Can someone explain the tone mapping variations?

[4kUltraBD]

Hello all,

As you can see in my signature, I have a budget 4K tv. It’s a 2018 4 series TCL.

Last night I was watching a YouTube review of the Hisense H9F, and the reviewer was using Godzilla KotM to demonstrate how that TV will clip bright highlights, therefore losing details with certain HDR content.

I decided to take a look at what my set was doing with these highlights, and much to my surprise, it didn’t clip them at all.

I’m not sure why my el cheapo TV is capable of displaying these highlights without clipping them, when a much more lauded set will clip them.

My guess comes down to two different things.

1. My TVs tone mapping sort of compresses the information on the disc to squeeze it into the available nits. Or,

2. My TV simply doesn’t have the nits available to blow out the highlights. I’m very aware that my TV has low nits, as it doesn’t get bright enough to properly display HDR highlights as the director intended, but I would assume that would only exacerbate the clipping issues. However, my set doesn’t seem to be clipping them at all. I can still see the details as I’m supposed to (they just aren’t super bright compared to the rest of the screen).

I hope all this makes sense.

[Geoff D]

It's all about the tone mapping, it's doing the dictating as to how the highlights are being handled, not the nits on the TV alone. If the TV was just clipping the content according to how many nits your TV can handle then the highlights would be blown out to buggery.

The flipside of such aggressive tone mapping is that it can make content seem darker. Some movies which have a lot of what's called average brightness (which I refer to as APL, average picture level) i.e. a lot of brightness in the scene in general and not just the brightest highlights, are fine with this as they've got enough brightness in reserve to counteract the mapping. But some movies have a low APL, some are already at SDR-type levels so when tone mapped they can appear to be much much darker than they really should be.

To get around this then some manufacturers prefer to clip the content to keep the APL where it should be and keep the image nice and bright, this is why some of those more lauded sets will clip such stuff. Does that make any sense at all?

[4kUltraBD]

Awesome! Thank you! That’s exactly what I was wondering, and your explanation was perfect.

I’m thinking my sets tone mapping must be super aggressive, in order to keep from clipping things on discs that would otherwise be blown out. This set is noted for not being very bright, but it has decent contrast, and I’m watching in a light controlled room. Plus, it was cheap.

Your APL explanation makes a lot of sense as well. I’m now thinking that may be why I thought Solo looked dimmer than expected. If that disc doesn’t have a high APL, then that means my TV is probably crushing it even further, and it doesn’t have enough overall brightness to overcome the aggressive tone mapping, to accurately display what is on the disc.

Things are making so much sense now. Thanks again Geoff. Your help and knowledge is greatly appreciated.

[Geoff D]

Originally Posted by 4kUltraBD

I’m thinking my sets tone mapping must be super aggressive, in order to keep from clipping things on discs that would otherwise be blown out. This set is noted for not being very bright, but it has decent contrast, and I’m watching in a light controlled room. Plus, it was cheap.

Your APL explanation makes a lot of sense as well. I’m now thinking that may be why I thought Solo looked dimmer than expected. If that disc doesn’t have a high APL, then that means my TV is probably crushing it even further, and it doesn’t have enough overall brightness to overcome the aggressive tone mapping, to accurately display what is on the disc.

Things are making so much sense now.

Spot on. Solo will still look dimmer and darker than most other content anyway because it's supposed to, but tone mapping may yet take it down further. Other torture tests are Arrival (shot by Bradford Young, the same cinematographer who lensed Solo) and Goodfeathers, both with extremely low APL that's down into the double digits for many scenes. Still very watchable, beautiful even, with the proper APL in a light controlled environment but tone mapping can badly affect them.

What doesn't help in GF's case is that the mastering metadata 'container' isn't at 1000 nits but 4000 nits, so if a display is relying on that data alone to tone map then it can destroy an image that's only got 247 nits peak brightness and 66 nits for the brightest average scene!

[4kUltraBD]

Lol! Animaniacs reference for the win!

