View previous topic :: View next topic |
Author |
Message |
garyfritz
Joined: 08 Apr 2006 Posts: 12026 Location: Fort Collins, CO
|
Link Posted: Thu May 01, 2008 12:11 am Post subject: Primaries & secondaries |
|
|
I measured my primaries & secondaries, and I'm a bit confused by what I see.
See the chart below. Magenta is nearly dead on, only off by 0.001 xy. Green is almost as good, off only 0.007. Red and blue are off a bit more (.025 & .012, respectively.)
What confuses me is yellow and cyan. If Green is almost perfect, and red is "too far out," then why is the yellow point off toward the green side? I would have expected yellow to be halfway between R & G. Would bringing the red in to its proper SMPTE C location bring yellow closer to red??
I assume you want the three dotted lines (between primaries & secondaries) to cross at D65. Mine don't, and I'm trying to decide what that means. Is the intersection point the place where the display **actually** displays 6500K white? If so, and if that intersection point isn't actually on the D65 point at 0.313/0.329, does that mean calibrating to .313/.329 puts you in the wrong place? Or does the colorimeter actually read .313/.329 at the point where the lines intersect, even if that isn't on the color-temp curve?
This probably won't change anything, since I can't fix the primaries or secondaries, but I'm curious...
|
|
Back to top |
|
|
kal Forum Administrator
Joined: 06 Mar 2006 Posts: 17860 Location: Ottawa, Canada
TV/Projector: JVC DLA-NZ7
|
Link Posted: Thu May 01, 2008 12:43 am Post subject: Re: Primaries & secondaries |
|
|
garyfritz wrote: | I assume you want the three dotted lines (between primaries & secondaries) to cross at D65. Mine don't, and I'm trying to decide what that means. |
Yup - that's the way I understand it too. Mine go almost exactly through the D65 point.
Take at mine from my Zenith 1200/Barco Cine8 Onyx (LC with tinted c-elements on red/green):
You're using HD144 tinted lenses with tinted red/green if I'm not mistaken.
I've never thought about how accurate primaries are or are not, so I just spent a few seconds thinking about it and here's what makes sense to me:
The primaries are obviously measured with only one colour being displayed so the light output or brightness (luminance Y value) doesn't really matter. But when you measure secondaries, it's all about the Y levels of the two colours in addition to how accurate the primaries are. If the Y balance is wrong, then the secondaries will be wrong too even if your primaries are bang on
The primary and secondary patterns you used are window patterns just like the 0 - 100 IRE ones for setting greyscale right? Which means they're probably also 100 IRE window patterns. So what are your 100 IRE white window pattern x/y values? My guess is that you're not exactly at 0.313/0.329. Looking at your graph, your cyan point is closer to green than blue so at 100IRE your green level's probably higher than blue right?
At 100 IRE my xy values are 0.311/0.336 which gives a DeltaE of 6.6. Not bad but not super-great either. (Good enough IMHO). Can you post your chc file? I'd be curious to see it. I've uploaded mine if you want to look.
Quote: | Is the intersection point the place where the display **actually** displays 6500K white? |
No, I don't see how that's possible. Otherwise your greyscale readings would actually show that wouldn't they? You'd measure and get readings that clustered around that new intersection point.
Quote: | This probably won't change anything, since I can't fix the primaries or secondaries, but I'm curious... |
Understand it is important to knowing what's going on in case you need to fix something!
_________________
Support our site by using our affiliate links. We thank you!
My basement/HT/bar/brewery build 2.0
|
|
Back to top |
|
|
garyfritz
Joined: 08 Apr 2006 Posts: 12026 Location: Fort Collins, CO
|
Link Posted: Thu May 01, 2008 3:23 am Post subject: Re: Primaries & secondaries |
|
|
kal wrote: | You're using HD144 tinted lenses with tinted red/green if I'm not mistaken. |
Right. With Marquee 180 tubes.
Quote: | The primaries are obviously measured with only one colour being displayed so the light output or brightness (luminance Y value) doesn't really matter. But when you measure secondaries, it's all about the Y levels of the two colours in addition to how accurate the primaries are. If the Y balance is wrong, then the secondaries will be wrong too even if your primaries are bang on. |
Actually I only had a little bit of time to do anything with these measurements -- I just re-hung the Marquee and only had an hour or so to play with the Spyder before I had to leave on a business trip. And I hadn't gotten the Marquee fully set up yet. I did a quick pass at setting the grayscale at 30% & 80% but it was still off. When I did my quickie primaries measurement, I didn't have **ONLY** the one tube displaying. I just displayed the primary/secondary 100% windows from my GetGray test disk. Since my cutoffs aren't set right yet, the other two tubes weren't completely cutting off when they should. That might have skewed my primary/secondary readings.
