance of black bars. In this instance you may prefer to use a low APL black bar pattern. On the other hand, one could argue that a higher APL pattern would compensate for the light scatter.

AVIA supplies Black Bar patterns with black background or a half white background, color bars, and in the Needle Pulses patterns. You can follow the recommendation for most situations, but you also have other choices. Just think about your goals in selecting a different APL black bar pattern.

So in short, the varying appearance of black on your display is probably normal for your display and not an indication of improper function.


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:21 PM
Post (#12 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



Should I Set my DVD Player to Enhanced or Normal Black?

NTSC video signal levels are commonly measured in IRE units which can be converted to voltage levels. The conversion isn't important to this discussion. IRE units are easier to use so we all tend to refer to signal levels in IRE rather than volts.

100 IRE represents the brightest white. 7.5 IRE, rather than 0 IRE, is black due to historical limitation in equipment. This is true in North American video equipment. In Japan, 0 IRE is black. The offset from 0 to 7.5 IRE is referred to as "setup." Your DVD player adds the setup to the output signal to make "black" come out at 7.5 IRE if you use normal settings.

Your player also gives you the option of making black come out of the player at 0 IRE. This is the "enhanced" black setting. Why might you want to do that? Well, normal NTSC video has 100-7.5 IRE or 92.5 IRE of dynamic range from black to white. By selecting "enhanced" blacks the dynamic range is 100 IRE, a little bit larger of a voltage swing. Is this a good thing? It depends.

Your television must be told what level IRE is black. That's essentially what you are doing when you adjust the black level control (brightness). The AVIA test patterns include portions which are black and very near black so you can tell what you are doing with a known target. As long as you set the television to display black (whether it is 7.5 or 0 IRE) truly as black you're fine. This would seem to indicate that it's good to use the enhanced black setting since black gets correctly displayed. You get a larger signal dynamic range to avoid degradation, but there is a catch. The standard level for black for all your other sources is 7.5 IRE, not 0 IRE.

If your display allows you to independently set black level for each video source, then it's no problem to set the DVD video input settings to display 0 IRE as black. You'd then calibrate the other video input settings to show 7.5 IRE as black. Unfortunately, some displays don't let you independently set black level for each input. Getting one right makes the others wrong.

If the display is calibrated to display 7.5 IRE as black and you view an "enhanced" black signal, all the shadow details which between 0 and 7.5 IRE will be lost as pure black.

If the display is calibrated to display 0 IRE as black and you view standard American NTSC material which has black at 7.5 IRE, the picture will be somewhat washed out because the blackest black will actually be a dark gray.

The answer depends on whether or not your display remembers separate black level settings for each video input. If it can, then you could go ahead and use the enhanced black setting. You'll still have to calibrate your other inputs for 7.5 IRE black. For practical purposes, the increased dynamic range won't really make a difference because a no matter the dynamic range, the calibrated display still displays black as the same black and white as the same white.

Another pitfall may occur if you use video processors. Their inputs must also be configured to recognize black at the setup level you choose at the player. If you decide to use enhanced blacks, you need to think about how that affects the rest of your system. Selecting normal, 7.5 IRE black means less to worry about.


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:22 PM
Post (#13 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



Setting Speaker Level with AVIA

Let's go back and examine what is happening when setting a system to reference level. You are attempting to set sound reproduction level such that for any given sound you produce the same loudness as was heard in a mixing studio whose system is also set to reference level. Maximum SPL level is 105 dB, but a reference tone recorded at maximal SPL would be difficult to use so reference tones are usually recorded at least 20 dB softer. In the case of AVIA's level setting tests, the encoded sound should produce a measured level of 85 db SPL when the amplifier and speaker are set to reference levels. In a larger room, the amount of energy needed from the amplifier would be larger but the same SPL level of 85 dB would need to be attained. With VE, the test tone is recorded yet another 10 dB softer thus the target SPL measurement is also 10 dB softer, 75 db SPL when the system is at reference level. Each channel needs to be adjusted to attain that signal level for the entire system to be at reference level.

Now for the oddities. Unless your receiver is THX certified, the built in test tones are not guaranteed to be the right level to set reference level. In higher end processors, the built-in tones are set to the correct level such that they can be used to set reference level, usually targeting 75 dB SPL. You have to check with the manufacturer or manual to see what target SPL to aim for when using built-in tones. For all receivers with built-in tones, you can use the built-in tones to balance the channels relative to each other, but not necessarily to set absolute level to reference.

There are complications. We don't all have identical sounding speakers and room acoustics. Some systems have large speakers and satellites. The room may absorb some frequencies more from some speakers than others. For this reason, test tones for setting channel level are usually not pink or white noise. Instead, a shaped noise whose spectral distribution is centered at a frequency which most speakers can reproduce is used. That way the effect of non-identical speakers can be partially compensated. Room effects are more difficult to avoid. This is important when one goes from one set of test tones to another. Although both sets are encoded at the correct levels in a perfect system, real world speaker and acoustics can cause the measured SPL's to slightly differ. The frequency energy distribution would have to be identical between the tones to achieve exactly the same result. This is why your calibration can be a bit different between built-in tones and those in AVIA or VE. Nothing is wrong with your equipment, it is an effect of acoustics.

Where the master volume control ends up when you are at "reference" level differs from model to model of receiver. Some processors let you set the master to 0 dB (or it automatically sets master to 0 dB) and then you adjust the individual channel levels until they each produce reference level SPL's. In such a system, 0 dB on the master volume is reference level. On many receivers, one cannot place the master volume at the 0 dB point and still have enough range in the individual channel adjustments to achieve reference level. On such receivers, the master volume level for reference is whatever setting you had it at when you set your individual channels to reference. If that setting was -12 dB, then reference level on that receiver would correspond to -12 dB on the master volume.

A feature built into AVIA's tutorial probably adds some confusion. Listening at reference level is very loud. In most home settings, a master volume of about 10 dB below reference is more comfortable for listening. You'll note that the beginner's tutorial in AVIA has you initially balance your channels at 75 dB SPL rather than 85 dB SPL. This sets the system 10 dB below reference and automatically places you in an appropriate level for home listening. It isn't until the next section that you learn to target 85 dB SPL to actually be at reference level. I think it's even easier to simply calibrate all channels to reach 85 dB SPL, then turn down the master volume to comfortable listening level, and finally recheck channel balance. Most receivers track the channel levels correctly with each other as one alters master volume, but you may wish to recheck individual channel SPL levels at your usual master volume setting to double check that your receiver tracks all the channels together as master volume is changed. If you find the individual channels are not tracking together, you may wish to readjust the channels to make them match each other at your normal listening level.

On occasion, you may be told that your subwoofer should be calibrated to produce an SPL which is 10 dB higher than the other channels. There is no need to do this with AVIA's subwoofer test tones because they are recorded 10 dB softer to automatically produce a 10 dB higher subwoofer setting when you target the same 85 dB SPL. If one wishes to be technically more correct, the 10 dB offset actually is only true if one is using an RTA to measure sound levels. We're using an SPL which would read about 3 dB lower for the same subwoofer setting. If you want to be completely correct, place sub level 3 dB lower at 82 dB instead of 85 dB SPL.

Some receivers have built-in test tones and force you to listen to them during adjustments rather than using those from a test disc. In such receivers, one can do an initial setup with the built-in tones, then re-measure with tones from the calibration DVD. For each channel, one would then note the amount of error in dB. Next go back and adjust each channel the amount needed to correct the error. It may take a few rounds to get everything right if the receiver doesn't let you use external tones.

The LFE channel level is normally already set in the receiver. On a few units you can adjust the relative LFE level. If I recall correctly, DD has a suggested LFE offset of 0 dB, and DTS suggests a 0 dB offset for music and +10 dB for movie sound tracks.


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:22 PM
Post (#14 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



Using AVIA Phase Tests to Fine Tune Speaker Distance and Delay

AVIA's speaker phase testing signals are also useful for very accurate adjustment of speaker delays and distances. You'll need an analog RS SPL meter set to fast response in order to take advantage of this tidbit. This may seem a bizarre way to check delays and speaker distances but it is surprisingly accurate.

The phasing tests work by playing noise in the two channels being tested in phase and 180 degree out of phase intermittently. If the speaker distances and delays are both set correctly, then the in phase sounds from both speakers reinforce each other at the prime listening positioning. During the out of phase (diffuse) portion of the test, the sounds cancel. An SPL meter set to fast response can readily show the magnitude of the cancellation/reinforcement.

Start by playing the Phase left front/right front signal. Move your SPL meter slowly left and right at your listening position. If you have set distance and delays correctly the maximal SPL delta will occur in the middle of your sitting position. I get about a 6 dB needle bounce on my system. If it happens right of center, then your right speaker is either too farther away than the left speaker or delayed more than the left speaker. Conversely, if the peak SPL delta occurs left of your prime listening spot, the left speaker is too far or excessively delayed.