GF sounds like it could be the ultimate torture test for sets like mine. I love that movie, but I haven’t upgraded to the UHD yet. I’m still salty about the not-so-great BD release it had. I’m now extremely curious to see how the UHD looks though.

I find it odd that there isn’t a single standard for the metadata container. You mentioned some are 1k nits while others are 4k nits. It makes me wonder how TV manufacturers are supposed to work around this.

Perhaps have their tone mapping algorithms ignore the container altogether, extract the metadata, and then apply their tone mapping algorithms to that information? Is that even possible?

I apologize for the noob questions, but I’ve recently been watching TV reviews, and now my curiosity has been piqued in regards to all of this stuff. However, none of the stuff I watched really dug into this sort of information at all. They essentially just give an overview on weather or not they like what they saw, or RTings will give measurements of peak brightness, color accuracy, contrast, etc. Then tell how those measurements compare to other sets they’ve tested.

It’s all rather fascinating stuff, once you start peeking under the hood.

I haven’t seen anyone really talking about tone mapping, unless they were discussing Dolby Vision specifically. Even then, they generally don’t talk about how this information may be processed, or the container the metadata is contained in. It’s more about comparisons to HDR10.

[Geoff D]

Those are very good points, some TV reviewers are still concerned with specs and measurements rather than discerning what the TV actually does with HDR mapping. As for there being no mapping standard it's a source of immense frustration to all of us, so welcome to the club.

The BDA (Blu-ray Disc Association) did mandate in the UHD disc spec that content shouldn't exceed 1000 nits peak and 400 nits average peak for the first two years of the license but that was disregarded immediately by the likes of Warners who put out Fury Road in a 4000-nit mastering container with almost 10,000 nits of peak brightness! They and Sony have used 4000-nit mastering metadata for HDR10 ever since (a handful of exceptions aside for Warners), while most other studios cap HDR10 to 1000 nits.

Sony TVs have historically ignored the metadata completely, remember that clipping I mentioned? They just clip the upper end of the content according to how bright the TV can get, this is fine for something like the ZD9 which can hit ~1800 nits so that's where it clips when set to the proper contrast and luminance levels that follow the HDR EOTF correctly. But for a lower nit Sony TV that can only hit 500 or 700 nits then this clipping approach can be much more deleterious to the highlights on certain movies. Recently though their higher end sets now have their own 'dynamic' HDR10 processing which analyses the incoming image itself and remaps it on the fly, other manufacturers like LG also have a similar system in place.

[4kUltraBD]

Thank you for another detailed response. I’m seriously fascinated by all this stuff.

From what you’re saying, It almost sounds like it’s a bit of a free-for-all with UHD authoring standards, as well as TV’s tone mapping algorithms and technology. If the 1k nits standard is being ignored by the studios, it may as well not even exist. I’m guessing that TV manufacturers are probably more frustrated than even us consumers, in regards to this happening. That’s probably why we’re seeing these different implementations of how HDR is being mapped.

Manufacturers are most likely trying to pick the algorithm they hope will produce the best results, regardless of the metadata container, or the amount of nits used for peak brightness on the discs. I wouldn’t want to be the engineer trying to figure all of this out. Lol!

The amount of R&D that would need to be poured into figuring all of this out has got to be quite substantial, I’d think. Especially when you start adding in Dolby Vision, HDR 10+, etc. They’re probably trying to figure out what trade offs to make, and hope it works out for them. Pick a poison and roll the dice type of thing.

So, now that I understand this better, I looked up my TV, and RTings states it has an HDR real scene peak brightness of 191 cd/m², or a 191 nits. I suppose it “technically” has higher dynamic range than SDR (which I believe is 100 nits, no?), but it’s clearly not delivering highlights as intended.

Truth be told, it doesn’t look as dim as you might think though, especially in a completely dark room and playing a higher APL disc. Really, the only 2 discs that seem kinda dim are the aforementioned Solo and Godzilla ‘98. 2049 looks good, Ready Player One, Pacific Rim 2, Godzilla KotM, etc. Even 2001 has nice bright white hallways...er, bright white rooms I mean.

Thanks again for your assistance and knowledge. I genuinely appreciate it.