the worst video on full/limited range RGB you can find

[Fudoh]

must be today's video by EposVox. https://www.youtube.com/watch?v=7kjZZNT5js4

I get the intention behind it (next to getting as many views as he can) and somewhere down the road I don't disagree with the notion of "using limited range in capture/streaming applications", but the whole explanation around it is so bad, it's really hard to grasp. Full of misconceptions and simply wrong statements that don't help the average user to understand the logic behind it in any way.

A bit of background on the two types of signals might have helped, for example one coming from earliest DVI specs, which tried to offer a digital solution to show the whole 8-bit color range on a computer, the other coming from digital broadcast, the earliest days of YCbCr and the purpose of digital video codecs. It would also have been a good start to demonstrate why clipping and padding is the wrong way to convert one into the other and compressing or stretching ranges is indeed something that's technically not completely unheard of.

Good intentions, but in the end a terrible video.

[Udderdude]

It still amazes me that full/limited is even a thing. How did they manage to fuck that up?

[Fudoh]

Imo limited RGB should have never been allowed. The conversion between RGB and YCbCr should be triggering a mandatory range conversion. Source manufacturers adopted the idea of limited RGB to accomodate badly engineered projectors and the existence of said source adjustment made display and projector manufacturers offer the function by default. A vicious circle.

This still doesn't excuse the mistreatment and the miscommunication about the settings, the reasons why these formats exist and especially how they interact with each other.

[nissling]

It still amazes me that full/limited is even a thing. How did they manage to fuck that up?

Actually it's very easy to understand why both of them exist and it would be surreal if there was just one standard considering how analog signals work.

With so much archival footage having both BTB and WTW information it would essentially be doomed not to be able to include that information in a modern format. Also considering that during broadcasting the signal may be passed through many additional steps that will be more or less impossible to get a complete overview of, and if you were to use full range on such broadcasts where one of the units manipulate the signal. If you attempt to add one step to 255 (if we're talking 8 bit of course), you'll go back to 0 and white would then get black. Having the limited range means you will have much more headroom before this kind of issue would ever appear.

As for computer graphics on the other hand, those are not generated by a sensor or photochemical film where you may have information all over the place. Instead you have essentially complete control over the signal itself (as far as hardware goes, since software can mess this up big time) and there's no information above white or below black since those are... The limits, obviously. When going over to digital it makes no real sense to only utilize the SMPTE Legal range for graphics, as you will always lose information.

While it may be a complete hassle for many users I'd say that those engineers who made the standards were very aware of what they were doing. I do agree on Fudoh however that RGB limited is not favorable at all.

[Joelepain]

I'd say that those engineers who made the standards were very aware of what they were doing

Funny, I have the feeling that it's completely the opposit.
I always had the impressions that for exemple the first spec of HDMI 1.0 is just a big joke. The guys just took the best consumer video link available at the time (DVI) and try to stick audio and other legacy stuff in it with a hammer (no really, using video blanking to send audio, wtf)
Oh and don't forget to remove the dual link possibility so your new standard is less futurproof. Oh and don't forget to ask for fees and royalties for the new crap you just made.
No really, if you look closely to the evolution of HDMI, you could argue that HDMI 1.4 could have been HDMI 1.0 if it was well though from the start.

Now to go back to full/limited range problem. Everything you're discribing may be true in a production environnement, and even then it could be argued (like doing processing and conversion of a 8bit integer signal in a 8 bit toolchain that involve conversions with decimal numbers, yeah you will end up with problems).
But as far as the medium has reached the consumer side, limited range is COMPLETELY USELESS. They're absolutely no point for it in the digital domain.
I can understand for some legacy stuff, like a DVD encoded in limited range (and even then you could argue that it should be the DAC task to convert the digital information at the right voltage level).
But what's the point of loosing 1/3 of the color palette of a Blu ray when the final information should be displayed "as it" and not manipulated anymore ?
What's the point of crushing the range of video games when everything has been calculated in hdr internaly for almost 20 years now ?

