AVS Forum banner

Status
Not open for further replies.
1 - 19 of 19 Posts

·
Registered
Joined
·
638 Posts
Discussion Starter #1
Using the DVE color bars pattern, I've been calibrating the color on my 60v500 over the last two days. Remember that the DVE color bars pattern contains six colored squares containing each of the primary colors and their mixes: red, green, blue, cyan (blue + green), magenta (blue + red), and yellow (red + green). When the TV is set to display a single primary color, three squares will be colored in (the three squares containing the primary color) and the remaining three squares will be black (because they don't contain the primary color). The goal of color calibration is to make all of the visible squares have equal brightness when their primary color is selected. This will ensure that the color decoder is putting the proper amount of red, green, and blue in each color.


For an example, when you isolate the red "gun" (these TVs don't have "guns" technically, but you can simulate isolating a gun from the menus, and it's a convenient shorthand, so I'll say "gun" even though the TV doesn't have one), the red, magenta, and yellow squares will be colored red (because they contain red) and the blue, cyan, and green squares will be black. Note that the "yellow" square isn't actually colored yellow because you've isolated the red "gun". But the "yellow" square contains red, so when you isolate the red "gun", it shows up as the color red. I'll continue to refer to the squares by their color, with the understanding that the square's actual color depends on which gun is being isolated.


Anyway, say you've isolated the red gun and you notice that the "red" square is dimmer than the "yellow" and "magenta" squares, which are evenly bright. Your goal is to make the red square brighter until it matches the other two. Etc... If you can do the same thing--make the visible squares match brightness--for all three primary colors, you'll have correctly calibrated the color decoder.


Today, I put up the DVE color bars pattern. Then, I isolated each primary color (R, G, and B) and then wrote down what each control did to each colored square. The Hitachi v500 has four color controls: Red, Green, Color, and Tint. Here's what I found:


When the BLUE "gun" is isolated:
  • Red has basically no effect on any square.
  • Green has no effect on any square.
  • Color caused all three of the blue-containing squares (blue, magenta, and cyan) to get brighter or dimmer, but not at an equal rate. As Color got higher (above about 40 or so), the "magenta" and "blue" squares became brighter than the "cyan" square, whereas at lower settings, the three blue-containing squares were roughly equally bright. "Color" can be considered as a way of balancing "magenta" and "blue" with "cyan", but "color" would not work at, say, balancing "magenta" with "blue". However, NO combination of color settings AT ALL caused "magenta" and "blue" to be out of balance with each other, and even at extreme settings (color set to 100), these two were roughly similar to "cyan". In other words, blue seems like a pretty stable color. You have to really try to get the three blue-containing colors imbalanced on the blue gun.
  • Tint caused the "cyan" and "magenta" squares to get inversely brighter and dimmer. In other words, as "tint" was moved to the left, "magenta" got brighter (because it contains red) and cyan got dimmer (because it doesn't). As "tint" was moved ot the right, "cyan" got brighter and magenta got dimmer. Tint can be considered as a way of balancing "cyan" with "magenta" when the blue gun is isolated.


When the RED "gun" is isolated:
  • Red caused the "red" square to get brighter and dimmer. Not surprising. It had minimal effect on the other red-containing squares (magenta and yellow). Red can be considered as a way of balancing the "red" square with the "magenta" and "yellow" squares.
  • Green had essentially no effect on any square.
  • Color caused the "red" and "magenta" squares to become brighter. In addition, as "color" was turned up, the "red" square became brighter than the "magenta" square. Color can be considered as a way of balancing the "magenta" and "red" squares with the "yellow" square, but turning "color" up might require turning "red" down to compensate for the difference between the rate of increase/decrease between "red" and "magenta". In other words, if "magenta" and "red" are equal birghtness, but dimmer than "yellow", you could turn "color" up until "magenta" equals "yellow", but then red will become brighter than "magenta", which can be compensated for by turning "red" down.
  • Tint caused the "yellow" and "magenta" squares to get inversely brighter and dimmer. In other words, as "tint" was moved to the left, "magenta" got brighter and "yellow" got dimmer, and vice versa. This doesn't entirely make sense because both of those colors contain red, but I guess only "yellow" contains green, which Tint also adjusts. Tint can be considered as a way of balancing "yellow" with "magenta" when the red gun is isolated.