Once you have the front left and right speaker distanced and delayed exactly right, the SPL meter position at peak delta will be in the middle of your prime listening position. Note that position carefully. You'll need to be able to refer to that point within half an inch during the next step.

Now comes the trickery that gets the center speaker also precisely phased and delayed. The AVIA disc also has a Phase Left Front/Center test. We can take advantage of it to bring all three front speakers into very tight phase alignment. From the previous step we already know where the two front main speakers are in phase. Leave the left and right delays and speaker positions alone now. We'll next adjust the center speaker to be in phase with the left front. This places all three into phase.

Play the Phase Left Front/Center test and once more move the SPL meter left and right to find the maximal SPL delta point. Compare this new position to the one for the front mains. If all is perfect, they exactly coincide. If the left/center maximal SPL delta point is left of the left/right point, then the center speaker is either too close or insufficiently delayed. If the left/center max delta point is right of the left/right max delta, then the center speaker is too far. Move or adjust CENTER channel delay as needed to get the left/center max SPL delta to occur at the exact same place as for the left/right channels.

Your left, center, right speakers are now in phase. You'll probably note that a 1 msec adjustment in channel delay makes for a considerable shift in max SPL delta position. After all, that is about a 1 foot speaker distance equivalent. Use very small speaker movements to fine tune the center speaker into phase alignment.

Put your head at the center of the max SPL delta position and listen to some stereo and 5 channel material. You will be pleased with what has happened to sound imaging in your system.

Moving your speakers to achieve exact phase match isn't the entire story. One must also position the speakers with relation to room acoustics to smooth frequency response. Sometimes, moving speakers into exact phase also moves one or more of them into positions that yield uneven frequency response. In such cases, some compromise is needed to address both imaging and frequency response concerns. Happily, the home theater sound processor does have delays and these can sometimes help bring speakers into phase, while still keeping them closer to best tonal balance position.


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:23 PM
Post (#15 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



Most people obtain AVIA for calibrating their systems, but very few realize how a test disc helps separate displays from each before purchase. Ask nicely. You may get a chance to objectively evaluate a display instead of relying on how the store set up to make certain displays look better.

Here are a few things I like to check on each display?.


1. Power Supply Adequacy and Reserve

Display AVIA's Needle Pulses + Steps pattern. Turn the display white level to minimum and then back up noting when the white lower half of the pattern just changes from gray to white. Also note at what points the limits of blooming and line bending begin to appear. The more reserve available between the point of adequate light output to appear white and the blooming and geometry limits, the happier I am with the display's ability to provide enough light output when new and in the future.

If I start to see blooming even before sufficient light output is attained (in normal lighting conditions), I don't recommend the display. If line bending is only mild, I'll forgive the shortcoming.


2. Black Level Stability

Set White Level to correct viewing position (white just appears white and without blooming) using AVIA's Needle Pulses + Steps pattern. Then switch to the Black Bars + Half Gray pattern and set black level such that the black background is black and the two gray bars are just visible. Switch to the Black Bars pattern and the Black Bars + Half White pattern and see if the appearance of the black bars changes with average picture level. On a display with perfect black level stability, the black bars don't change in brightness as the rest of the screen changes.

Most consumer sets have imperfect black level stability. More modern sets often have superior black level stability. I don't disqualify a display with this test, but I remember it as a good characteristic if a display has perfect black level stability


3. Gray Scale

It's handy to bring along a type D fluorescent flashlight and a neutral gray card to do a quick check of gray scale. Note: Most type D fluorescents are slightly green relative to the desired D65 color of white. Display the Crossed Steps pattern and see if the display produces the same color of gray throughout the entire range from black to white. Next recheck using the gray ramp patterns and look for obvious banding.

A display's gray scale if usually too blue from the factory but this can usually be corrected by a calibrator so I don't worry too terribly much if this is off. However, some displays have digital image processing circuits that produce severe visible banding in gray ramps. I would shy away from displays with that problem as virtually nothing can be done to fix the problem. By the way, it's normal for rear projection screens to appear somewhat bluer on the left and redder on the right side of the screen. Simply take note of color shift severity.


4. Ringing

Display AVIA's Sharpness pattern and adjust peaking to make the bottom vertical line patches as equal in brightness as possible. The rightmost set is often slightly dimmer than the rest. Note the position of the sharpness control. Next pay attention to the vertical and sloped lines in the central portion of the Sharpness pattern and see if there is any ringing (false white lines) around the black lines of the pattern. Decrease peaking until false outlining is virtually gone.

A display with well designed peaking circuitry can almost equalize the brightness of the line patches while causing minimal false outlining. There is very little which can be done about a display which produces a lot of ringing other than to turn down the peaking control. If this also dims the more close spaced vertical line patches, it suggests the display has a problem with handling high frequency video components and I tend to avoid the display.


5. Horizontal Luminance Resolution

Display AVIA's 200 TVL Resolution pattern. Looks at the vertical wedges (with peaking set previously to avoid excess ringing) and see how closely spaced you can discern the vertical lines. Read the scale number too find the display's horizontal resolution in TV lines resolution. You many wish to turn down the color control to perform this test). Check both composite and s-video inputs to see if there is a difference.

A modern TV should resolve the entire 540 TVL of the pattern. If not, expect the display to soften and blur images. I find little reason to accept a display that cannot fully resolve as much image resolution as possible. If you are dealing with a cheap set with a notch filter, you may find a patch in the vertical wedges is gray instead of composed of lines. I would definitely avoid a display that had problems with luminance resolution.


6. Color Decoder Check

First set saturation and hue correctly using a blue filter and AVIA's Blue Bars pattern. (Be sure to turn OFF all auto flesh tone "features" on the display." The switch to AVIA's Color Decoder Check pattern and view it through the red, green, and blue filters. For each color filter, try to find the same color patch that matches the brightness of the gray background. The number next to that patch indicates the amount that color is exaggerated by the display's color decoder.

Severe red push is frequently found in consumer displays. This degrades color fidelity and often cannot be corrected. Some late model displays have adjustable color decoders which can be set to produce NTSC accurate decoding by a calibrator, but I would check before buying. If you get a display which has red push and lacks the ability to have its color decoder adjusted, you're stuck with only one solution -- dropping overall color saturation.


7. Gamma Check

Display AVIA's Gamma pattern and find the gray patch which best matches the background in brightness. An NTSC display should be between 2.2 to 2.5. Otherwise images will appear too light or too dark even if black point and white point are correctly set. If a display has gamma problems, I avoid it because you usually cannot correct the problem.


8. Overscan and APL Bounce

Display the Overscan Bounce pattern and see if image size changes as APL goes up and down. A perfectly stable image without excess overscan is desirable.


9. Electron Beam Focus and Convergence

Display the 100 IRE Dot Hatch pattern and examine all the dots throughout the screen. They should be sharply focused and a single white dot. If convergence problems are present you'll see the dots split apart into their component red, green, and blue parts. Also check the vertical and horizontal lines for proper convergence.

Convergence and focus problems can often be minimized by a professional calibration. I am most concerned if there are marked differences in focus between corner and center of the display as this may indicate a focus problem which cannot be fully corrected.


10. Geometry

Display the 100 IRE Circle Hatch pattern and observe the circles for roundness. If they are not round, the display has a geometry problem. If the display has an anamorphic mode, use a Widescreen Enhanced Circle Hatch pattern to test anamorphic geometry.


11. Scan Velocity Modulation Presence

You can check this two ways. One is to observe the Needle Pulses pattern and see if the white and black portions of the vertical lines are roughly equal in width. If SVM is present, the black portion will be considerably wider. Another test is to view the 16 Rectangles pattern and see if the corners of the white rectangles just touch each other. If the corners are horizontally displaced (white rectangles horizontally shortened) then SVM is present. You'll probably want SVM disabled for best image fidelity.


12. Color Separator Check

Using a COMPOSITE input, view the Zone Plate pattern. Look for cross color (rainbow) artifacts. The fewer the better the display's color separator. Only the most advanced comb filters are able to avoid cross color in the slanted line sections of the pattern. Switching to the Moving Zone Plate also allows you to look for motion artifacts and how well the separator handles motion.


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:25 PM
Post (#16 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



Blue Defocus

Defocusing the blue gun increases the efficiency of that gun and allows greater blue output. This is not an issue with direct view sets, but projection CRT displays often require some degree of blue gun defocus to achieve a grayscale which tracks correctly from black to white.

If you plot the light output of each gun vs input signal level, you'll find that the red and green guns have a nice exponential curve. The blue gun, due to its phosphor reaching maximal output much sooner, has a more sigmoidal shaped curve. This doesn't match the other two guns and if nothing was done, the color of gray would vary with the signal level. Novices often forget this and have a terrible time getting the color of gray correct for the entire 7.5 to 100 IRE signal range. They end up with yellowish highlights if the middle IRE is set to D65. Either that or they get the 100 IRE highlights to D65, but end up with the middle of the scale being too blue. That's because the blue gun quits putting out more light to keep up with the red and green guns.

