Thursday, April 15, 2010

If you’ve made your way here, then you’re already well aware that Panasonics recent plasmas have been experiencing a rise in the level of light output at predetermined intervals (known as Minimum Luminance Level, or MLL for short). This rise means that the TV’s ability to render deep black is reduced significantly.


I won’t pretend to know the ins and outs of the technology responsible for this phenomenon, but it appears to be a side-effect of plasmas’ need to increase voltage to the panels cells as the phosphors age. If they don’t get enough juice, they misfire. All plasma manufacturers have this problem to overcome, but it seems Panasonic has inelegantly implemented their software/hardware in a way that leads to this unsatisfactory change and far too soon into the panel’s lifecycle...


If you have clocked up thousands of hours on your plasma and have no idea what the fuss is all about, then lucky you! Walk away now. This blog is for the benefit of those new owners that have concerns/curiosity about the issue, existing owners who suspect they might have a problem, existing owners who definitely do have a problem and are sending their precious TVs away for ‘repairs’ or having a tech come see them. It is about good record keeping and practices when gathering evidence and keeping a journal of the TV over time.




Bye Bye black, I hardly knew thee...



I believe that these TVs aren’t suffering from bad batches, regional/factory variations or any other factors that would lead to isolated cases. The TV’s behaviour should be entirely predictable. It should be possible to nail down patterns of behaviour. What is essentially true for one person with a particular model should be true for someone else with the exact same model with the same hours clocked.



The problem is that we simply aren’t all on the same page. We’ve all got different models and sizes and displays in different states of decay. We also have different ambient room conditions, different calibration settings, different source material and the biggest one.... unique brains which lead to different interpretations of the same thing from person to person.



Another thing we don’t all have is light meters. Many of us have cameras and they can be useful in capturing light. Done right, a photo can give us some insight into what state our TVs are at verses someone else with more/less hours on the clock. Done incorrectly, a photo can not only be completely useless, but can lead to misrepresentation, misinformation, FUD, etc.



Individual photos that have no point of reference have no value. None. Zip. Zero.

There are many people in online forums that have posted miscellaneous shots of their TVs with various settings and various source material on the screen. Those shots aren’t helping anyone and are a massive waste of not only the time of the person producing them, but the time of others trying to decipher them.
The first problem is that we can more or less portray anything we like with a photograph. You can make the TV look like it has deeper than Kuro blacks, or you can make it look like its washed out beyond the worst LCD you can possibly imagine. Different settings and different source material = different result.
The second problem is that people will ultimately view them on monitors with a wide range of qualities and settings. They’ll draw false conclusions from already potentially false representations.
There is virtually nothing that can be garnered from these random photos.


Cameras only become useful when they are used to measure the light output of the TV in its darkest achievable state. You cannot do this when your favourite scene from The Matrix is taking up most of the frame.

Photos are also worthless if they don’t have a reference point - A point of comparison. The best point of comparison is your own photos taken when the TV was in a different state (pre-rise). Most people will miss the boat on this simply because it is not something most of us have had the chance to consider, let alone document. By the time we realise we needed to do it, it’s too late.


If you are starting fresh, have a new or recently fixed TV or are about to send one out and hope to be objective about what comes back, then you should be planning on keeping a record of it. If your goal is just to have your own journal of your TV, then you can use any camera settings and any conditions you feel happy with that will allow you to note changes from one state to the next. Conditions need to be as close to identical as you can get them, same ambient lighting, same source (black frame) on the TV, same warm-up period/IR clearance, same camera distance/angle and exposure settings.



However, I think that we’d all benefit from following a standard so that at least there is some point of comparison if people want to compare what a fellow owner has.



The settings I used were already choosen by multiple AVS forum members, so I followed suit.

Sadly, there is only a very small pool of owners who have before and after shots to share. I’m (un)lucky enough to be one of them.


This first shot was taken soon after the break-in period at around 120 hours
-


I was slapped with the first rise less than a week later... (approx 160 hours)


The EXIF data that accompanies these photos allows you to see what settings I used on the camera.
Save the photos to your desktop and right click on it and choose ‘Properties’, then go to the ‘Details’ tab.


F-stop 2.8, ISO 200 and 2 second exposure time are what essentially gives the shots their final value. In a dark room these show luminance from the screen area while keeping light in the surrounding area dark.


Many cameras have different levels of JPEG compression and its best to choose the finest setting available to you. Don’t forget to set the date and time on your camera too, it will make logging the results a lot less confusing.



Room conditions were as dark as possible. I was careful not to choose a moon-lit cloudy night to take one shot and then a clear black night to do the next. If you don’t have sufficient blackout conditions in your room then be mindful of the light that is coming in. One night can be very different to the next.



These TVs are a big sheet of glass which will reflect any light that makes its way onto them. You have an LCD PC monitor nearby? Turn it off. The display on your AVRs and STBs etc will also be spitting out light that will end up in the frame.

Any phosphor based display needs to warm up before it settles. Not only do you need to warm the display up, but you have to be able to clear any potential image retention that will raise the light output and fudge the results.
You are going to want to have your black source on for a good half hour before even considering taking a snap. Use a 100% black frame either from the TV’s built in card reader, a PC or games console or other source that you can load a simple JPEG or PNG image to. If you have to use an empty input then you’ll probably have for example ‘HDMI 1’ displayed in the corner. Its going to add an unwanted value to the end result but is better than nothing.