When the GREEN gun is isolated:
  • Red had essentially no effect on any square.
  • Green caused the "green" square to get brighter and dimmer. Not surprising. It had minimal effect on the other green-containing squares (cyan and yellow). Green can be considered as a way of balancing the "green" square with the "cyan" and "yellow" squares.
  • Color caused both the "green" and "cyan" squares to get brighter or dimmer, roughly equally, but it had minimal effect on the "yellow" square. "Color" can be thought of as a way of balancing the "green" and "cyan" squares with the "yellow" square when the green gun is isolated.
  • Tint caused the "yellow" and "cyan" squares to get inversely brighter and dimmer. In other words, as "tint" was moved to the left, "cyan" got brighter and "yellow" got dimmer, and vice versa. Tint can be considered as a way of balancing "yellow" with "cyan" when the green gun is isolated.


Final thoughts:
  • Turning color down below about 50 caused the squares that should be black in any color to change from black to shades of colored gray, due to overall color desaturation. Optimally, color would be set to a value above 50 (which I believe would give better contrast between blacks and colors than a lower color setting), but other considerations might dictate a lower color value. For example, you might need to lower color in order to balance magenta and blue with cyan when the blue gun is isolated.
  • Turning "red" or "green" down below about thirty caused the red and green-containing black squares to become shades of colored gray, due to overall desaturation. Ideally, these settings will be high enough that the black squares on any gun are solidly black.
  • Tint is most effective for balancing the "yellow" square between red and green, since yellow contains red and green and tint affects the red/green balance in the picture. If the "yellow" square is too dim on the green gun and too bright on the red gun, adjusting tint towards green will fix both problems. Unfortunately, the "yellow" square is one of the hardest to get right, in my experience, because it is so heavily affected by tint, and because tint also affects cyan, magenta, and yellow.
  • DVE recommends setting blue, then red, then green, and that's a good idea, but in my experience, blue was the hardest to mess up, while red and green were more tricky. A change that made red perfect would tend to mess up green, and vice versa. I found that the best method for me was to get blue roughly right, then red, then green, then go back and forth between red and green until they're both good, then go back to blue and check if it's been messed up. Repeat.


So now you should have everything you need to color-calibrate your v500 set. In the end, I couldn't get everything perfect. On the blue gun, cyan is slightly dimmer than blue and magenta. I could fix this by turning color down (which would cause magenta and blue to get dimmer relative to cyan), but the overall desaturation of the picture is undesirable. I could also address the issue by adjusting tint. This would cause cyan to get bright and magenta to get dimmer, but blue would still be brighter. In addition, adjusting tint that far towards the green caused yellows to be noticeably green. So I just decided to live with a slightly dim cyan. Cyan was also dimmer on the green gun, but again, the tradeoffs to fix this weren't worth it.
 

·
Registered
Joined
·
638 Posts
Discussion Starter #2
Using the principles above, I tweaked the color decoder settings on my Hitachi tonight until I got it just about as good as I think it's going to get. Currently, all colors are as balanced as I think they can be except for yellow. I can't get yellow right. Either it's too red or too green. Currently, I'm opting to have it set correctly on the red gun and too dim on the green gun, because I prefer a slightly redder yellow to a slightly greener yellow.


For those who are curious, here are my current settings:


Red: 51

Green: 55

Color: 20

Tint: 7 notches to the left (towards red) from center.


The extremely low color setting surprised me at first, but it was the only way to pull some of the combined colors into balance with their primary color--for example, magenta and red and cyan and blue. The low color setting meant that all colors were somewhat desaturated, which was when I figured out what the Hitachi's "Color Management" menu could be used for.


The Hitachi v500 series (and perhaps others, but I don't have one of those) has two color menus: color decoder, which I've described in depth, and color management. Color management contains controls for all six of the main colors: red, blue, green, cyan, magenta, and yellow. I determined through experimentation that turning the color management controls up and down did not change the balance of colors at all. For example, if you look at the color bars test pattern with the red "gun" on and then you turn red all the way up in color management, red doesn't get any brighter. But it does get more saturated when you view the test pattern will all "guns" on.