How do you get around this? One way is to drop total light output by about 30 to 40% to make the blue gun curve more exponential in shape. That occurs if the top end is no longer above the phosphor's linear output limit. That's also usually too dim to keep anyone happy. Pioneer adds a gamma correction circuit to the blue gun circuit which hypes up the top of the blue output curve. This helps, but still means the blue phosphor gets hammered a lot harder than the other two guns. The gentler solution used in many CRT projectors is defocusing of the blue electron beam. This spreads the beam energy over a larger area of blue phosphor and means that any area of phosphor is excited to a lower extent, hopefully staying within linear limits. This works because the spot is scanned across the surface and at any time only a phoshor area is excited. There is reserve emission capability which is brought into use by defocusing the blue gun. This allows one to bring down the beam current but still get good blue light output. The more linear operation of the blue phosphor then allows better grayscale tracking.

Does this make the image blurry? Yes, but to a much smaller extent than one might imagine. Our eyes are conveniently much less atuned to seeing blurring of blue than red or green. Blurring of the blue gun causes minimal perceived decreased in image sharpness from normal viewing distance. You notice it from up close, but the amount needed to get that extra 20 or 25% more blue light output shouldn't be noticed from across the room. If it is, then the beam defocus may be too much.

Precisely defocusing the blue gun to hit the best balance between sharpness and grayscale tracking is a skill even some ISF calibrators have not mastered. Regular service technicians who don't even attempt to attain the flat grayscale HT viewers demand simply are not in need of any further finesse than the vague instruction listed above. Doing it better requires use of a colorimeter and that is hardly something you'll find your service tech carrying about.

So, do you complain about the blue gun defocus? Probably not. That is something you'll need a GOOD ISF calibrator to set correctly. A service tech may well decide to accede to your wishes and simply maximally focus the blue gun without realizing that grayscale tracking will mostly likely go out the window.

You can do a rough check of your grayscale tracking by viewing the AVIA Black Bars + Log Steps pattern. Do this with the color saturation turned all the way down. If grayscale tracking is good, then all the rectangles will have the same color of gray and vary only in brightness.


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:26 PM
Post (#17 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



Some tips on setting beam and optical focus very well.
As mentioned above, display a dots pattern. Set dynamic astigmation controls to center position. This allows you to set the magnets to do most of the work and just fine tune later with the dynamic controls. Binoculars are very handy but must be of large light gathering power.
You'll need to significantly increase contrast setting to above normal to see things clearly during astigmation adjustment. Also work with one gun at a time. Cut-off the other two. Once astigmation is complete, don't forget to return contrast to normal before final beam focusing.
Turn e-beam focus to max one direction then the other. In one direction (underfocused)the dots will become large circular or oval blobs with uniform brightness within the blob. In the overfocused direction, the spots will have a central luminous portion and a dimmer halo. These both need to be examined when doing astigmation.
The 2 pole magnets are the rearmost of the shaper magnets. These are used to center the electron beam in the middle of lens. Place the system into overfocus to show the central luminous point and surround halo. The 2 pole magnets deflect the luminous point. Twist the knob to alter the amount of deflection. Rotate the knob about the tube neck to alter the angle of deflection.
If you use a center cross pattern instead of a dot pattern when centering the e-beam it's easier to see when the beam is centered in the lens. A center cross shows up as fat lines with a central luminous line. Just adjust the 2 pole magnets to make the luminous portion of the vertical and horizontal lines centered within the fat linear halos. It's easier to see than staring at a point with a halo.
The 4 pole magnets are frontmost of shapers. It is used to alter beam shape to a circular shape. One sets the tube to underfocus and displays a dot patter. This turns the dots into a larger, uniform blob whose shape is adjusted using the 4 pole magnets. Twisting the knob adjusts degree of oblongation. Rotating the knob about the CRT neck alters the axis of the effect. Make things as dead round as possible for the center dot.
Turn focus up and down and make sure things track correctly. You may need to make some minor tweaks to the magnets to get things just right.
Focus the dots as well as possible. Then go on to the next gun. When all three guns are completed, you now have "static" astigmation optimized and this means the dynamic astigmation system is under less stress. Perform dynamic astigmation for each zone as well as beam focus.
Now that beam focus is optimized, it's time to recheck optical focus for center, corner, and Schempflug (horizontal and vertical flapping)
You can check lens flapping by using the center lens focus (rear wingnut). Display an inverse + pattern and adjust center focus to make the center of the top screen edge perfectly focused. Note the lens control position. Next adjust the lens to make the center of the bottom screen edge perfectly focused. Is it the same position. If yes, vertical lens flapping is dead on. If the two positions are not identical, you need to readjust vertical Scheimpflug (flapping). Similarly compare the perfect focus positions for left vs right edge. Finish by adjusting center focus to perfection and touching up the corners using corner focus (front wingnut).
Now both e-beam and optical focus are perfected and you can proceed with convergence and redoing grayscale.


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:27 PM
Post (#18 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



Yes the forums are a wonderful resource. The immediate feedback, ability to reach many people at once, addition of information by others, and correction of errors & omissions are why I prefer to answer questions in the open forums rather than private mail or messages. Writing a single person helps just that one individual. If others can read the same information, then there is the hope that the same work can be useful to more people. Often, someone else will have an even better solution than mine. On the other hand, the forums often leave more difficult to write and salient postings virtually unnoticed.

BTW: Does the Fidelio brand velvet happen to have red threads securing its edges? If so, then that is the one I'm using from Joann Fabrics.

Now for something to chew upon.....

LCD projection differs from CRT in obvious ways. The fixed pixel spacing begs for precise match from the video scaler. Otherwise, the built-in scaler is brought into play. That's probably the first thing I would verify. Then it becomes a game of how can the gamut and transfer function of the LCD be made to best match CRT.

We already know that the darkest black from a LCD simply isn't as dark as a CRT. Consequently, dark scenes normally appear somewhat washed out or foggy. A technician may be able to adjust the internal polarizer plate to maximize blackness, but this isn't something for the consumer to attempt. The gamma curve may be alterable on some models. Even so, there remains the vexing problem of hiding those not quite black blacks. You can't fix it by lowering "brightness" control. The dark gray simply doesn't get any darker and you just end up clipping shadow details into black. On the top end of the scale, dialing up contrast too high clips near whites. AVIA's moving black bars and moving white bars let you easily detect when either control is moved beyond the points at which shadow or highlight details get clipped. That is an important feature I added to assist with digitally processed video signals.

I'll now suggest something for your consideration. This is a means of making the LCD dark grays appear black and increase apparent gamma. Note, I'm saying APPARENT gamma, not contrast ratio. We can't eliminate the physical presence of bypass light in darkened portions of the LCD. If LCD projection had greater lumens output, then a gray screen would help by dragging the dark grays down towards the visual threshold at which our vision accepts as black. Again, contrast ratio isn't helped. A neutral filter or f-stopping down the projection lens is less helpful because it lacks the other contrast assisting effect of gray screens - reduction of backscattered light. Unfortunately, LCD's are not all bright enough to take this tact so some other means is needed. Why not alter the gamma response of the viewer?

The tact one normally takes with front projection is to surround the image with a dark or black wall. I'm not talking about the thin frame around the screen fabric. This leaves a bright image surrounded by a dim environment - a dark surround condition. When we view images with a dark surround, the perceived gamma is lower. If we view the exact same image with a bright surround, the apparent gamma is higher. A classic demonstrate of this can be done with two pieces of poster board, one black and one white, and two identical copies of a photo. Cut the boards as mattes to surround each photo. Place the two matted photos side by side and see which one appears to have greater contrast. Despite the photos actually being identical, the one with a bright surround appears to have greater contrast. I think you see where I'm headed. Add the right amount of light in the right places and create a bright surround condition without washing out screen.

I'm suggesting a little experimentation with lighting the wall to each side or above/below your screen with neutral gray light. The intensity of the light will need to be adjusted to hit the best balance between increasing apparent gamma through bright surround effect and not washing out the screen. The bright surround will also tend to constrict your pupil and push black darker. I once this to good effect with my old Sharp LCD projector. I don't think this tact has been adequately exploited in the HT world. We keep going all black as if we were still using CRT projection. The situation is different for LCD and other not-quite black technology. Why use a CRT projection tactic, which worsens the situation?

An initial experiment could be done with a lamp and dimmer control. Arrange the lamp so it lights just your wall and not the screen. Slowly increase light level and see how it affects apparent contrast. Another, easy way to try this in a darkened room is to use a laptop computer screen. Open a big window and place the laptop screen so it is just below the projection screen. Note how the projection screen contrast changes as you fold the laptop screen down and eliminate its effect on your vision.

Where else but this forum would one write such heresy?


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:29 PM
Post (#19 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



Another Grayscale Answer for a User

Let?s first review how the NTSC picture is best conceptualized so the calibration process makes sense.


