Well, I think I found some tiny little (possible) bugs in the "official" break-in DVD / slide-show-pack.
Let me show you the first like this...
The RGB stands for a Red, Green and Blue sub-pixel triple ("a pixel"). A thin R means that it's OFF and a bold R
means it's ON. (The actual content has some steps between OFF and ON, so I mean OFF = 0%, ON = anything else but not zero.)
Let's see a part of the "infinite" loop:
||| RGB RGB RGB RGB RGB R
||| RGB RGB RGB RGB RGB
Every sub-pixels are ON for 10 slides but the longest continuous ON time of the Red and Blue sub-pixels is 10 slides and the longest continuous ON time of the Green sub-pixels is only 5 slides.
Now let's talk about the other one: Single channel (fully saturated Red, Green, Blue) slides.
Well, you may think an input with R=100, B=0, G=0 numbers will result in an R
GB state (with completely OFF Blue and Green). But no... this isn't that simple anymore.
Many LCD and plasma manufacturers use wide color gamut phosphors (of RGB LEDs) to make "Joe" to say "WOOOWWW I WANT THAT" on the shop.
And a benevolent manufacturer will offer the standard (I mean standardized, like the ITU BT.709) gamut with software based color space emulation. (This is a real problem alone but I don't wish to discuss the observer mesmerism [which makes this emulation noticeably imperfect] and the problems with the "of-the-shelf" colorimeters. That's another topic...)
So, if you run these slides with a picture mode which applies gamut emulation, then an R=100, G=0, B=0 input can actually be converted to (let's say as a non-real example) R=85, G=3, B=5. (And the reasonable picture modes often apply gamut emulation because the native gamut looks weird, so it's possible that the user will use this kind of picture mode which looks right...)
I don't say that these are real problem in practice (yah, many people used these slides in the last... decade?), but this is the AV Science forum, so please don't hurt me if I am too theoretical and analytical.
My suggestion is to use gray shades only. (It's safer and could be more effective.)
As of the efficiency... I am not sure if the original author of the colored slides took it into consideration (or if it was a "happy accident" because it's a great thing or a simple accident which only makes the whole process less efficient...) that a single-channel slide will let the other two sub-pixels rest.
But I also considered that we want to avoid a possible overheating (I think we actually want to simulate some "efficient but not too unrealistically hard" case. I do fear about the dilatation when we have gas and glass...), so I put some 0% black slides to my package.
(And my earlier package used randomized gray slides but I reconsidered that. I think it's better to make it smoothly -> smoother dilatation and voltage increase....)
Here is my current recommendation: breakin_v2
(I didn't make any approximated calculations about it [too many unknown variables] but I think it is not only safer but a bit more efficient. -> requires less time)
Of course, it can't help with the fluctuating white balance across the grayscale (these are perfect grays but the device colors will randomly fluctuate). So may be it is not a bad idea to calibrate the TV first (not perfectly but "more or less" very fast...) if you have something like "10 point gray balance".
What do you think?