Here is how I used color management. When a color became desaturated due to low "color" setting, I used the color management menu to increase the saturation of that color without affecting the color balance that I had achieved in the color decoder menu. For example, when I pulled "color" down to 20, the cyan squares became especially desaturated. I could measure this objectively by viewing only red and then looking at a cyan square (which, of course, didn't contain red). Since cyan doesn't contain red, the cyan square should have been black when the red "gun" was isolated. Instead, the cyan square was a kind of reddish gray. I bumped "cyan" up in the color management menu until the cyan square was a good solid black. I did the same for other colors that were reddish gray instead of black. Then I switched to the green "gun" and re-saturated any colors that were showing up as grayish green instead of black. And so on for blue.


It's important to only turn the color up in the color management menu just to the point where its squares are fully black (fully saturated), and no further. If you over-saturate the colors, your picture will look unnatural. Still, I'm curious what would happen if, having achieved good color balance, I bumped each of the color management controls up an equal amount. Perhaps the more-saturated picture would look better? Can't hurt to try...


Here are my current color management settings:


Magenta: 55

Red: 51

Yellow: 55

Green: 63

Cyan: 66

Blue: 71


This is what it took to re-saturate the colors after balancing red, green, and blue using the color decoder menu.


Remember that adjusting color management does not affect the balance of red, green, and blue at all. So if your cyan square is too light in the green gun, don't expect to fix it by turning cyan up in the color management menu. It doesn't work that way.
 

·
Registered
Joined
·
638 Posts
Discussion Starter #3
Taking into account everything in the previous posts, I will now summarize the steps I took to calibrate my Hitachi's color decoder:


1. Put up a color bars pattern. I like the DVE color bars, but the SMPTE color bars work just as well (and better in some ways). Whatever. As long as the pattern contains one section of each of the six main colors (red, green, blue, cyan, magenta, and yellow), it'll work.

2. Isolate the blue "gun" from the Hitachi's color decoder menu.

3. Adjust "tint" to balance the brightness of the cyan and magenta squares. Adjust "color" to generally balance the blue and magenta squares with the cyan square. If the blue square and the magenta square are not equally bright, then I recommend adjusting "tint" so that the magenta square is balanced with the blue square. The cyan square will be dimmer than those two. Your blue-greens will be slightly green.

4. Isolate the red "gun" from the Hitachi's color decoder menu.

5. Adjust "tint" to balance the brightness of the yellow and the magenta squares. Remember that "tint" makes one brighter and the other dimmer, so there will be some middle point where they are equally bright. Once they are equally bright, adjust "red" up or down so that the red square matches the yellow and magenta squares. You should be able to make all three red squares match using this method.

6. Isolate the green "gun" from the Hitachi's color decoder menu.

7. If you're lucky, the yellow and cyan squares will already be equally bright. If they're not, adjust "tint" so that they are. At this point, you will have messed up the yellow and magenta squares on the red gun, so you might have to find a happy medium between green and red.

8. Adjust "green" so that the green square matches the brightness of the yellow and cyan squares.

9. Go back to red. Repeat adjustment of "tint" and "red".

10. Go back to green. Repeat adjustment of "tint" and "green".

11. Continue steps 9 and 10 until you've got red and green as good as they're going to be. If you're stuck and things just aren't getting better, consider lowering or raising "color" and see what effect that has. You can see that "tint" is a double-edged sword since it affects red and green inversely. As you fix one, you're inevitably screwing the other up. "Color" can sometimes trump this process. Look at the chart above to see which squares "color" will affect on each gun and consider whether that'll help you.

12. Isolate the blue gun. Have you messed anything up?


At this point, you've just got to do your best. I couldn't get things past a certain point until I allowed myself to turn "color" down to a ridiculously low point. If I didn't have the option of re-saturating the colors using color management, I would have been much more hesitant to do this.


Once you've got the color decoder as good as possible (all colored squares equally bright on all guns), the next step is to re-saturate colors using color management.