Think of the image as a black and white picture upon which is overlaid colorization information. This is represented on the signal lines as Y (luminance) for the black and white image (actually composed of various intensities of gray. The colorization information is carried either o the C line (in s-video) or the Pr & Pb lines for component video. It is also possible to combine everything together as compositve video. It really doesn?t matter which type of connection is being used as the end result of calibration is that you take a known signal and adjust the display to give the expected result. If you use several different types of connections, it is normal for the display to behave slightly differently for each type of connection and you should do the calibrations for each type of connection separately. This assumes that the display maintains separate memories for each input.

Now on to the ?white balance? portion of the equation. You recall from above that the base of the image is a black and white picture. The color of white and gray is built in the display from red, green, and blue light. The ideal NTSC display produces a specific color of white know as D65. This approximates the color of mixed direct and indirect outdoor lighting at mid-day. Unfortunately, that doesn?t help you a great deal. Indeed it is nearly impossible to describe the correct color of white without instrumentation. The purpose of the white balance controls is to adjust the color of white and grays so they are all that specific D65 color. You may have heard of using a ?daylight? fluorescent bulb as a reference light against which to match your color of white, but I?ve found that D type bulbs vary tremendously in color. The best is to use a real optical comparator, but you can get fairly good results using a Lumichrome 1xx 6500 kelvin fluourescent bulb as a reference for the proper color of white. That bulb is sllightly too blue but is as close as you will find short of a commercial optical comparator. For most accurate setting of the white balance, one needs to use a spectrophotometer ($10K +!). The idea is to display a grayscale image and adjust white balance to get the right color of white. This is a tricky process and best left to a professional calibrator.

Notice that nowhere in the above paragraph do I mention any of the color controls. That is because the color controls (hue & saturation) are laid ON TOP of the gray picture. Any deviations in the gray naturally affect the final color, but the way the color bars are built automatically compensates for any grayscale errors. This is because the colors are built in the patterns out of the same intensity red, green, and blue light which is used to build gray. Since you do the AVIA calibration by making the intensity of blue equal that in the gray, and also equal in the other two mixed color bars, the process automatically compensates for any excess or minus blue error in the grayscale. In short, it doesn?t matter if the grayscale is off, the setting of saturation and hue is done the same way.

Now I understand your picture still looks off. The first thing to check is that indeed the shades of gray on the display are actually all the same color. Display a crossed gray ramp from AVIA and you should see that all the intensities of gray are the same color of white (preferably all D65). If not, then the grayscale (color balance) must be calibrated against a physical white reference or using a colorimeter. A professional is highly recommended for this as virtually all novices will render things much worse the first dozen or so times they attempt this.

Once grayscale is correctly done, the rest of the picture should fall into place using the usual AVIA calibration procedures for the basic controls. (BTW, in case you are using the Sony Special Edition of AVIA, be sure to use both thickness of blue slides as the blue filter. The retail version of AVIA uses a different filter which is not a slide and can be used as a single layer. On occasion, we have a Sony edition owner use only a single blue slide and that definitely gives poor results.)


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:30 PM
Post (#20 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



Reply re Scaler Question


If I understand the question correctly, the problem is that the AVIA resolution patterns don't seem to look different despite changing the HTPC scaling resolution up and down. This seems puzzling, but is a natural consequence of your equipment's resolution. Despite AVIA's having resolution patterns to the limit of DVD resolution 720 x 480 (540 TVL) and higher than in VE, it is still probably well below what your HTPC is capable of resolving.

The vertical wedges in AVIA resolution patterns are used to estimate horizontal resolution. People are taught to do so by looking for the point at which the vertical lines blur together. This is a rather crude way of considering things, but in a regular TV, the high frequency loss is often severe enough that complete line loss does occur somewhere in the wedge. The high performance displays and systems of home theater enthusiasts, typically have less high frequency loss. HTPC's are particularly high performance in this respect. The old, easy look for lines disappearing just doesn't work. More sophistication is needed to observe changes.

The actual thing one is looking for is how fine details (high frequency) portions of the signal are attenuated in amplitude. High frequency roll-off reduces the amplitude as frequency increases and we observe a dimming of the lines as they come together. One should pay attention to how the lines dim not just if they disappear. On a HTPC, even this tactic isn't complete effective. A good HTPC/display combination has a higher potential resolution than the 720 x 480 limit to DVD video. It may be downright tough to get a HPTC to make the lines disappear or dim. That's why you aren't seeing obvious changes in the patterns. The changes induced by altering scaling are subtler.

Image sharpness and resolution are interrelated but not identical. We often forget that and simply assume that increasing resolution necessarily increase sharpness. Resolution is closed related to how fine details can be and still be represented. Sharpness has more to do with the contrast change rate at edge transitions. It's possible to have high resolution but low sharpness. As you look at changes in scaling rates, keep this in mind. You are usually scaling 720 horizontal pixels to a higher number of pixels. This will tend to preserve all the original resolution but the sharpness of edge transitions may suffer. That is why some higher end scaling solutions apply an edge sharpening function after the scaling to recover some of the image sharpness. As you change scaling, pay attention to how sharp the edges of the black lines remain.

Scaling introduces some unavoidable spatial artifacts. Pixels in the original image are not exactly represented equally in the scaled image. In the frequency domain, this appears as peaks and valleys in the frequency response. I find it is easier to think of the scaling artifact in the spatial domain. Let's assume you scale a 9 pixel wide image up to 10 pixels. As you assign values to the new 10 pixels, you use information contained in the original 9 pixels. You may even interpolate information from more than one original pixel to figure out what each new pixel should contain. Not matter what you do; there are 10 spatial positions to represent 9 evenly spaced positions in the original image. Somewhere in the scaled image, there will be a problem since you can't place anything between the new 10 pixels. If the final number of pixels is much larger than or a multiple of the original pixels, then the spatial error becomes very small or nil. In HTPC's we are scaling a 720 wide pixmap to resolutions nearly double the original. That's still close enough to create noticeable spatial defects. You can detect these in the 540 TVL (every other pixels) circular patch of AVIA resolution patterns. Without scaling artifacts, the vertical lines in the patch are simply uniform vertical lines. I invite you to see how the scaling artifacts decrease in number as you adjust horizontal scaling resolution closer to 1440.

Does this mean you should blindly adjust scaling to minimize that particular artifact? No. You have other things to consider. The actual spot size of the projector is another limit to resolution. If scan lines overlap, you will reduce image sharpness. Again, apparent resolution of the 720 x 480 pixels may not change. Even with 50 percent overlap of scan lines, you'll probably still see some evidence of the smallest details in the resolution pattern. Therein lies a trap people fall into. They pay attention only to the presence of small details (resolution) when checking if a particularly high scaling factor is appropriate for their display. Even worse, they look for the presence of bright small details. The image is being drawn with a bright spot so even small bright details will appear. The thing to look for is small dark details and how dark they appear. If spot size is too large for the resolution then light from the adjacent areas blurring into the dark detail will make the dark detail lighter.

Remember what I said before about the difference between image sharpness and resolution. If you plot the brightness across an edge transition, you find it lose steepness when the requested resolution exceeds the point at which the spot size overlaps the previous sample. This is a drop in sharpness. Setting scaling too high doesn't make the details completely disappear until you go to very very high excess. However, you do begin to blur edge transitions and small dark details reduce in contrast modulation. Those are the things you should look for as you increase HTPC scaling resolution. Examine edge transition sharpness and how dark small dark details remain. If the edges drop in sharpness or small dark details begin to lighten, then you've gone too far.

The above assumes a CRT display. On a digital display, a specific final resolution is needed to match the native resolution. Matching that resolution is needed to allow the HTPC's superior scaling to do the work instead of the usually less capable scaling available in the projector. I have also ignored the temporal advantages of various refresh frequencies on HTPC.


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:32 PM
Post (#21 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



Grayscale Rescue for an Overly Enthusiastic Tweaker

As you've discovered, adjusting gray scale is not an easy matter. That is why most people leave it to professional ISF calibrators who can tune it in very tightly. At the very least you need an optical comparator or D65 light source against which to compare the color of gray while making such adjustments.

This is only intended to get you closer to normal. Doing this sort of stuff is best left to a technician. You can get yourself into a ton of trouble doing this. Since you've already gotten yourself there, I'll try to help. I'm not going to be specific to model so the actual name of the service menu items may not match your set.

I hope you haven't adjusted any of the variable resistor type "screen" controls. These are set at the factory to yield precise grid voltages for the CRT's. They should be left alone and adjusted only with instrumentation and the service manual. If you have messed these up, then at the very least, display a black field pattern and look directly into the CRT's to make sure the raster (area scanned by the electron beam) is just extinguished and no diagonal retrace lines are visible. Unfortunately, if you have already significantly altered the bias (cutoff) settings for the tubes, the proper setting for screen may not look right. You might have lightened the blacks of a tube using bias causing you to incorrectly adjust screen too dark. Setting screen controls incorrectly can accelerate CRT wear or burn out so I REALLY hope you have not touched those controls.

