AVS Forum banner
Status
Not open for further replies.
1 - 20 of 23 Posts

·
Registered
Joined
·
277 Posts
Discussion Starter · #1 ·
I've come up with a way to adjust greyscale on a projector for only about $60 worth of materials, rather than paying $900 for a TVSPro Optical Comparator, or more for a color analyzer.


An optical comparator is a source of light at a very specific color temperature (color). The basic idea is that you hold the comparator up near the screen while showing test patterns that have various values of grey, and adjust the gain and bias controls on the projector until all the greys are the same color temperature. Joe Kane has written at
http://www.avstandards.com/page3.html that optical comparators can actually work better than color analyzers with some of the newer display technologies.


The parts I used were:

- $16.95 An 18" full spectrum 6500K light bulb from http://www.cinemaquestinc.com . This bulb is actually made by Lumiram http://www.lumiram.com and is the Lumichrome 1XX. It was chosen because it has the highest color rendering index available, CRI 98, which makes the color less viewer-dependent. Lights with lower CRIs would look like different colors to different viewers who might adjust the projectors differently to compensate.


- $7.95 A plastic GE 18" flourescent light fixture with diffuser, switch and cord from Home Depot. To get the right one look for one that handles F15T8 bulbs. This was chosen because of its compactness and smooth diffuser (frosted plastic bulb cover) that doesn't seem to change color temperature much. I actually preferred this unit over the fixture that cinemaquest sells (which I'm returning), because the diffuser smooths out the light, making it easier to see just one intensity of light at a time


- $22 for four Neutral Density filter gels from www.filmtools.com to make different parts of the comparator more dim than others. These were branded as "Lee filters" and were numbers 209, 210, 211, and 299, in strengths of .3, .6, .9, and 1.2 respectively. After I bought these I noticed that cinemaquest literature says they sell a set of Rosco ND gels for $12.95 complete, so you might want to investigate that first; they are probably fine.


Building it was a matter of putting the bulb in the fixture, and then cutting the filters to size and putting them between the diffuser and the bulb. My goal was to divide the 18" into quarters, with the left quarter about 100 IRE, and the right quarter about 20 IRE, and the middle two values in between. Determining which combinations of filters will achieve the correct levels is done by putting the test patterns up on the screen with brightness and contrast ballpark correct, and then experimenting with covering the light with different filters or combinations of filters. It will be different for different projectors depending on the tube brightness, but doesn't have to be exactly spot on. The filter stack I ended up with was like this:


Full width: 1.2 filter

3/4 width: .9 filter

1/2 width: .6 filter

1/4 width: .3 filter


I then put the molded diffuser on over the filters, and put black electrical tape around the edges of the diffuser to cover the places where light was leaking around the edges of the filters. I also covered most of the fixture in black electrical tape, wherever light was leaking through the plastic. It didn't seem too hot, but do this at your own risk; it is theoretically possible that these modifications could cause the fixture to overheat and be dangerous.


So then I put the completed comparator on a speaker in front of the screen, and got to work calibrating my warmed up projector. I brought up the Avia DVD on my HTPC, and switched my Sony 1272 projector into the service mode to adjust the electronic bias and gain controls. By holding down the pattern button I got it to show the Avia test patterns rather than the projector's test patterns. After trying a number of different patterns, I learned that my high-index Draper M2500 screen actually did shift the colors so that the right hand of the screen was more red, and the left was more blue. So it worked best to adjust with patterns that had the greys in the middle of the screen. I alternated between a vertical grey step, and the central rectangles of solid greys of different values. I sometimes used towels to cover the parts of the comparator that were of different intensity from what I was working on... make sure the towels are not colored so they don't add a color cast to the comparator (I confused myself for a little while using orange towels).


I would first adjust the bias (Red and Blue) to some of the darker values on the comparator, and then adjust the gain (Red and Blue) to match the lighter values. There is definitely a lot of interaction between the two controls, so it is necessary to go back and forth a few times to get everyting to look right.


Then I saved the settings to memory, and slightly tweaked the brightness and contrast settings to get those back to where they should be. The proof that I had gotten the greyscale linear was that I could turn the contrast from all the way down to all the way up, and the overall color cast of the image would remain constant (except for when it approached 100%; then it would be more orange because the blue gun goes into saturation).


Because the ND filters on the comparator allowed me to adjust both highlights and lower values, I didn't need to get the greyscale tracked with the more laborious Guy Kuo Lenscap/solarcell/Voltmeter method.


I had adjusted the greyscale manually in the past without a comparator, so it was somewhat close, but not quite right. The difference for me was not night and day, but I definitely noticed that colors (especially fleshtones) looked more accurate, especially when watching NTSC sourced material. One surprising thing is that I saw more shadow detail, probably because the shadows are now correctly grey rather than losing one or more gun's output.


With the diffuser and ND filters on top of the tube, I am not positive I have calibrated to exactly 6500K; It might be off by a few hundred degrees, but that absolute value makes almost no difference as far as I am concerned. The important thing is to get close, and to get the greyscale linear so it is all one temperature from 1 IRE to 100 IRE.


In the couple of years I have owned this projector I had always said I would get an ISF calibration "when the system is finally settled down". But I've realized that I will probably keep changing DVD players, HTPC video cards, HDTV receivers, or line doublers throughout the life of this thing, so it is never going to be totally "done". By having the ability to adjust greyscale (and everything else) myself, I don't have to worry "I'd like to change something, but that would mean another $400 ISF visit"... I can just redo it myself on an as needed basis.


-Tom
 

·
Registered
Joined
·
3,502 Posts
Bravo. Excellent intiative. It is great to see people moving forward and becoming self sufficient. Your optical method isn't as accurate as a fully instrumented setup, but it goes a LONG way towards giving you both a working grayscale and the ability to do your own. Love it.
 