13. Isolate the red "gun". Examine the squares that are supposed to be black. Are they good, solid black, or are they a little bit (or maybe a lot) reddish? Make a note of which squares are reddish.

14. Go to the color management menu and enable "user colors". All colors are set to 50 by default. Bump the colors that were reddish up to 55. You might not even want to bump them up that high--maybe just go one notch at a time. But keep in mind that some colors will need much more re-saturation than others, so if you're not getting anywhere, try bigger jumps. Trial and error is your method here.

15. Go back to the color decoder menu and isolate the red gun again. Are the black squares a good solid black?


Repeat until all black squares on the red gun are a good solid black.


16. Turn on all "guns" and check to make sure you haven't made any color garishly saturated. If you only turn the colors up until their squares are good and black--and no more!--you shouldn't have made any garish colors.

17. Switch to the green gun. Repeat steps 13-16.

18. Same for the blue gun, although in my experience, all the black boxes were fine on the blue gun.


You are done.
 

·
Registered
Joined
·
638 Posts
Discussion Starter #4
By the way--make sure you have set your color temperature BEFORE you go through this hassle, because color decoder settings are stored for each color temperature, so if you have to change the color temperature you'll have to repeat this process. Personally, I prefer the "middle" color temperature. The "high" one is clearly too blue for movies, although it is nice for when you use the TV as a computer monitor. I thought that "standard" was too red, even though it might be "standard".
 

·
Registered
Joined
·
638 Posts
Discussion Starter #5
A final thought before bedtime... in case you're wondering whether this is all worth the hassle (but who on this forum would wonder that?), let me say that the difference in picture quality is dramatic, impressive, and completely un-subtle. It's just not possible to describe the amazing difference that calibrating the color decoder made. What's more amazing is that the changes between "better-than-default-but-still-not-perfect" and "oh-my-god" were relatively small. You can go a mile and get 50% of perfection, and the other 50% comes in the last half inch. I'm more convinced than I was that ISF calibration really can make a huge difference in a set's image quality even if the calibrator makes relatively small changes.
 

·
Registered
Joined
·
1,951 Posts
Thanks for such an in depth post. I added a link to this thread to the "Master Thread". I really appreciate the time you took here. I'm surprised by your choise of color temperature, but you ended up with the one I didn't try.


Cheers
 

·
Registered
Joined
·
1,734 Posts
loraan,


thanks for all the effort! you did a very good job explaining the difference bet color mgt and color decoding. i suppose for me, the only thing that confused me was your selection of Medium instead of Standard for your White Balance. Joe Kane in DVE suggests a starting point as close as possible to 6500K which in our case is the standard setting. That is, assuming our sets reflect what is in the spec sheets. I would be curious to read what your final results would have been if you used standard as your starting point instead of medium. Maybe you tried this and did not like where it took you.....don't know. Anyway, great posting as usual from you and glad you ended up with a Hitachi.
 

·
Registered
Joined
·
871 Posts
Loraan-

Great effort. I've been trying to get to finishing all this for a few weeks now. We've been recently discussing this in the various hit threads- color decoder vs. color management and what does what. I'll print out your write-up see what happens.


Regarding what manpig wrote- I wonder if it's possible that you had to turn color (blue) down so much because you used med color temp- which tends to add more blue, so you had to compensate for this by turning down color. I suppose it's possible that your med is at a lower color temp than mine is. I see no red added by using standard CT, but definitely do by using b and w. I don't have down exactly how all this works so I don't know if that would be the case at all.
 

·
Registered
Joined
·
120 Posts
Quote:
I suppose it's possible that your med is at a lower color temp than mine is.



When I was having the pink push issue I went into the SM to try to compensate. I really screwed up my drives and cuts for the various color temp modes. I, like a moron, did not write down my originall settings. I went to 2 CC's and 1 Sears, went into the SM's and verified the drives and cuts are all the same on each one. This is contrary to what mfusick told me, and makes me believe no factory cal is done for each individaul tv.
 