But assuming you had the cutoffs right, then how would you tweak the Y balance between the primaries? The secondaries are measured at 100%. You could tweak the secondaries by changing the gain on the corresponding two primaries, but that would mess up your D65 readings. And tweaking secondaries by messing with gain won't help the secondaries at lower IRE levels.
Quote: | So what are your 100 IRE white window pattern x/y values? My guess is that you're not exactly at 0.313/0.329. Looking at your graph, your cyan point is closer to green than blue so at 100IRE your green level's probably higher than blue right?
At 100 IRE my xy values are 0.311/0.336 which gives a DeltaE of 6.6. Not bad but not super-great either. (Good enough IMHO). Can you post your chc file? I'd be curious to see it. I've uploaded mine if you want to look. |
Umm. I'm trying to remember exactly what I did. I didn't get a chance to measure a full grayscale, and didn't bother anyway because I have more setup to do before it's anywhere close to right. I also have problems with my Spyder that cause it to produce totally whacked readings, so I was taking readings with my Minolta and trying to calibrate the Spyder to that. If I remember right, I measured the primaries with the Minolta and hand-entered them into a file for calibration purposes. Then I generated a correction matrix and used that to correct Spyder readings so the primaries matched the Minolta readings. The Spyder-read secondaries don't exactly match the Minolta secondaries but they're very close. The resulting file (reading primaries, secondaries, and white with the Spyder, and correcting the Spyder readings with the matrix) is attached. I didn't measure a full grayscale but the 100% white in the primary readings is way off (.307/.355) which, hmm, is just about exactly where the three dotted lines intersect. (30% was .313/.329, 80% was .313/.330.)
I'm still out of town (just happened to have the .chc file on my laptop ) so I won't be able to take any more readings for a couple of weeks. Maybe once I get a chance to properly calibrate the projector, so 0% actually fully cuts off the tubes, I might get more accurate readings.
|
|
Back to top |
|
|
kal Forum Administrator
Joined: 06 Mar 2006 Posts: 17860 Location: Ottawa, Canada
TV/Projector: JVC DLA-NZ7
|
Link Posted: Thu May 01, 2008 1:44 pm Post subject: |
|
|
Gary,
You don't have greyscale readings in that file but I can get it from the "White" reading in the primaries/secondaries screen. Your greyscale isn't right. Your 100IRE white window x/y points are 0.307/0.355 instead of 0.313/0.329. That's a DeltaE of 23.4. Compare that to the my file: My 100IRE white window is at 0.311/0.336 which gives a DeltaE of 6.6.
To have accurate secondaries your greyscale has to be set correctly first *and* you have to have a display who's primaries are accurate too. So your blue/green levels aren't right which means that your secondaries will not be right either.
When displaying primaries (100% red, green, or blue) your other two tubes should be pretty much off - if they're not it's not that the cutoff isn't right but that the tube's G2 (or whatever it's called in Marquee's) isn't set right - not to be confused with the Marquee RGBLowEnd control which is actually called G2! (Confusing!). The real "G2" setting is a pot that controls at what point the tubes turn on/off.
But they'll never be completely off - you should still have the raster lit up just a bit even at 0 IRE. Try measuring your primaries with the other two tubes blocked with a piece of paper or something (or turn them off in the projector menus). I bet the primary readings will be pretty much the same a before. You'd have to have pretty screwed up G2s to mess up primary readings.
garyfritz wrote: | But assuming you had the cutoffs right, then how would you tweak the Y balance between the primaries? The secondaries are measured at 100%. |
Cutoff controls the low end not the high end so cutoff has little to no effect on measuring primaries or secondaries. You tweak the Y balance between the primaries by simply getting your greyscale correct in the first place.
Quote: | You could tweak the secondaries by changing the gain on the corresponding two primaries, but that would mess up your D65 readings. |
Your 100IRE D65 reading is already messed up! That's why the secondaries are not right!
Quote: | And tweaking secondaries by messing with gain won't help the secondaries at lower IRE levels. |
Exactly right. Which is why secondaries depend on (a) a flat greyscale and (b) accurate primaries to begin with. Primaries are the starting point. Greyscale then defines how secondaries map along the Y scale.
Your solution is to do your greyscale and then remeasure your secondaries. They will measure a lot better.
The mistake you're making here is measuring your secondaries even before you do your greyscale.
I'll have all this in my guide soon enough... and measuring primaries/secondaries comes last.