And the biggest problem IMO with this full/limited range stuff, is that it's there, but it's seems like there's not really a concrete standard for two device to negociate the appropiate format. So it's up to the end user to understand the thing and do the right setup (when he even can). And it's confusing for everyone, even people who are implementing theses stuff on devices.

[Fudoh]

Let's not forget that we're not really talking modern achievements here. I don't really know when the limited range for digital video was first defined, but it's certainly included in the rec.601 specs from the very early 80s.

So, just to be clear and not to be misunderstood: I get the concept of having both ranges available, I just oppose the use of limited range in RGB colorspace. YCbCr designed around the idea, but RGB never was and while there are several good reasons to have YCbCr in limited range with wtw and btb being options, there's simply no good reason to introduce a dead zone on both ends of the RGB range that could be misinterpretated in any way.

[Joelepain]

To be clear I completely understand why it exists in certain circumtances too (mostly production) but not why it reached the consumer level, and even less why it persists even today (it's there in rec2020 !!! )
And for me it's not a RGB/YCbCr legacy problem. For exemple Canon camera produces h264 video in YCbCr full range (recognised as "YUV full scale" by VLC) which is strange because it's shouldn't exist at all.
But it's completety understandable in this context : you are already crushing the 14bit of raw data from the sensor into a 8bit compressed signal, why loose another 1/3 third of the possible colors with no benefit on the bandwith ? In this case Canon is given you the maximum information it can in a h264 stream even if it's not the standard.
And on the other end there some stupid stuff that are doing the exact opposite for no reason.

[Konsolkongen]

If that video pisses you off, you should check out this guy explaining VRR:

https://www.youtube.com/watch?v=JKRL44S ... e=youtu.be

I've only watched a few minutes, but his explanation of screen tearing on consoles was so far off that I had to stop. No fucking way screen tearing on PS3 was caused by the console outputting too many frames

[nissling]

What's the point of crushing the range of video games when everything has been calculated in hdr internaly for almost 20 years now ?

I'm sorry but I really don't get this argument. All current video game consoles (Switch, PS4 and Xbox One) are providing options for SMPTE Legal Levels and full range for you to chose. With HDR you will sooner or later run into limitations depending on refresh rate, chroma sub-sampling and bit depth. The latter is especially tricky, considering just how extremely few consumer TVs that actually can display more than 8 bit color depth.

To be clear I completely understand why it exists in certain circumtances too (mostly production) but not why it reached the consumer level, and even less why it persists even today (it's there in rec2020 !!! )

It makes less sense to force video content to RGB rull than maintaining it in YCbCr as BTB and WTW are supposed to be intact in the final signal. Also, WTW should be visible on the display all the way up to 255 so leaving that out is a no-no. Full range in YCbCr is risky and by no means whatsoever the way to go engineering wise.

I am not familiar with Canon products but I do know that many Nikon DSLRs (if not all) use the full range available for video recording. While this may give slightly better bit precision (although H.264 is still complete shit for source), you are losing the failsafe feature of primarily WTW with the reference levels set at peak. In some cases it may even become unusable due to the risk of misfires. All of the Panasonic Varicams and BMDs I've worked with have maintained WTW and BTB which makes post so much easier.

TBH I think that the only real reason RGB limited even exist is because some TVs (especially older ones) could have difficulties displaying the entire range due to poor signal processing. Basically offering a 16...235 option for RGB can in some cases - very few nowadays - be the only way to get the levels displayed properly on some sets. In 2020 I think this issue is next to non-existent with possible exception for no-name brands where you have literally no real control on the picture that you're offered.

[Joelepain]

I'm sorry but I really don't get this argument.

My point was just to give an exemple where this concept of limited range does not give any benefits, whichever way you look at it. Yes I know consoles have the option to switch between the two, but the option is there because the standard is there, and half the TV sold since the early 00's just f**k this up and they have to accomodate with that.
Remember that the xbox 360 and ps3 didn't have this option when they were released, it was added through a firmware update. The WiiU never have the chance to get this option. That's just show how the concept is a complete non sense and quite well hidden in the deepest corners of the hdmi specs, if people at Microsoft or Sony didn't get it right in first place.
For quite some time I have a lot of problem with my Asus PC monitor if I plug it with DisplayPort instead of DVI because it reports a "1080p" resolution instead of "1920x1080" and my graphic driver assume it was a TV set and Nvidia took like litteraly 5 years to implement a simple switch between full and limited range.