·
Registered
Joined
·
638 Posts
Discussion Starter #10
Amazing Smooth and manpig: My choice of color temperature was totally based on my personal preference. I initially selected "standard" but it looked too red. However, bear in mind that that was before I calibrated the color decoder, so that could explain it. I chose Medium because it was just a little cooler than Standard but not as blue, blue, blue cool as High. Can anyone confirm that Standard actually does track closest to D65? After all, there's no guarantee that just because Hitachi called it "standard" it's actually close to the real standard.


That being said, now I'm curious and I'll go back and calibrate Standard and see what results I get.


About black-and-white: yeah, it is definitely reddish. I noticed that if you set Color to zero, the set makes an abrupt transition into grayscale mode. I mean, it goes from "simulating grayscale by combining colors" to "full grayscale with zero color". At least, that's my impression from fooling around with the color decoder. I'm considering setting Color on my black-and-white color temperature to zero. Then, when I watch a black and white movie, I could really watch it in black and white. Only problem is that I'd miss color effects like in Schindler's list (the red-dress girl).
 

·
Registered
Joined
·
638 Posts
Discussion Starter #11
manpig: Color decoding vs. color management really confused me when I first got the set, but the scientific method prevailed :) I determined the effects of color management by adjusting one control at a time and looking at the change in the color bars pattern with all guns on vs. with a single gun selected. The surprising thing was that, when I turned a given color all the way up in color management, squares of that color didn't get any brighter at all with a single gun selected. I concluded that color management couldn't be used to work around problems with imbalanced colors.


With all guns on, there was a clear visible difference in the squares when I turned a single color all the way up, but I didn't know how to describe it. I noticed that one difference between color management and color decoding was that color management ONLY affected the square of the color you were adjusting--for example, turn magenta all the way up in color management and only the magenta square changes, which is different than color decoding, in which many color settings interact.


Finally, I noticed that turning a given color down in color management caused desaturation of just that color, which was the key to figuring out what it really does. I was sitting there, not wanting to turn "color" down as low as it needed to go, because it would desaturate the colors, when it hit me that I might be able to use color management to re-saturate them afterwards. Sure enough, it worked.


In conclusion, I believe that color management adjust the saturation of individual colors. It should be used to compensate for color decoder settings that result in excessively desaturated colors. For example, if you need to turn "red" down in color decoding to the point where the mixed-red squares (magenta and yellow) are not sold black on the green and blue guns, you can compensate by adding magenta and yellow in the color management.


I also experimented with lowering saturation in color management, but found that I didn't like the results. Nor did I like the results of raising color management--all colors in equal increments--after calibration. The overly-saturated picture looked muddy and unnatural. IMHO, color management should be adjusted JUST to the point where all black squares are solid black--no more, no less.
 

·
Registered
Joined
·
1,734 Posts
Quote:
Originally posted by loraan
Can anyone confirm that Standard actually does track closest to D65?
SM page 13


White Balance


High: 14,700 + 23MPCD


Medium: 7500K + 0MPCD


Standard: 6500K + 0MPCD


Black and White: 5400K + OMPCD


there are some X, Y tolerences but they don't appear to be significant


Assuming Hitachi is close to these figures OTB then I would assume the best starting point would be standard before adjusting anything else.

One Hitho had his set ISF'd and the calibrator used Standard as his starting point.


Wonder why Hitachi chose such a freekin high temp jump bet Medium and High? It's almost double the medium setting. Geeze.


I sure would be curious to see what you would come up with (Loraan) using std as your starting point.


PM me if you are sans service manual.


Pig
 

·
Registered
Joined
·
1,951 Posts
I don't think the color chart is linear so that a doubling of temperature may not correspond to a doubling of light color--I don't even know what that means.
 

·
Registered
Joined
·
638 Posts
Discussion Starter #14
I just finished calibrating with standard color temp. Here's what I got:


First number is standard color temp, second is medium.


red: 68, 51 -> decrease 17

green: 55, 55 -> same

color: 8, 20 -> increase 12

tint: 7 notches left, 7 notches left -> same


magenta: 80, 60 -> decrease 20

red: 56, 50 -> decrease 6

yellow: 70, 55 -> decrease 15

green: 85, 66 -> decrease 19

cyan: 90, 71 -> decrease 19

blue: 90, 71 -> decrease 19


All color management settings decreased when going from standard to medium. This was definitely due to the overall desaturation caused by standard's very low color value.