The camera should always be on a tripod or stable surface. 2 seconds is too long to try and hold a camera steady. If you don’t have a tripod, use the timer function on the camera to take an unassisted shot.

Fortunately plasmas are a lot more forgiving than LCDs when it comes to angle, so you’ll be able to get away with a little bit of difference in height/angle/distance to the display (as you can see in my shots). Try match these as best you can though.


And finally, write everything down. The camera settings, the measurements of distance, how many light sources you blacked out...everything. When you go to take follow-up shots days/weeks/months later, the last thing you want is to be 2nd guessing the results with avoidable variables. Consistency should be paramount.



All of what I just said would apply to a light meter reading too. Light meters are just like cameras they are a tool. In the wrong hands, they both give equally questionable outcomes.




So you’ve taken your photos, now what?



To address the problems created with assessing these images on a monitor of questionable value, or as I recently discovered, a browser that has a gamma bug that crushes black, it helps to take these shots to a higher authority. Photoshop.



Photoshop has some nifty tools to allow us to read the luminance levels of photos to a degree beyond what our brains can tell us. We can look at individual pixels, or groups of pixels, or even averages of an area of pixels. Not only luminance, but we can also see colour shift, so if we have a unwelcome light source in the room, or a very pink sky night compared the the more blue-grey one we had last time, the color picker can alert us to it, even if it is a subtle shift.



Even an application as basic as MS Paint allows you to check the luminance of pixels (although it has a different scale to Photoshop, from 0 (black) - 240 (white). Photoshop measures from 0 – 100.
Low light images are of course a bit noisy, so picking an individual pixel isn’t going to tell you much as they will differ from one to the next. However, when you average those pixels together, you’ll get their mean value.


Here are my two shots averaged and in swatch form -


Photoshop tells me that the luminance of the 120 hour photo (the screen area) has a value of 4. The 160 hour photo gives a value of 7. The outside area (the bezel of the TV and beyond) gives a value of 1 in both shots.


Many people make the mistake of using exposures that are too short and F-Stops that are too high that don’t capture enough light and the results make luminance impossible to measure. This is the same if you are using a light meter that isn’t capable of reading your displays lowest black level. You may as well not bother. If you start off low and then you take more shots at a later date that do show a rise, you’ll never know what the percentage difference is between them since your first sample has an immeasurable value. The screen will measure 0, the surroundings will measure 0 and you’ll have nothing to work from
.



Your results vs others



The difference between one mll level to the next is significant enough that I’m confident anyone comparing with similar hours on the clock, similar testing conditions etc will fall within either one or the other of my examples here. And someone who has more hours on the clock and is at a more advanced level of mll rise will not be within tolerance of these examples. They’ll be in a different category again.



There are a huge amount of factors that can make these types of comparisons very difficult, even with best practices and adhering to the same standards. But the same is essentially true for light meter readings too. The more unified data that gets collected, the better we can average it out and find common ground.

There is no doubts that different models behave differently from one another and the rises aren’t set to behave the same in the different generations that are affected (from 2006 onwards apparently) and you cannot directly compare a young Neo PDP with an aged 2007 model etc, so be mindful of that you might not be comparing apples with apples, or apples with fruit period.


Let my experience be a warning to you. You can blaze through 160 hours of usage in under two weeks of heavy viewing. This is the honeymoon period where the TV is new and shiny and for many of us, a bit of an enigma. It would be very easy to miss changes in this period, either dismissing change as a quirk or something that is ‘just meant to be’, or by actively fiddling with calibrations etc that distract from the fact that change has occurred.

160 hours is a shamefully short period in which a TV can degrade. If you want to gather evidence then do it right and do it now. Tick-tock!

5 comments:

  1. Bla bla bla...

    I suppose you take your car back to the dealer as soon it doesn´t go as fast as when it was new, your fluorescent lights when they have dimmed, your phone and laptop once the battery doesn´t last as long as it once did.

    ReplyDelete
  2. It is spelt 'Blah'.

    Where did you see me state that I was, or was suggesting anyone else return the TV?

    Poor quality troll.

    ReplyDelete
  3. As a pending Panasonic G20 owner, I'm obviously very interested in the MLL issue and this makes for interesting reading.

    However, I can't get past the fact that the two LEDs in the second image are considerably larger than in the first. I loaded the two JPEGs into Photoshop as layers and set the blending mode for the top layer to 'exclusion' to see the discrepency. This is a pretty clear indication that the two exposures are different. Either that or your LEDs are getting brighter, too, in which case, you need to take it back, pronto!

    The luminance reading of 1 for the bezel tells me nothing other than the inability of your camera to resolve blacks accurately at that level, which is to be expected.

    Were you using manual colour balance or automatic?

    I'd suggest that, for accuracy, you need to include in the frame a non-black reference.

    As it is, while you're obviously being very thorough, I don't think your test is consistent enough to be conclusive, yet.

    I'd welcome your thoughts.

    ReplyDelete
  4. Sorry for the late reply Steve, the answer is actually very simple. The flaring of the LED in the second shot is because of the one factor I didn't have tight control of, which is the angle of the camera relative to the TV. The distance is set by placing the camera on the tripod directly behind my couch. Having the camera ever so slightly off axis won't affect the levels from the raster area, so its not a deal-breaker.

    Please check the EXIF data for yourself if you are still not convinced.

    ReplyDelete