The controls for setting gray scale are the gain (drive) and bias (cutoff) controls for each gun. The gain and bias controls interact with each other, much as the black level (brightness) and white level (picture) controls. The bias (cutoff) control sets the appearance of black for a tube. The gain control sets the intensity of light output for a tube. One sets gray scale by displaying a 90 or 100 IRE window and adjusting gain. Then display a 20 IRE window and adjust bias. The process is repeated back and forth between the high and low IRE windows until both 20 and 90 IRE windows have the same, correct color of gray. Professional calibrators use a colorimeter or a optical comparator to tell when the color of gray is correct. It is near impossible to eyeball the color of gray without a reference white to compare against because human eyes continuously white balance and change our perception of gray.

Barely acceptable, readily available references for the color of white are a computer monitor set to 6500K (not very accurate as these tend to be all over the place in actual color), a high CRI 6500K fluorescent tube such as a Lumichrome 6500K (good accuracy), or a daylight fluorescent bulb (tend to be green tinted). If you've really messed things up, then I guess any of these would be better than nothing for a reference. Shine the bulb on a neutral gray or white card like the Kodak gray card to vary the brightness of the light. Compare the color carefully against what the display is producing.

With a colorimeter, accuracy is far better and the ISF calibrator can measure the color for points along the entire gray scale from dark to light. Most displays don't track the entire scale exactly right and an experience calibrator knows which compromises to take to make overall grayscale acceptable.

One piece of advice I would give is to always leave one of the guns (green) alone while making adjustments. If you alter all three guns, you'll soon be in a deeper quagmire. You only need to vary two of the guns since it is the RATIO of light between the guns that yields the color. At the very least, don't go raising the gains much above their factory levels.

One could go commando, take off the screen and measure the light output from each projection tube using the solar cell technique to get better precision than by eye, but that is probably more surgery than most users can safely endeavor. I'd just eyeball the color of gray of the 20 and 90 IRE windows and adjust bias and gain until both match the color of the reference light. Varying the distance between the lamp and card lets you change the brightness of your reference.

I hope this helps. The big lesson here is that one must not get carried away making adjustments in the service menu unless one fully understands their implications. Some people have even misunderstood how things work so far that they try to adjust red tube gain and bias (gray scale controls) in an attempt to fix a totally unrelated red push problem in their color decoder! The best policy is to leave service menu items to the pros or wait until one learns considerably more about how a display functions.


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:33 PM
Post (#22 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



Another Grayscale Rescue

(Posted as an abject reminder to stay within your limitations)

The service level controls inside a RPTV can create some real problems. I'll try to help, but frankly this is very difficult to do without a professionally trained eye actually seeing the set. It appears you have adjusting the potentiometers on the set's convergence board. Setting convergence on a RPTV is difficult without prior experience. You often end up with a worse picture the first few times you try. Doing such adjustments without a service manual and experienced guidance is fraught with difficulty and some danger. It sounds like you've also accidentally altered the focus controls and perhaps touched the tube drive controls.

Here's what I would do in an attempt to correct your adjustments.

1. Temporarily adjust contrast and brightness while viewing the Needle Pulses + Steps pattern. Set brightness such that the black background just barely extinguishes and becomes black. Set contrast to make the white portions of the pattern intentionally darker than white. This temporarily brings the image into linear range so you can go on to the next step.

2. Open the display and access the convergence board. The next step is refocusing the CRT electron beams. On the vintage Mitsubishi display that you are using, I believe the focus pots are actually on the convergence board rather than a separate high voltage assembly. This makes them easy to accidentally tweak while adjusting geometry and convergence.

Display the Gray Field Plus pattern and look directly into each CRT (through the lens) and adjust beam focus for maximal sharpness of the pattern. Try to make the raster scan lines and vertical elements of the pattern as distinct as possible. You'll probably find that the blue gun does not focus as tightly as the others.

3. I hope you have not touched optical focus on the display. If you have, you'll need to place a piece of paper at the focal plane of the normal screen and refocus each lens. I use two piece of plastic (a pair of floppy discs) to act as offset mounts for a long piece of packaging tape stretched across the front of the opened projector. The floppies are taped to each side of the cabinet to protrude exactly the same distance forward as the normal screen sits. A piece of paper on the stretched piece of tape then serves as the focus plane. This allows access to the lenses for adjustment. Each optical focus is then adjusted to sharpen the image. It's best to cover the other two lenses while making adjustments. I sincerely hope you have NOT touched the factory lens settings as this WILL require full geometry and convergence adjustment if you alter optical focus.

4. Next set CRT gun centering. This is done by displaying a center cross pattern, setting the user centering controls to middle position and adjusting the centering magnet rings on the CRT necks. These are usually factory sealed so I doubt you've messed them up. I'd leave them alone if you haven't broken the seals. Excessive force can snap a CRT tube neck so this is not something to mess with unless one has to do so. Each gun needs to be adjusted to place the center of the pattern exactly in the center of the screen. Personally, I'd just leave the magnets alone in your set and use the user centering controls to align the three guns on each other.

5. Cover the red and blue guns. Set green gun size, linearity, pincushion, and keystone to make the circle hatch pattern fit on the screen with about 3 to 5 percent overscan on each edge. Get the green gun as well aligned as possible for it will act as the template for the other guns.

6. Uncover the red gun and adjust geometry and convergence controls to best align atop the green gun image. The crosshatch pattern is best for this. First concentrate on the center vertical and horizontal lines. Get skew and bow correct for the center lines. Then work on pincusion, pin balance, keystone, and keystone balance. You set may not have all the controls or may have additonal ones. The caveat here is not to accidentally change controls for other guns while adjusting red.

7. Cover the green gun. Uncover the blue gun. Adjust blue gun convergence and geometry to align on the red image.

8. Uncover all three guns and fine adjust red and blue geometry and convergence as needed.

9. Next, set sub brightness and sub contrast to make the default or center positions of the remote control bring up the correct black and white levels. Set the user bright and contrast controls to midposition. (If the remote has a reset button set to the "reset" position)

Find the sub-brightness and sub-contrast controls inside the display.

Use the Black Bars + Half Gray pattern while using the sub-brightness controls to make the black background black, and the two dark gray moving bars just barely visible. This should be done in subdued room lighting!

Use the Needle Pulses + Steps pattern while adjusting the sub-contrast control to correct levels. Keep things below the point of blooming (blurring of the electron beam). Correct level is the lowest setting that makes the white portion of the pattern appear white (rather than a shade of gray) and below the point of blooming.

You may need to adjust subbright and subcontrast alternately to achieve correct levels.

10. At this point, I would adjust the display's gray scale tracking using the bias and drive controls, but I hope you have NOT tweaked those. Setting gray scale requires a reference white for observations and a LOT of practice to do correctly. This step would take a very long discussion.

After gray scale is set, the sub-bright and sub-contrast controls will probably need retweaking.

11. Finally, close things up and proceed with the normal user level adjustments that ARE described on the AVIA disc to achieve the final image.

If the above seems too difficult, get a professional as you are probably in over your depth. I hope this helps, but you've far exceeded the adjustments described on the AVIA disc and pretty badly distorted the image of your RPTV. This is clearly not a problem with AVIA but overenthusiasm.


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:35 PM
Post (#23 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



DIY 6.1 using AVIA

(Thankfully Home Theater Tune-up has the 6.1 tests as discrete signals)

Here's how you can set the rear left, center, and right surround channels using AVIA. I'm assuming you are using a 5.1 processor and feeding the left surround and right surround line level outputs to a separate ProLogic receiver which actually drives the three rear speakers. If you have an official 6.1 processor, go ahead and use its internal channel setup tones. It?s much simpler than setting up a homebrew, two receiver setup as I describe below.

1. Set the right and left surround levels on your 5.1 receiver to equal levels as indicated electrically, not by equalizing the SPL outputs. The signal intensities feeding the Prologic decoder of the second receiver must be equal for the steering logic to work best.

2. Set the Prologic receiver to no surround, only 3 front channels. The three front channels of the Prologic unit then are used to feed the three surround speakers. The rear speaker outputs of the Prologic unit are not used.

3. Calibrate the main 5 channels as usual with AVIA's test tones but with the following nuances.

a. The front left, center, & front right speaker SPL's are set using the main volume and channel controls of the 5.1 receiver.

b. The rear surround and rear left speaker SPL's are set using the main volume and channel controls of the Prologic unit, NOT the 5.1's channel levels.

That takes care of the main five channels. The next step is to calibrate the center surround channel relative to the left surround and right surround channels. Do this with the built-in test tone of the Prologic receiver. Adjust ONLY the "center" channel level to make the SPL of the center surround speaker match the left and right surround speakers. Do not alter the master volume or left/right channels levels of the Prologic unit. Use ONLY the center channel level control.

Now all 6 channels are dialed in and ready for testing. Play the Left Surround/Right Surround Phase test on AVIA (not available on VE. The in phase portion of the test should come out of the center surround speaker.

Finally, go on to set your subwoofer phasing, crossover point, and output level using the subwoofer setup section of AVIA.