Red decreased when going from standard to medium. Color (which generally affects blue the most) increased. We know that medium color temperature is bluer than standard, so intuitively we would expect red to increase and color to decrease. That the exact opposite occurs suggests that we don't properly understand the interaction between color temperature and color decoder.


Green and tint did not change at all between the color temperatures. This is interesting, because one would expect that a change in red might also require an inverse change in tint to compensate.


Color increased by 12 steps. All of the colors in color management decreased by nearly 20 steps, except for red. It might be worth investigating whether this relationship is coincidental or not. Intuitively, it seems as though turning color up should increase saturation to some degree for all colors in addition to its effect on the color decoder.
 

·
Registered
Joined
·
638 Posts
Discussion Starter #15
Here's another data point...


In the numbers that I posted above, you'll notice that, on both Standard and Medium color temperature, "color" is very low (more so on Standard). The reason for this is the need to balance cyan and magenta on the blue gun. Tint is the control that balances cyan and magenta on the blue gun, but tint also balances yellow/magenta on the red gun and yellow/cyan on the green gun. The result is that if tint is set to properly balance cyan/magenta on the blue gun (about three to five clicks towards green), yellow is dramatically out of balance on both red and green.


But turning down color while viewing the blue gun causes magenta and blue to get dimmer without affecting cyan. Therefore, by turning color down, I have been pulling magenta and blue down to cyan's lower level, then using tint to balance yellow on the red and green guns. This worked pretty well, with all colors being essentially equally balanced, but the entire picture was dramatically desaturated. I compensated for this by turning colors up in the color management menu.


And things looked pretty darned good. But I'm well aware that a pretty crappy picture can look "good" when you don't have much comparison, and my sanity-check-alarm was going off like crazy every time I saw "color" down below 20 and all of the color management settings up around 70 to 90! Things were further complicated when, after a little watching, I decided that I really did like the redder "standard" color temperature better than the "medium" color temperature on most movies. "Standard" requires even a lower "color" setting, if cyan/magenta are to be balanced on blue.


So I decided to see what happened if I gave up on having cyan/magenta being balanced on blue. I set "color" to a point where blue and magenta were roughly equal to the background on the blue gun and ignored cyan. It wasn't very far off from magenta anyway. I then calibrated red and green as before. The resulting settings (again, on standard color temperature) were:


red: 46

green: 46

color: 33

tint: three notches towards red


magenta: 50

red: 50

yellow: 50

green: 55

cyan: 55

blue: 50


As you can see, these settings are much less extreme than my previous settings. Unfortunately, it's not possible for me to a/b the two settings to really see which I like better, but I think I like these settings better than I did the settings with low color and balanced cyan/magenta on blue. The picture seems more saturated overall without being noticeably unbalanced.
 

·
Registered
Joined
·
97 Posts
Loraan,


Yet another great scientific post from you! This should be in Hit's owner's manual. Great work.


Like any great work, it provokes more questions and thoughts...


I am glad you ended up with the 'standard' color temp-- at least it sounds like you did.


One point that you only briefly mentioned in your last post was: the goal of setting the color decoder is to match each of the lighter colors to the reference grey background. By default, they should match each other (Or, be as close as you can get).


I am looking for this information, but what I have read, think I read, dreamed?? (someone please correct me if I am wrong) is that the color bars on DVE are:


Gray background => 50% R, 50% G, and 50% B

Red Square => 50% R

Green Square => 50% G

Blue Square => 50% B

Cyan Square => 50% B and 50% G

Magenta Square => 50% B and 50% R

Yellow Square => 50% R and 50% G


Therefore, when you isolate say, the red 'gun', all of the squares that contain red AND the reference background should show 50% R. Changing the settings that you mentioned above adjust what your display shows. Example, if your Yellow square is too red, your display is showing Yellow as, maybe 55% red.