·
Registered
Joined
·
4,137 Posts
Talk about ingenuity! What a great post.


--Jerome
 

·
Registered
Joined
·
833 Posts
Great post, kudos!


A picture of that beast would be really priceless. As they say "A 1000 pictures says more than a single word". Eh, reasonably close, right? :D
 

·
Registered
Joined
·
277 Posts
Discussion Starter · #9 ·
For comparison purposes, I tried the Guy Kuo lenscap/solarcell/voltmeter method to linearize the greyscale once I had used the comparator to determine the proper ratios of the guns to give 6500k white. As Guy said, that is a bit more accurate than the purely optical method I did above. I found that the optical method had yielded R/G and B/G ratios that were off by as much as 15% some levels. Once I did the measurement-based calibration I got the ratios to within about 8% (I couldn't get closer because my projector doesn't have gamma controls). So for best results it probably is a good idea to do the analytical calibration too. If you already have a voltmeter it costs less than $10 for the materials.

http://www.avsforum.com/avs-vb/showt...&threadid=3891
 

·
Registered
Joined
·
3,874 Posts
Here's a bit of an update for potential improvements on Tom's gismo. CinemaQuest, Inc. has revised the filter kits to replace Rosco with GAM neutral-density filters. Joe Kane found a new source for filters while researching a supplier for 'Digital Video Essentials' that he says are better than the Roscos for this kind of purpose. He found the Roscos had a bit of a green tilt. When he shared this info with CinemaQuest, Inc. the filter kits were upgraded immediately.


There is a new bulb on CQ's site from GretagMacbeth that is the most accurate 6500K fluorescent available. It is larger, being a T12 (twelve eighths of an inch, or 1 1/2" if you live in Palm Beach County, FL) and 24 inches long. It's Spectral Power Distribution (SPD) is much better than the Lumiram product. SPD is a more precise method of measuring color performance of a light source than Color Rendering Index (CRI). It's improved performance is due to using a seven-phosphor mix in the bulb. Lumiram's bulb uses four phosphors.
 

·
Registered
Joined
·
316 Posts
Quote:
Determining which combinations of filters will achieve the correct levels is done by putting the test patterns up on the screen with brightness and contrast ballpark correct, and then experimenting with covering the light with different filters or combinations of filters. It will be different for different projectors depending on the tube brightness, but doesn't have to be exactly spot on. The filter stack I ended up with was like this:

Full width: 1.2 filter

3/4 width: .9 filter

1/2 width: .6 filter

1/4 width: .3 filter
I just ordered all of the parts to make this optical comparator and am a bit confused by the above. When covering up the light with different filters, am I suppose to try to match it to the X IRE pattern on Avia? Also, my setup is very similar to Tom's. A Sony 1252 (instead of his 1272) and a Da-Lite Hi-Power 2.9 gain screen instead of his Draper 2.5 gain screen. I would anticipate that I would end up with a similar combination of filters. Does the comparator that Tom constructed have the 1.2 filter for the full width of the bulb and other filters added on top at the appropriate width or the 1.2 filter is only covering the bulb from the 3/4 width to the full width?

Thanks in advance!
 

·
Registered
Joined
·
254 Posts
Hi Guy's ,

This is a great starting place, but there are a few things to consider when trying to set the grey scale with an optical system. The first problem is in the Genes, men tend to have much worse color perception than women and a lot of us are color blind in red green or red blue world. This makes setting up a grey scale hard if you don't have a great color eye. The second problem has to due with the true color index of the the target color. The reason Tom picked a bulb with a CRI of 98 is that you can get to 6500K and still have too much red or green in the color. The true d6500 is measured in an xy value system, it's when the x and y values match the target that you have a true d6500.

This is what the analyzer type systems give you that the optical and voltmeter can't. If you have used one of the monitor calibration tools now available , you will understand this. I have found a way to use a device that sells for around two hundred dollars to due

a good job of setting the color temp and the Grey scale tracking. I would be happy to explain to anyone who is interested just send me an email to Dfried1 @uswest.net
 

·
Registered
Joined
·
475 Posts
Vrinder... glad someone is making use of this. Answers to your questions below.

Quote:
When covering up the light with different filters, am I suppose to try to match it to the X IRE pattern on Avia?
Yes. The idea is to get your brightest quarter of the light about the same brightness as the 100 IRE, get the dimmest quarter of the light about the same brightness as the 20 IRE or so, and the two middle ones in between.

Quote:
Does the comparator that Tom constructed have the 1.2 filter for the full width of the bulb and other filters added on top at the appropriate width...
Yes.


-Tom Morrow
 

·
Registered
Joined
·
316 Posts
Hi Tom.

Thanks for the reply and to the great optical comparator plans!

My last question would be about using the unit. In your original post, you mentioned that you set the comparator on top of a speaker which was in front of the screen. I was a bit confused by this because I would expect that using the unit in that way would result in light hitting the screen and altering the image. Wouldn't it be more appropriate to set the unit somewhere where you could look at it quickly and then look at an unaltered projected image? Hopefully my question makes sense.

Thanks again!

Vitaly Rindner
 

·
Registered
Joined
·
475 Posts
I find that the amount of light hitting the screen is small enough not to make a difference, as long as the comparator is facing the viewer (not the screen). Of course if you can mount it somewhere that no light can hit the screen that would be better. If you have to ability to place it behind and above or below the screen (wiith backing) that would be better.


I think it's best to find a place where you can see both the comparator and the screen simultaneously; small variations are easier to observe that way.


-Tom
 
1 - 20 of 23 Posts
Status
Not open for further replies.
Top