Kal
_________________
Support our site by using our affiliate links. We thank you!
My basement/HT/bar/brewery build 2.0
|
|
Back to top |
|
|
garyfritz
Joined: 08 Apr 2006 Posts: 12026 Location: Fort Collins, CO
|
Link Posted: Thu May 01, 2008 4:20 pm Post subject: |
|
|
kal wrote: | Your greyscale isn't right. |
I know. I did a quick-and-dirty calibration at 30%/80% but that's all I had time for.
Quote: | Your 100IRE white window x/y points are 0.307/0.355 instead of 0.313/0.329. That's a DeltaE of 23.4. Compare that to the my file: My 100IRE white window is at 0.311/0.336 which gives a DeltaE of 6.6. |
Right. My 30%/80% points were almost exactly .313/.329, but no guarantees beyond that. Since I never did a grayscale reading I don't know what the rest of the IRE range looks like. I don't *think* I changed anything between doing the grayscale and doing that 100IRE white reading, so it's probably off because of blue hump.
Quote: | To have accurate secondaries your greyscale has to be set correctly first *and* you have to have a display who's primaries are accurate too. |
I was about to say that the grayscale shouldn't affect the primaries. But then I realized the grayscale will adjust the gains, and the result of summing the two primaries will be affected by the gain settings. So yes, you're right.
Quote: | When displaying primaries (100% red, green, or blue) your other two tubes should be pretty much off - if they're not it's not that the cutoff isn't right but that the tube's G2 (or whatever it's called in Marquee's) isn't set right - not to be confused with the Marquee RGBLowEnd control which is actually called G2! (Confusing!). The real "G2" setting is a pot that controls at what point the tubes turn on/off. |
That's one detail I haven't figured out yet. As far as I know, the Marquee only has the RGBLowEnd control, which it calls G2. I don't know of any way to set the tube's actual "G2 cutoff pot" setting.
Quote: | But they'll never be completely off - you should still have the raster lit up just a bit even at 0 IRE. Try measuring your primaries with the other two tubes blocked with a piece of paper or something (or turn them off in the projector menus). I bet the primary readings will be pretty much the same a before. You'd have to have pretty screwed up G2s to mess up primary readings. |
The all-black tube state was bright enough to be quite visible on the screen, so I wouldn't be at all surprised if it skewed the readings. I just hadn't gotten a chance to get it fixed yet.
Quote: | Cutoff controls the low end not the high end so cutoff has little to no effect on measuring primaries or secondaries. |
It could if the cutoff is too high so the 0% tube state is still lit up.
Quote: | Quote: | And tweaking secondaries by messing with gain won't help the secondaries at lower IRE levels. |
Exactly right. Which is why secondaries depend on (a) a flat greyscale and (b) accurate primaries to begin with. Primaries are the starting point. Greyscale then defines how secondaries map along the Y scale. |
I hadn't thought about the grayscale affecting the secondaries, but it makes sense. When I get a chance to play with this again, I'll re-measure the primaries **after** I set the grayscale.
|
|
Back to top |
|
|
Gino
Joined: 22 Apr 2006 Posts: 1363 Location: Trinity Beach, AUSTRALIA
|
Link Posted: Fri May 02, 2008 4:33 am Post subject: |
|
|
Kal, can you explain why 30%/80% is used? Rather than say 20%/80% or 30%/70%?
_________________ ( B ) ( G ) ( R ) Blendzilla Down Under ( R ) ( G ) ( B ) - Tubes of Fury
|
|
Back to top |
|
|
garyfritz
Joined: 08 Apr 2006 Posts: 12026 Location: Fort Collins, CO
|
Link Posted: Fri May 02, 2008 1:53 pm Post subject: |
|
|
There's no particular magic about 30/80%. Some people try to use 10%/100% but that's not a good idea -- better to use something closer to 1/3 / 2/3 so you minimize the error across the IRE range. (Tune at 10%/100% and you might have a huge hump in the middle. Tune at 30/80 and the hump will shift down, being above in the midranges and below in the highest/lowest IREs, and the net error is less.)
|
|
Back to top |
|
|
kal Forum Administrator
Joined: 06 Mar 2006 Posts: 17860 Location: Ottawa, Canada
TV/Projector: JVC DLA-NZ7
|
Link Posted: Fri May 02, 2008 2:45 pm Post subject: |
|
|
What Gary said. 10IRE is also too low to reliably measure with most sensors, especially if your gamma is off (it'll be really dark).