The reason that you set it to the background is because the background signal is only luminence. Therefore, changes to the color decoder and color management will not change the reference background. As you mentioned, this is why you need the background to be 'just' grey and not have any red or blue cast to it-- it is your reference.


Also, I agree with your assessment of the order. I used the 'decoder' settings to work on the balance of the colors-- mainly the 'hue' of the secondary colors. Then I used the 'management' for final touch-ups that I could not correct in the 'decoding' menu.


I have not had a chance to re-run my calibration with the 'black' squares in mind. I have my light color squares matching, or as close as I can get, to themselves and the reference background. However, I did not do any adjusting to the black squares. When you raised the saturation in the 'management' menu to make the squares black when they were supposed to be, did this affect the colors in the other 'guns' when they were supposed to match the background (when they were supposed to be light)?


My quick test on 'Tweak' thread seemed to show that the R and G in the 'decoding' menu had the same effect as the R and G in the 'management' menu. Your assessment in your first post seems to support this:

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

When the RED "gun" is isolated:


Red caused the "red" square to get brighter and dimmer. Not surprising. It had minimal effect on the other red-containing squares (magenta and yellow).


et al

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

Intuitively, R and G in the decoding menu should reduce R and G in all of the squares that contain that color, but it did not seem so in my calibration.


Thoughts?


Thanks.


Dutch
 

·
Registered
Joined
·
638 Posts
Discussion Starter #17
Quote:
Originally posted by dutchparker


The reason that you set it to the background is because the background signal is only luminence. Therefore, changes to the color decoder a color management will not change the reference background. As you mentioned, this is why you need the background to be 'just' grey and not have any red or blue cast to it-- it is your reference.
Good point.
Quote:


Also, I agree with your assessment of the order. I used the 'decoder' settings to work on the balance of the colors-- mainly the 'hue' of the secondary colors. Then I used the 'management' for final touch-ups that I could not correct in the 'decoding' menu.


When you raised the saturation in the 'management' menu to make the squares black when they were supposed to be, did this affect the colors in the other 'guns' when they were supposed to match the background (when they were supposed to be light)?
I think of "decoder" as affecting color balance, while "management" affects color saturation.


When I turn "red" down in "management", the red square desaturates, but the magenta and yellow squares are not affected. Try this. When red is at zero in color management, the red square is completely gray, while all other squares are unchanged. If I do this with the red "gun" isolated, the red square does not get dimmer until "red" gets very low--to the point where desaturation causes it to gray out. If I turn "red" up in "management," then the red square doesn't get any brighter, except at the very end of the scale (close to 80, 90, etc...) and even then, the extra brightness seems to be due to overall saturation of the red square, not because the color balance is being changed.


If "management" were affecting color balance, then adjusting "red" in "management" would affect all three red-containing squares, not just the red square. In addition, I think that it would affect the brightness of the red square much more dramatically than it does.

Quote:


Intuitively, R and G in the decoding menu should reduce R and G in all of the squares that contain that color, but it did not seem so in my calibration.
R and G in the decoding menu did affect all squares that contained that color, but not to an equal degree. The R and G squares were affected much more than the C, Y, or M squares. My general tactic was to use tint to balance Y/M (on red) and C/M on green, then use R and G to bring the red and green squares into balance with the Y, M, and C squares. However, the R control also affected the M square to some degree, and the G control affected the C square to some degree. Y seemed more affected by tint and much less affected by changes to the R and G controls (in decoder).
 

·
Registered
Joined
·
97 Posts
Quote:
Originally posted by loraan
R and G in the decoding menu did affect all squares that contained that color, but not to an equal degree. The R and G squares were affected much more than the C, Y, or M squares.
Got it. Thanks.


This is what I did:

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

In calibrating with DVE, I arrived at values for R and G in the 'decoding' menu. I rechecked these values by pulling up the DVE color bars and cycling through the 'guns' on the TV. They were correct. I then set R and G to 50 in the 'decoding' menu. Finally, I moved my R and G values (43 and 45, I think) to the 'management' menu and rechecked the DVE color bars. They were still right.