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:38 PM
Post (#24 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



For those who are making Pronto macros for using AVIA, here are the Title and Chapter
numbers for both the audio signals and video test patterns. GK


Title Chapter Audio Test Signal

6 2 Channel Identification (5.1)
6 56 5 Channel Speaker Balance
6 3 Left-Front Level
6 4 Center Level
6 5 Right-Front Level
6 6 Right-Surround Level
6 7 Left-Surround Level
6 43 Phase Left-Front/Right-Front
6 44 Phase Left-Front/Center
6 45 Phase L-Surround/R-Surround
6 46 Phase L-Front/L-Surround
6 9 Subwoofer Level, Left-Front
6 10 Subwoofer Level, Center
6 11 Subwoofer Level, Right-Front
6 12 Subwoofer Level, Right-Surround
6 13 Subwoofer Level, Left-Surround
6 14 Subwoofer Phase Filtered Pink Noise, Left-Front
6 15 Subwoofer Phase Filtered Pink Noise, Center
6 16 Subwoofer Phase Filtered Pink Noise, Right-Front
6 17 Subwoofer Phase Filtered Pink Noise, Right-Surround
6 18 Subwoofer Phase Filtered Pink Noise, Left-Surround
6 19 Subwoofer Phase Warble Sweep, Left-Front
6 20 Subwoofer Phase Warble Sweep, Center
6 21 Subwoofer Phase Warble Sweep, Right-Front
6 22 Subwoofer Phase Warble Sweep, Right-Surround
6 23 Subwoofer Phase Warble Sweep, Left-Surround
6 25 Wideband Pink Noise 5 Channel Pan
6 26 150 Highpass Pink 5 Channel Pan
6 27 Circulating Ambience Generator Clicks
6 55 Pink Noise Match of Center Speaker
6 47 Low Frequency (200 to 20 Hz) Sweep, Left-Front
6 48 Low Frequency (200 to 20 Hz) Sweep, Center
6 49 Low Frequency (200 to 20 Hz) Sweep, Right-Front
6 50 Low Frequency (200 to 20 Hz) Sweep, Right-Surround
6 51 Low Frequency (200 to 20 Hz) Sweep, Left-Surround
6 52 Low Frequency (200 to 20 Hz) Sweep, LFE
6 28 Low Frequency Pink Noise, 5 Channel Pan
6 29 Low Frequency Pink Noise, 6 Channel Pan
6 31 Wideband Pink Noise, Left-Front
6 32 Wideband Pink Noise, Subwoofer Level, Center
6 33 Wideband Pink Noise, Subwoofer Level, Right-Front
6 34 Wideband Pink Noise, Subwoofer Level, Right-Surround
6 35 Wideband Pink Noise, Subwoofer Level, Left-Surround
6 36 Wideband Asynchronous Pink Noise, 5 Channels


Title Chapter Video Test Pattern

1 1 Needle Pulses
1 2 Needle Pulses + Steps
1 3 Black Bars + Log Steps
1 4 Black Bars
1 5 Black Bars + Half Gray
1 6 Black Bars + Half White
1 7 Vertical 10 IRE Steps
1 8 Horizontal 10 IRE Steps
1 9 Crossed Step Scale
1 10 Vertical Brightness Steps
1 11 Horizontal Brightness Steps
1 12 Black
1 13 10 IRE Window
1 14 20 IRE Window
1 15 30 IRE Window
1 16 40 IRE Window
1 17 50 IRE Window
1 18 60 IRE Window
1 19 70 IRE Window
1 20 80 IRE Window
1 21 90 IRE Window
1 22 100 IRE Window
1 23 20 IRE Window
1 24 10 IRE Field
1 25 20 IRE Field
1 26 30 IRE Field
1 27 40 IRE Field
1 28 50 IRE Field
1 29 60 IRE Field
1 30 70 IRE Field
1 31 80 IRE Field
1 32 90 IRE Field
1 33 100 IRE Field
1 99 Black Bars
1 102 Vertical Gray Ramp
1 103 Horizontal Gray Ramp
1 104 Crossed Horizontal Gray Ramp
1 105 Crossed Vertical Gray Ramp
2 1 Center Cross 30 IRE
2 2 Center Cross 50 IRE
2 3 Center Cross 100 IRE
2 4 Crosshatch 30 IRE
2 5 Crosshatch 50 IRE
2 6 Crosshatch 100 IRE
2 7 Crosshatch Inverse
2 8 Dot Hatch 30 IRE
2 9 Dot Hatch 50 IRE
2 10 Dot Hatch 100 IRE
2 11 Dot Hatch Inverse
2 12 Circle Hatch 30 IRE
2 13 Circle Hatch 50 IRE
2 14 Circle Hatch 100 IRE
2 15 Dots 30 IRE
2 16 Dots 50 IRE
2 17 Dots 100 IRE
2 18 Gray Field Dots
2 19 White Field Dots
2 20 Black Field Plus
2 21 Gray Field Plus
2 22 White Field Plus
2 23 Checkerboard 30 IRE
2 24 Checkerboard 50 IRE
2 25 Checkerboard 100 IRE
2 27 Crosshatch 1.66 30 IRE
2 28 Crosshatch 1.66 100 IRE
2 28 Crosshatch 1.66 50 IRE
2 29 Crosshatch 1.85 30 IRE
2 30 Crosshatch 1.85 50 IRE
2 31 Crosshatch 1.85 100 IRE
2 32 Crosshatch 2.0 30 IRE
2 33 Crosshatch 2.0 50 IRE
2 34 Crosshatch 2.0 100 IRE
2 35 Crosshatch 2.35 30 IRE
2 36 Crosshatch 2.35 50 IRE
2 37 Crosshatch 2.35 100 IRE
2 38 WSE Crosshatch 30 IRE
2 39 WSE Crosshatch 50 IRE
2 40 WSE Crosshatch 100 IRE
2 41 WSE Crosshatch Inverse
2 42 WSE Circle Hatch 30 IRE
2 43 WSE Circle Hatch 50 IRE
2 44 WSE Circle Hatch 100 IR
2 45 WSE Dot Hatch Inverse
2 46 WSE Dot Hatch 30 IRE
2 47 WSE Dot Hatch 50 IRE
2 48 WSE Dot Hatch 100 IRE
2 49 WSE Dots 30 IRE
2 50 WSE Dots 50 IRE
2 51 WSE Dots 100 IRE
2 52 WSE Resolution
2 154 Crosshatch 1.78 30 IRE
2 155 Crosshatch 1.78 50 IRE
2 156 Crosshatch 1.78 100 IRE
3 1 Resolution 100 TVL
3 21 Resolution 200 TVL
3 22 Sweep
3 23 Sweep 50%
3 24 Multiburst
3 25 Multiburst (Labeled)
3 26 Multiburst 50%
3 27 Multiburst 50% (Labeled)
3 28 Sharpness
4 1 Blue Bars
4 2 Red Bars
4 3 Green Bars
4 4 Split Color Bars
4 5 Split Bars with Gray
4 6 Crossed Bars
4 7 Minimum Phase Bars
4 8 Maximum Phase Bars
4 9 Full Bars
4 10 Split 100/75 Bars
4 11 Full 100 Bars
4 12 Split 100 Bars
4 13 Full 50 Bars
4 14 Split 50 Bars
4 15 Yellow Field
4 16 Cyan Field
4 17 Green Field
4 18 Red Field
4 19 Magenta Field
4 20 Blue Field
5 1 Color Decoder Check
5 2 Y/C Delay
5 3 Zone Plate
5 4 Zone Plate (30 frames/sec)
5 5 Gamma Chart
5 6 16 Rectangle
5 7 Overscan
5 8 Backlight Levels


--------------------------------------------------------------------------------

Guy Kuo
Director - Imaging Science Foundation Research Lab
Video Test Design - Ovation Multimedia / Home of OpticONE Colorimeter, AVIA and Avia PRO




Posted: November 25th, 2001
01:41 PM
Post (#25 of 28 posts)


Guy Kuo

Seattle, WA USA
Member

--------------------------------------------------------------------------------
Local Date and Time:
October 16th, 2005
11:07 PM
--------------------------------------------------------------------------------

Member Since:
March 6th, 1999

--------------------------------------------------------------------------------



That should get you an idea of some things that can be done with AVIA.
ChrisWiggles is offline  
post #4 of 57 Old 10-19-2005, 06:28 PM
 
ChrisWiggles's Avatar
 
Join Date: Nov 2002
Location: Seattle
Posts: 20,730
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 15
quotes are my own:

Quote:
On a digital display you want to maximize your contrast range as much as possible. On an DLP like the x1, you can use mirror dithering to set black level with extreme precision. Throw up any of the black patterns, and observe the black background, which is at video 16 (black). You should set this so that the mirrors are off, or one click above this so that you see them dithering. You should see both black bars on the Avia patterns, as they are encoded above black. There is no below-black elements on Avia.