Think of the RGB curves from 0 to 100 as sprint-loaded wires or something. If you hold them in place at 30 and 80 you sort of minimize your average error across the range. Do it only for the middle (50) and it'll likely bend down or swoop up at either end. Do it at 10 and 100 and the middle will bump up or down.
Kal
_________________
Support our site by using our affiliate links. We thank you!
My basement/HT/bar/brewery build 2.0
|
|
Back to top |
|
|
Gino
Joined: 22 Apr 2006 Posts: 1363 Location: Trinity Beach, AUSTRALIA
|
Link Posted: Sat May 03, 2008 5:57 am Post subject: |
|
|
so using that logic, wouldn't 30/70 make more sense?
_________________ ( B ) ( G ) ( R ) Blendzilla Down Under ( R ) ( G ) ( B ) - Tubes of Fury
|
|
Back to top |
|
|
kal Forum Administrator
Joined: 06 Mar 2006 Posts: 17860 Location: Ottawa, Canada
TV/Projector: JVC DLA-NZ7
|
|
Back to top |
|
|
VideoGrabber
Joined: 09 Apr 2006 Posts: 933 Location: Michigan
|
Link Posted: Mon May 05, 2008 3:01 pm Post subject: |
|
|
Gino,
> so using that logic, wouldn't 30/70 make more sense? <
Not really. You can't take a (meaningful) reading at 0, so the measurable/adjustable range is 10-100. 30-80 is +/-20 off that base. That's the logic anyway, but Kal's comments do apply.
_________________ - Tim
|
|
Back to top |
|
|
kal Forum Administrator
Joined: 06 Mar 2006 Posts: 17860 Location: Ottawa, Canada
TV/Projector: JVC DLA-NZ7
|
|
Back to top |
|
|
Gino
Joined: 22 Apr 2006 Posts: 1363 Location: Trinity Beach, AUSTRALIA
|
Link Posted: Tue May 06, 2008 5:05 am Post subject: |
|
|
VideoGrabber wrote: | Gino,
> so using that logic, wouldn't 30/70 make more sense? <
Not really. You can't take a (meaningful) reading at 0, so the measurable/adjustable range is 10-100. 30-80 is +/-20 off that base. That's the logic anyway, but Kal's comments do apply. |
Cool, that makes more sense
_________________ ( B ) ( G ) ( R ) Blendzilla Down Under ( R ) ( G ) ( B ) - Tubes of Fury
|
|
Back to top |
|
|
garyfritz
Joined: 08 Apr 2006 Posts: 12026 Location: Fort Collins, CO
|
Link Posted: Sat May 10, 2008 11:44 pm Post subject: |
|
|
I did a more careful calibration on the full grayscale on my color-filtered Marquee. At first, after a touchup at 30%/80% but no blue defocus, I got a REALLY confusing display: my blue went UP at 90% and 100% !!? (First pic below.) I have NO idea what's up there.
So I defocused the blue a bit, and it looks much better. I still have a bit of a hump in the blue but not much, and I don't want to make the blue even fuzzier to fix that small hump. For the 8500 owners out there, my blue focus was at 47, though I think sharpest is actually about 50. I bumped it up to 59 for the final measurements.
DeltaE is 7.13 or below for 20-100%, color temp is 6374 at 100% but 6458-6668 for 20-90%. Gamma is actually a bit low (I have never gotten around to adjusting my Moome ext box) but I think I'll try it there for a while.
And now, with the grayscale correct, my secondaries are basically dead on.
(The measurement dot waaay off to the left is IRE 0. That's way below what my colorimeter is capable of measuring, but I took the value it gave me. BTW these measurements were all done with my old Minolta TV2150, since the Spyder seems to be much less accurate. It's a pain to type in all the numbers but I think the result should be a lot better.)
Strangely enough, the final results show a bit of a hump in the RED. I didn't expect that. I've never heard of red overdriving the way blue does. I'm running the Marquee at 70 contrast (and still only getting 7.0 ftL off the screen, according to the Minolta!), but I don't think the red should be topping out. Anybody else seen humps in the red?
Gary
|
|
Back to top |
|
|
kal Forum Administrator
Joined: 06 Mar 2006 Posts: 17860 Location: Ottawa, Canada
TV/Projector: JVC DLA-NZ7
|
Link Posted: Sun May 11, 2008 6:48 am Post subject: |
|
|
Not bad at all Gary. Very nice. Similar to mine in that I can get everything from 20 to 100 nice but 10 IRE is off (no big deal).
Don't be afraid to defocus blue even more. If 50 is the sharpest and you're only at 47 that's nothing. For me, ~42 is the sharpest and I'm at 0. Yes, 0. If I put up a white crosshatch pattern you'll see blue halo's when standing up close to the screen but they're not visible at the seating position (12/13 feet back or more from my 8 foot wide screen).