*** Conclusion: The R and G in the 'decoder' menu is the same as the R and G in the 'management' menu.

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


Perhaps the change in the secondary colors was so slight that I could not detect it. I stand corrected.


I am also trying to find info on the 'black' boxes portion of your calibration. I am nowhere near an expert, but I have read many JKP articles, different forums, etc, and have not seen anything about getting the 'black' boxes black. I will keep looking.


Dutch
 

·
Registered
Joined
·
638 Posts
Discussion Starter #19
Regarding the "black" boxes--that was my own observation and conclusion, not Joe Kane's. I think that DVE doesn't consider overall color saturation when discussing the color decoder. It only considers color balance. But I found that when I turned color down past a certain point, it didn't just affect blues, but it also desaturated all colors to some degree. What this means, objectively, is that some of each primary color starts to leak into colors that aren't supposed to have any of that primary color--e.g. on the red gun, cyan starts to show a little reddishness; on the green gun, magenta starts to show a little greenishness. On the other hand, if a color is fully saturated, then its square will be completely black when its primary color is not selected.


Now, there's no rule that says that slightly desaturated colors are bad! In fact, maybe Joe Kane's thinking was that a properly-balanced, slightly desaturated picture is better than a fully-saturated, badly-balanced picture. Most TVs don't have controls to manipulate the color decoder, nevermind saturation! And it's true that an over-saturated picture looks unnatural. If you don't believe me, go to color management and turn all values up to 100! Your squares will be SOLID color and SOLID black on the color bars, but everything that isn't a test pattern will look muddy, neon, etc... To me, it's like an oil painting with too-bright paints. But in theory, it's bad for one color to be noticeably less or more saturated than any other color.


That's the basis of my "adjust the black squares with color management" method. The goal of adjusting the color decoder is to balance levels of red, green, and blue in the pciture. The goal of adjusting color management is to equalize the saturation of the six colors. In order to equalize them, we need some objective "zero-point", which I define as, "the point at which the black squares on any single gun appear as fully black and not some shade of the gun's primary color". I define that as "fully saturated" and use color management to set each color's black squares to that point.


In reality, "completely black" is a judgement call, and some blacks are blacker than others. Just like when you're balancing the colors with each other: maybe they're not EXACTLY equal, but at some point you've got to say they're close enough. Maybe the black squares aren't 100% black--you can still see a little green or red in them. The goal, to me, is to get them roughly equally black.


Once I've done that, I pop in a DVD and check out the colors. I especially look at flesh tones (they show up oversaturation in reds, I think) and grass, trees, etc... (they show up oversaturation in greens). If the peoples' skin tones look overly colored or grass looks neon, then my next step is to bump all colors down equally in the color management, which I hope will evenly desaturate the entire picture without emphasizing any one color.

By the way, I especially like these DVDs for checking picture quality:


LOTR 1, chapter 2 (where Gandalf comes to the Shire). Just the first 30 seconds or so, until Frodo laughs and jumps into Gandalf's cart. Check the greenness of the grass in the opening where Frodo is reading; check Gandalf and Frodo's skin tones in the head-shots of each of them as they talk. The colors in certain scenes of this DVD seem overall a little more pronounced than other DVDs, and I think this might be an artistic choice by the filmmakers. For examle, the Shire seems slightly oversaturated and warm, while Moria is very desaturated and cool.


Sleepy Hollow. Not good for checking colors because it's very muted throughout, but very good for checking brightness for shadow detail. Look for detail in the trees during the chapter where Johnny Depp is riding in the coach going to Sleepy Hollow. If the trees are black, darkness is too low; you should be able to see details of the bark.


The Matrix, Lobby Shootout scene. Setting bright/contrast using details in Neo and Trinity's black clothing. Good for finding the right balance between a dark picture with some lost shadow details and a too-bright picture that shows all details.



I can't overemphasize the value of experimentation when performing this kind of adjustment. Not sure if saturation is too high or too low? Bump all the colors up to 100 and then down to 20 and watch what changes about the picture. Now you know what to look for as you're making finer adjustments: the same thing you saw during large adjustments, just in smaller increments.
 
1 - 19 of 19 Posts
Status
Not open for further replies.
Top