Originally Posted by JohnAd
While I have no plans to do that as I think everybody does agree on where to set black, I will make the following potentially important comment.
If you do look at the distribution of BTB signal in an image where there is some present, there can be quite a lot. It is also very interesting that there is often lots of BTB around any black bars in the film and also near any sharp transitions. My guess when I looked at it before is that the BTB is introduced mostly via the resizing algorithms used and is not really signal.
I agree this is an important point and one that I've tried to make for years. How can it possibly even be a candidate for information when so much of it appears off frame or out of the active image area? Rhetorical.
I think it's also an important point to note the history. Originally, one of the purported benefits of StudioRGB was all about the BTB. There were all the images like "Talented Mr. Rippley", "LOTR", "I-Robot".... We posted histrograms, waveforms, inverted and highlighted pixels, counted occurrences, etc, etc. The same things that are being done here, except, all about the BTB. It was recommend to make allowances for that too.
Most everyone has forsaken the notion of BTB being useful for anything other than calibration and that can be done just fine without it. I don't think anything technical has changed in this regard, material is encoded how and where it has always been. It's only peoples notions about it, and being aware of the costs of an elevated black level that has changed. Once open a time, it was advocated to keep some of it. Just like with WTW.
I've also noted that clamping out BTB can be a form of error correction. All of the BTB pixels that should be black, but aren't, are an error. Clamping will correct these errors before they are passed or propagated.
I do agree that BTB has little or no value, and always have, but that hasn't always been the case here by many in the past. I do think BTB/WTW are related however, I've always felt they were two sides of the same coin, but I know not many others do not and have no issue trashing or dismissing one half, yet clinging to the other half.
Originally Posted by JohnAd
Those pictures I've looked at so far the WTW is either small blown out patches or also looks consistent with scaling ringing.
Also agreed and I think a VERY important point. Perhaps one of the most important. I made similar observations in post 158, 181, and 184 here in this thread and have for years. AlluringReality also noted similar here. I noted that source encoded blowouts will still be blowouts regardless of the cst levels. I also noted the distribution as false echos, halos, and very EE like. Nobody seems to like EE here, and much like the BTB, filter and scaling ring may not want to be classified as valuable information. Clamping would remove this as well.
This again is where I find value in side-by-side A/B images. It affords the opportunity to access, using human perceptions, what it is and its value and contribution to the image. Otherwise, it's just data. As seen with BTB, errors are data too, but nobody want to keep them.
If it's information that adds to the human perceptual value of images, then it should be manifest as such using A/B images. If it's not, it wont.