Edit: That blue rise at 90 and 100 IRE in your first picture is very odd. Never seen anything like it. Weird!
Kal
_________________
Support our site by using our affiliate links. We thank you!
My basement/HT/bar/brewery build 2.0
|
|
Back to top |
|
|
garyfritz
Joined: 08 Apr 2006 Posts: 12026 Location: Fort Collins, CO
|
Link Posted: Sun May 11, 2008 10:57 pm Post subject: |
|
|
No, sharp is at 50, I started out at 47, and I ended up at 59. 9 off is pretty significant for a Marquee. I haven't done a lot of viewing with it yet but I suspect the blue defocus might be noticeable at viewing distance on test patterns and credits.
I strongly suspect the wacky behavior at 0-10% is due to the colorimeter more than the projector. It reads 0.0 down there and the display flashes, indicating it's out of its range. (It's only rated down to 0.2 ftL.) 20% is 0.2 ftL so everything below that is questionable.
I've never been quite sure whether to believe the ftL measurements from this colorimeter. It says I'm only getting 7 ftL at 70 contrast, and I got similar readings with my XG and my G70. With just an 85" wide screen, I would expect more than 7 ftL even if it's slightly under 1.0 gain. Maybe if I ever get around to upgrading my screen I'll finally be able to get up to 9-10 ftL or so...
|
|
Back to top |
|
|
GEBrown
Joined: 08 Mar 2006 Posts: 729 Location: Denver
|
Link Posted: Mon May 12, 2008 3:34 pm Post subject: |
|
|
garyfritz wrote: | . . . .
I've never been quite sure whether to believe the ftL measurements from this colorimeter. It says I'm only getting 7 ftL at 70 contrast, and I got similar readings with my XG and my G70. With just an 85" wide screen, I would expect more than 7 ftL even if it's slightly under 1.0 gain. Maybe if I ever get around to upgrading my screen I'll finally be able to get up to 9-10 ftL or so... |
Hmmm . . I get 14.5 ftL at contrast 65 on my 80" wide screen and your tubes are in much better condition than mine.
Something seems out of whack.
I am using the DisplayLT facing the screen about 1 ft away from the screen surface.
Maybe it's just the difference in the screen materials BOC vs WilsonArt.
My 2 cents.
Gary
_________________ Member of the Marquee Maniacs Club
|
|
Back to top |
|
|
garyfritz
Joined: 08 Apr 2006 Posts: 12026 Location: Fort Collins, CO
|
Link Posted: Mon May 12, 2008 3:54 pm Post subject: |
|
|
I have my Minolta sensor about 4" from the screen. My BOC probably has a gain of 0.9 or so. DW is 1.24, so the higher gain would only bump the reading up to 7.0*1.24/0.9 = 9.6 ftL.
I suspect the Minolta's ftL measurement is off. But could that be off if the color measurement is right (which it seems to be)? Seems unlikely, but...
|
|
Back to top |
|
|
Person99
Joined: 09 Mar 2006 Posts: 4901 Location: Flower Mound, TX
|
Link Posted: Mon May 12, 2008 5:45 pm Post subject: |
|
|
GEBrown wrote: |
Hmmm . . I get 14.5 ftL at contrast 65 on my 80" wide screen and your tubes are in much better condition than mine. |
If that figure was correct, then you would be getting about as much light output from your PJ as cliff is getting from each G90 (which have been professionally measured with high quality equipment). I highly doubt that. I bet you are getting closer to 10.
_________________ Dave
A train station is where the train stops. A bus station is where the bus stops. On my desk, I have a work station....
|
|
Back to top |
|
|
Person99
Joined: 09 Mar 2006 Posts: 4901 Location: Flower Mound, TX
|
Link Posted: Mon May 12, 2008 5:51 pm Post subject: |
|
|
garyfritz wrote: | I have my Minolta sensor about 4" from the screen. My BOC probably has a gain of 0.9 or so. DW is 1.24, so the higher gain would only bump the reading up to 7.0*1.24/0.9 = 9.6 ftL. |
I think BOC is closer to .85 gain. And I don't really believe the DW is 1.24 gain.
What is your screen size again? I think you are 45x80, right? That would mean about 225 lumens. Still seems a little low. I would expect close to 280 or so out of the 'Quee. I would think on the BOC you are getting around 10.
_________________ Dave
A train station is where the train stops. A bus station is where the bus stops. On my desk, I have a work station....
|
|
Back to top |
|
|
|
|