BTB and WTW are supposed to be intact in the final signal. Also, WTW should be visible on the display all the way up to 255 so leaving that out is a no-no

People will correct me if I'm wrong, but I don't think you're right here. The main concept of the limited range is that your display should be calibrated to display it's blackest shade at 16 and it's whitest at 235. So by definition, informations in BTB and WTW range will be purely and definitely lost. So my genuine wondering is what was the purpose of these information in the first place ?

All of the Panasonic Varicams and BMDs I've worked with have maintained WTW and BTB which makes post so much easier.

That looks a lot like a production environement to me, which I don't have any problem with, and i'm not familiar with so I prefer not to comment (or should I say not be pedantic about, because I know I can be )
But in this case shouldn't you be be using a different format for video recording, like a logaritmic scale ? Or at least set the settings of your camera so you're not always at the limit of the format you're using (in other word, making your own headroom, instead of relying on strange features that is legacy of a 50 years old video standard that i'm not sure was even meant for that in the first place)
And to be honest I am genuinely wondering, if your display is calibrated for limited range, how do you even know you have some stuff recorded in BTB and WTW because you don't see it. If your display is calibrated for full range, how do you know when you exports your project what will finish in the BTB and WTW ranges ?

TBH I think that the only real reason RGB limited even exist is because some TVs (especially older ones) could have difficulties displaying the entire range due to poor signal processing

I don't think the range has any impact of the processing quality, at least the way the range is defined in our case. If you want to do quality processing, with headroom and as few errors as possible, what you need is a complete chain that compute data and store buffer in larger bit depth than your source and destination. But otherwise the full or limited range won't change a thing, you will end up with the same problems in both case.

[nissling]

I'm not going to quote because of simplistic reasons, but to answer your statements.

If the people at Sony and Microsoft doesn't get this right then that's their problem, it's not due to the standards themselves. Microsoft aren't exactly famous for being experts when it comes to graphics and video standards. I honestly don't buy your argument.

When you calibrate to Legal range you're supposed to clip black at 16 and maintain all information within WTW up to 255 (meaning that 254 and 255 should still be distinguishable). Not all systems or sets allow for this however. Many Philips TVs messes this up big time for instance, and some converters will clip any information above 234 (at least for luma). Firstly it's, again, due to how analog video works which also has information below black and above white, with the information above white was always visible on CRT displays back in the day. In fact if you go through some old video tapes and look at them with a waveform monitor you'll realize just how much would get lost if you'd make hard clipping.

But then there's also the factor that if you adjust the white clipping on your display so that all information above 234, you will have a real difficult time getting the white balance right as the calculations will get completely messed up. I know for references that some plasma displays like the Pioneer Kuro did clip at 234 and could still provide good white balance, which are the exception to the rule. But if you're trying to achieve the same thing on a display that functions properly you are doing completely wrong. The standard itself is also called SMPTE Legal Levels 0-109, which also makes sense.

I mentioned the Varicam and BMDs as example to what does right, and I do in fact use logaritmic tone curves with ProRes or RAW whenever doing any somewhat serious work. And even so you will still have WTW and BTB information intact. But on the Nikon and Canon, which are by comparison pure consumer products, it makes really no sense for them to use YUV full range without any way to change it. I've never said that I'm relying on it but it does help for sure. And I really don't think it's worth it getting into any more detail about this as you don't seem to have any real interest in the subject.

When you are working with video and film you must assure that your workflow is function correctly. I've done my homework and are always including WTW and BTB on all exports. You must check all of the software that you're using. I'm with DaVinci Resolve which has clear controls for ranges and levels beyond references.

Poor signal processing also includes inability to actually detect the used range. Say you are using RGB full range on a set that always clips levels below 16 and above 234 with no way to change it, it's essentially FUBAR. And in that context it makes sense to have RGB limited available, tho I'd never use a set like that in the first place.