For white, on a digital display you are looking for clipping or colorshifting. Some displays will hard-clip white, and others will clip at one color before the others which will show as a colorshift. Note how this is very different than calibrating a CRT's white level. Both of the moving bars on Avia are below reference white(235) so they should ALWAYS remain visible. You should raise your white level until the larger white portion (which is brighter than the bars) begins to colorshift, or you begin to see that portion clip down until the bars below white are no longer visible as distinct from white. You should stay a few clicks below the point of clipping or colorshifting. This will maximize your white point.

Now, you may want to return to black, and go back and forth between black and white a couple times as the controls may interact, so it is an iterative process.

***advanced note: It is preferable for some demanding viewers, to align the peak white of a digital display to peak white 254, rather than reference white 235 so as to prevent any peak white highlight detail clipping. This will reduce the contrast range between black and reference white slightly, and your in-scene contrast will be slightly less, but you will maintain data the extends beyond reference white. This is a subjective choice by the user on displays that have limited contrast ranges, unlike the unlimited range provided by CRTs that does not prompt this choice. To do so, you will need to use patterns that extend to peak white, such as those on Avia PRO, or DVE.

Quote:
Basically what he's saying is that you should see BOTH moving vertical bars on BOTH the black and white test patteren, turning the black and white level up or down until one bar is visible and keeping the level at just before the point where the other bar next to it disappears into the background.

That is correct. All the moving bars in Avia are *within* the range from black to white. The black bars in Avia are both *above* black. The white bars are both below white. If the bars are clipped off, then you are losing shadow detail near black, and obviously you'll be losing white detail near white.

The black bars may become obscured using a very high APL pattern (like the half-white pattern in the advanced menu) if your system has very low ANSI contrast. With most digitals, however, this is not really a problem because the black level of the display is elevated enough, and ANSI contrast is high enough that the visibility of the bars doesn't really vary with APL. On CRTs, this is more complex, because if you throw up a full black pattern with the bars and calibrate, then switch to one with half gray or white, ANSI washout will render the bars invisible. On CRTs then it is a little more complicated because the ultimate black level you arrive at will vary depending on the APL of the pattern, and the pattern you choose can depend on your preference for total black-out, or rendering all the shadow detaisl even in bright scenes (high APL).

Quote:
And remember, if you have a DLP, use dithering to find your black point by going up to the screen and observing the dithering. This will provide extreme precision compared to paying attention to the bars. Just observe the black background and find the point where the pixels start/stop dithering, that's your display's black point. At this point, the bars should all be visible, because they are all above black (on Avia; on DVE there are additional bars that are below black).

Quote:
Dither is how a DLP chip creates a grayscale. The DLP mirrors only have an ON and an OFF position, so white is full on, and black is full off. Everything in between is a ratio of how much time the mirror spends flipped on to flipped off. This happens extremely fast, so you don't see the mirrors flipping, however very close to the black point, the mirrors are mostly completely off, but flick on just a few times over a period of time to be slightly above black. If you go right up to the screen, even on paused video, look at dark objects and detail and you can see a sort of "dancing" on the screen as the pixels try to create light output that is very close to black.

If you pay attention closely to this dithering right at the screen, you'll see that as you lower your black level, at some point the detail that you are observing will plunge below the black point on the display, and the pixels will stay completely in the off position and the dithering will dissappear. This is the black point of your DLP display. So, then use any pattern with video black, and use the black portion of this test to align black to this point on the display. I hope that helps and makes sense. You have to go right up to the screen and look closely at the pixels, and you should immediately see what I'm talking about, and then play around with your black level control and you'll see the dithering dissappear and re-appear as you move your black level down and up again. By paying attention to this dithering, you can identify very very precisly the exact black point of your display and align a test pattern to it.

Quote:
hehe, you look at the black *background* for dithering, as the background is black. The point where it starts dithering is the projector's black point, so you are aligning the black background with this point on the projector. At this point, BOTH Avia bars should be visible (except when they may be obscured by washout from bright parts of a pattern if applicable). Remember that both bars are ABOVE black and should remain visible if the projector's black point and the black background of the pattern is aligned correctly.

ChrisWiggles is offline  
post #5 of 57 Old 10-27-2005, 10:52 PM
Super Moderator
 
markrubin's Avatar
 
Join Date: May 2001
Location: Jersey Shore
Posts: 23,099
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 131 Post(s)
Liked: 480
this thread is for FAQ contributions only: will periodically split out any other threads

please post questions in another thread

split out threads were moved Here

Thanks

please take the high road in every post
if you see a problematic post, please do not quote it or respond to it: report it to the mods to handle
markrubin is offline  
post #6 of 57 Old 11-03-2005, 07:35 PM
AVS Special Member
 
Ursa's Avatar
 
Join Date: Jul 2003
Location: Inner Loop, Houston
Posts: 5,001
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Here is my contribution. It covers some of the basics of doing a true grayscale calibration, and includes charts to illustrate what happens when you calibrate a display.

Please feel free to send comments or edits my way.

Later,
Bill

 

Calibration How-To v1.pdf 456.3935546875k . file
Ursa is offline  
post #7 of 57 Old 11-03-2005, 08:17 PM
AVS Addicted Member
 
krasmuzik's Avatar
 
Join Date: Oct 1999
Location: NewPort, VA
Posts: 11,270
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Hey I got a thanks - glad we could bounce formulas off each other!
krasmuzik is offline  
post #8 of 57 Old 11-10-2005, 02:07 PM
AVS Special Member
 
Ursa's Avatar
 
Join Date: Jul 2003
Location: Inner Loop, Houston
Posts: 5,001
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by mdtiberi View Post

Bill:

I have read your Calibration FAQ, thanks it is most helpful. Couple of questions just to make sure the concepts are sinking in. Before calibrating my 34HS420 I plan on taking a lot measurements with the Spyder2 Pro. Were the charts in your FAQ generated by the Spyder software or did you have to extrapolate.

What I would like to do is replicate your charts to establish a baseline and then apply your correction methods accordingly.

When you say that you adjusted the red contrast would this be referring to the red cut-off voltage?

The screen shots were from a tool I created, and that is mentioned extensively in the SpyderTV review. Before anyone asks: no, I am not giving it out (read the STV review to figure out why). Included in the guide is a translation for the various synonyms used (see page 11). Cut = brightness in my terminology.

Later,
Bill
Ursa is offline  
post #9 of 57 Old 12-07-2005, 11:59 AM
AVS Special Member
 
cyberbri's Avatar
 
Join Date: May 2004
Location: San Jose, CA
Posts: 8,095
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 53
cyberbri is offline  
post #10 of 57 Old 01-14-2006, 09:01 PM
Member
 
kess91's Avatar
 
Join Date: Dec 2005
Posts: 36
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
I got the DVE DVD, but it won't work in my DVD player, and it is a brand new Sony????????
kess91 is offline  
post #11 of 57 Old 01-15-2006, 12:19 AM
Member
 
kess91's Avatar
 
Join Date: Dec 2005
Posts: 36
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
These pixels are driving me crazy. I love the tv, but even when i watch a dvd, i can see the pixels, like little blurry squares. I see it on regular tv also bad. I am using HDMI to HDMI on the dvd, and direct tv receiver. Can anyone tell me if there is a way to get a better picture please?????
kess91 is offline  
post #12 of 57 Old 01-15-2006, 08:30 AM
AVS Special Member
 
Ursa's Avatar
 
Join Date: Jul 2003
Location: Inner Loop, Houston
Posts: 5,001
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
kess91 - It sounds like you may have the "sparklies", which would necessitate a new/better cable. However, this really isn't the thread for those questions.

Later,
Bill
Ursa is offline  
post #13 of 57 Old 01-15-2006, 11:47 AM
Member
 
kess91's Avatar
 
Join Date: Dec 2005
Posts: 36
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
I have the best wires though?????
kess91 is offline  
post #14 of 57 Old 01-15-2006, 11:51 AM
 
ChrisWiggles's Avatar
 
Join Date: Nov 2002
Location: Seattle
Posts: 20,730
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 15
apparently not? Is it a long run, are the connectors fastened securely?
ChrisWiggles is offline  
post #15 of 57 Old 01-15-2006, 11:52 AM
 
ChrisWiggles's Avatar
 
Join Date: Nov 2002
Location: Seattle
Posts: 20,730
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 15
kess: I ran a quick search on your posts, it seems you are mis-describing your problem, are you seeing very small pixel-sized like artifacts, or larger blocks? It seems elsewhere you are describing visible macroblocks, i.e. large patches of squares, that will probably be especially visible during motion and such. This is a source problem, very common on a lot of cable and satellite material, unfortunately due to compression.
ChrisWiggles is offline  
post #16 of 57 Old 01-15-2006, 02:41 PM
Member
 
kess91's Avatar
 
Join Date: Dec 2005
Posts: 36
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Hi,
Yes i beleive you are correct. I do see them during the dvds, but less, and even less on High Def channels. They are smaller squares, but all together they look bigger, and almost blurry. Will the off air antennae help you think?
kess91 is offline  
post #17 of 57 Old 01-15-2006, 11:04 PM
AVS Special Member
 
Ursa's Avatar
 
Join Date: Jul 2003
Location: Inner Loop, Houston
Posts: 5,001
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
For macroblocking, Chris has already indicated the fix: better source material. DirecTV (D*) will get pretty bad macroblocking during rainstorms. OTA will do so as well, but not to the same degree. A better antenna and/or amp will help OTA. A bigger dish helps D*. If you have it on DVDs, then your DVD player is suspect. You really should start a new thread in the appropriate fora for help with the specific problems.

