I'm happy to cheer you up on this one. The update that changed it did two things. It implemented the dithering to work-around the chroma overshoot. It
also detected if the 1DLUT had been replaced, and if so, did not use the dithering so that the calibration remained the same as before.
The following a basically a re-post of an earlier post I made on this thread IIRC. The info first came from Vincent Teoh (who was instrumental in helping us to raise this with LG in the first place) and as I say below, it's easy to check whether dithering is being used.
firmware 4.10.31 (and later) behaves differently when a 1DLUT (either from an autocal session or a "unity 1DLUT") is uploaded to the TV. It adds a special flag.
firmware 4.10.31 (and later) therefore knows whether the 1DLUT was uploaded on firmware 4.10.31 (and later) or not, so it is able to behave in one of two ways:
- no flag; 1DLUT created on an older firmware => behave like the older firmware did, and use the white sub-pixel near black instead of the new dithering. Risk of the overshoot flashing, but this is required to match exactly the TV as it behaved on the day of calibration. Done on purpose so that calibrations remain exactly the same. <== this should make you happy
- flag; 1DLUT created when TV had firmware 4.10.31 (or later) => use the new dithering routines near black, as on the day of calibration.
During the period 4.10.31-4.10.55 we had a particular problem on our 2018 models. These firmwares had a "crushed blacks" gamma curve. This was fixed by LG in 4.10.55 (and later) so that the gamma curve matched 4.20.05 and before, but with the added benefit of the new dithering routines for better near-black detail.
- If you didn't have a custom 1DLUT uploaded (from Calman Autocal, or Lightspace, NB Autocal is not the only method of creating the 1DLUT and putting it in the memory slot in the TV, but it's a convenient "shorthand" Vincent uses in the video) - or a manually done calibration - then you'd have the same problem we all had with crushed blacks.
- Work-arounds to that were calibrate again (few could), or do manual tweaks like raise the 5% black level with test patterns etc which quite a few people did. Others shared their settings after doing do.
- If you manually calibrated again during this (relatively short) time, and didn't expect LG to fix this problem, ie thought they'd leave everyone with crushed blacks forever, then you'd find you would need to manually calibrate once more on 5.* firmwares.
- Or, if you just "waited it out", then fw 5.* restored the original gamma curves from launch used pre-4.10.31.
- If you DID have a custom 1DLUT uploaded (Autocal shorthand again), during that period (and indeed from then on), you were on a newer firmware which knew to operate one of two ways, and so your TV behaved the same as it always did in those modes which were calibrated ("1. no flag" situation above)
- Any calibration done using 1DLUT from fw 4.10.31 or later should look a bit better for dark detail, due to (2) above.
It's clever and pretty much the maximum LG could have done to appease people who had paid for Autocal calibrations on old firmwares with the overshoot problem not losing everything, whilst still working around the overshoot problem. This is dead easy to check in the dark on your TV with a jeweler's loupe which you can buy online for a few quid. Just check the difference of a grey patch between a mode that has been calibrated and one that hasn't (such as "Standard").
NB1: If your calibration didn't use the 1DLUT, and was instead a manual one where you can re-input the values into the TV's menus, then yes as
@Canary_Jules says above (his and my posts crossed over), if you manually calibrated during that one relatively short window of time, your calibration would indeed need to be re-done.
NB2: There is a separate firmware bug which means that if you ever use the 1DLUT for HDR10 or Dolby Vision,
you will always lose the new dithering routines and so it's not recommended to use the 1DLUT for those modes. This is separate to the bugs with the 1DLUT in SDR which are well documented by respected calibrator
Ted Aspiotis in a long post
>here< .
Ps. This is quite a complex set of scenarios, so you might have to read this post a few times, sorry about that but that's the detail of it all. HTH