Later,
Bill
Ursa is offline  
post #18 of 57 Old 01-20-2006, 11:11 AM
Member
 
jzdugolf's Avatar
 
Join Date: Mar 2005
Posts: 89
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
if my pj (Sanyo Z4) has in its adjustment menu: R,G,B; R Gain, G Gain, B Gain; R offset, G offset, B offset; and R gamma, G gamma, B gamma; what properties of the primary colors do they control? Thanks a lot for the info. Jim
jzdugolf is offline  
post #19 of 57 Old 01-20-2006, 12:01 PM
 
ChrisWiggles's Avatar
 
Join Date: Nov 2002
Location: Seattle
Posts: 20,730
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 15
They do not sound like they alter the chromaticities, sounds like they alter Gain, offset, and gamma response for each.
ChrisWiggles is offline  
post #20 of 57 Old 01-20-2006, 12:14 PM
AVS Special Member
 
cyberbri's Avatar
 
Join Date: May 2004
Location: San Jose, CA
Posts: 8,095
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 53
The R/G/B adjustments are used to adjust color temperature (greyscale). If the greyscale color temp is off, it affects all the color that is laid over top of it. If you have your projector hooked up with component cables, you can unhook the two that carry color and leave the luminance (B/W) cable - then try adjusting the R/G/B gains/offsets. You'll see how these affect the greyscale. If these are off, the color itself is off.
cyberbri is offline  
post #21 of 57 Old 01-20-2006, 12:47 PM
Member
 
jzdugolf's Avatar
 
Join Date: Mar 2005
Posts: 89
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
I see. ... But I am still wondering about the difference between R/G/B vs R/G/B Gain.
BTW I am using HDMI and I have only AVIA. With all test patterns availlable on the AVIA how far can I go in calibrating the display without any additional tools? It seems the DVD only covers calibration on B/W, sharpness, Color/Tint. After I went thru the color/tint the picture just looked horrible. But color/tint are only two out of some two dozen adjustments you can make on colors. so can you experts tell me if this type of AVIA calibration is simply useless (or making things worse) on these moden digital pjs?
jzdugolf is offline  
post #22 of 57 Old 01-20-2006, 03:42 PM
AVS Special Member
 
cyberbri's Avatar
 
Join Date: May 2004
Location: San Jose, CA
Posts: 8,095
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 53
The R/G/B controls are for color temperature / greyscale (or if you have any color temp presets already, you may not need to do it yourself). The color (saturation) and tint (color decoder, skin tones) are for the actual color.

Unless you use preset color temperature settings, if you want to adjust color temperature, you would display b/w greyscale patterns and use the r/g/b gains and cuts to get a neutral (slightly warm) greyscale, with no odd colorations. To check, like I said with the component cables, you can play anything and unplug the two cables that carry the color information and see how the greyscale looks. I have done this with Finding Nemo to see the gradations in the water (in b/w).

Once you get the greyscale (and gamma - I've never heard of gamma controls by color) down, then you move onto the color/tint. Color is just saturation, so if whatever results you get with Avia you aren't happy with or sure about, you can try adjusting this by eye. Tint is mainly skin tones, so besides the Avia tests, you can test this with people to get the skin color looking right.

Are you on the thread for your particular projector in the projector forum? Usually on those threads, people talk a lot about tweaks, settings, calibrations.
cyberbri is offline  
post #23 of 57 Old 01-20-2006, 07:29 PM
Member
 
jzdugolf's Avatar
 
Join Date: Mar 2005
Posts: 89
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
thanks a lot, cyerbri, for taking time explaining these concepts. now I understand that RGB are purely for grayscale(white's temperature) and have little to so with the actual colors. Then I suppose the offset and gain for each color would be like brightness and contrast levels for each color output.
now it makes a lot more sense. I have been reading the tweak thread on my pj, but I want to understand the basic concepts correctly first rather than trying to guess what other people are doing with the tweak. Thanks again for your help. This is a great resource for newcomers.
jzdugolf is offline  
post #24 of 57 Old 01-20-2006, 09:13 PM
 
ChrisWiggles's Avatar
 
Join Date: Nov 2002
Location: Seattle
Posts: 20,730
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 15
Quote:


Then I suppose the offset and gain for each color would be like brightness and contrast levels for each color output.

Correct.
ChrisWiggles is offline  
post #25 of 57 Old 03-05-2006, 08:02 AM
Senior Member
 
jbaracelona's Avatar
 
Join Date: Apr 2003
Location: river ridge, la.
Posts: 250
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
I have an Oppo and the Bravo DVD Players. I've tried to use them connected to my Yamaha 500LPX Projector, LCD, with a DVI connection. I can complete the the black and white, but nothing else. It won't allow me to calibrate color, saturation or anything else. Is there something I can do to further calibrate?
Thanks,
Joe

Joe
jbaracelona is offline  
post #26 of 57 Old 03-05-2006, 09:51 AM
AVS Special Member
 
Ursa's Avatar
 
Join Date: Jul 2003
Location: Inner Loop, Houston
Posts: 5,001
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Joe - Very few projectors provide color, tint, saturation, etc. controls when fed via a digital connection. To get more out of your Yamaha (aka Epson TW100, IIRC), you will need to see if Epson has some color editing utilities. However, I think that those may have been developed for the Y610/E500. Also, you may want to see about getting into the service menu.

Later,
Bill
Ursa is offline  
post #27 of 57 Old 03-05-2006, 11:27 AM
 
ChrisWiggles's Avatar
 
Join Date: Nov 2002
Location: Seattle
Posts: 20,730
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 15
Quote:


It won't allow me to calibrate color, saturation or anything else. Is there something I can do to further calibrate?

What Bill said, however I want to clarify that it's not that it is a "digital connection" but rather the signal format. DVI sends an RGB signal, it's already been fully decoded from a luma/chroma type signal, so there is no color saturation or balance adjustment to be made unless you take the signal format BACK into that type of signal, which is unecessary. Adjustments, if needed, would have to be made in the source.

HDMI can send component YCbCr, so you are more likely to have access to color adjustments with that signal format, however they still may not always be present.
ChrisWiggles is offline  
post #28 of 57 Old 04-01-2006, 01:37 PM
Advanced Member
 
rrhomes's Avatar
 
Join Date: Sep 2002
Location: Oklahoma City
Posts: 932
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Re: DVE Calibration disk. On Chapter 13:1 Where the levels of contrast start out at 0% black and then circle in 10% increments to 100% white. Then there are two extra squares that have off setting contrast smaller box's in them. Are those inner squares suppose to represent 10% above Pure black and 10% below Pure white, or are you suppose to adjust you brightness and contrast until the inner square are 1 click away from clipping - in essence making them 1% above Pure black 1% below pure white respectively. My image sure does look great when I treat them as the 1% shade but I think that may be wrong.

Does DVE have any 1% below white pattern. I will use the above ditthering method to set black, but I can't find any one percent below pure white unless it the cha[ter 13:1 of DVE but I thnk those are suppose to be 10% as stated above.
rrhomes is offline  
post #29 of 57 Old 07-03-2006, 11:02 PM
Member
 
SFXtreme3's Avatar
 
Join Date: Jul 2006
Posts: 45
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
R BIAS 64
G BIAS 150
B BIAS 104
R/G DRV 77
B/R DRV 74
R BIAS C 64
G BIAS C 142
B BIAS C 83
R/G DRV C 72
B/R DRV C 83
R BIAS W 64
G BIAS W 161
B BIAS W 127
R/G DRV W 87
B/R DRV W 66
R-Y GAIN 8
R-Y PAHSE 3
G-Y GAIN 5
G-Y PAHSE 0



My screen is incredibly green.

I understand that R Bias, G Bias, and B Bias (c for cool, w for warm, and nothing for medium) adjusts color temperature and is used for grayscale calibration.

What I do not understand is DRV, Gain, and Phase.



When my Xbox 360 goes into hibernation (dims everything) its very green. My default settings on green are really high too.



When using the blue bars screen I have tint/hue perfect, red is way off, green is a little off. If green is just a little and not nearly as off as red why is my screen green?

What do you suggest?
SFXtreme3 is offline  
post #30 of 57 Old 07-04-2006, 02:09 AM
AVS Special Member
 
cyberbri's Avatar
 
Join Date: May 2004
Location: San Jose, CA
Posts: 8,095
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 53
So you've calibrated the greyscale to as close to 6500 as you can get it in your judgment, and the screen is still green? What happens when you dial red in correctly? (The red BIAS number is way lower than the other two colors)

What TV/display do you have?
cyberbri is offline  
Reply Display Calibration



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