Support HDR images in your app [video]
developer.apple.com
developer.apple.com
https://www.iso.org/standard/81863.html
Note that, in this specification’s title, “HDR” refers to luminosity ranges that are broader than sRGB supports, and “WCG” refers to colorspaces that are broader than sRGB supports.
Apple’s video is focused on the HDR component of still images, which is relatively new and not yet finalized into any ISO or W3C standards.
Support for WCG images is already widespread courtesy of ICC profiles, though many holdouts (such as Slack and Discord) continue to damage WCG images posted to their services.
Quoting Apple’s video transcript at this time index:
https://developer.apple.com/wwdc23/10181?time=205
> This specification, TS22028-5, provides a structure for encoding HDR content into existing still image formats without compromising quality.
> The specification requires Hybrid Log-Gamma, HLG, or Perceptual Quantizer, PQ, as the encoding transfer function. These are functionally analogous to the gamma curves used in SDR images.
> The color primaries for ISO HDR files are the BT.2020 primaries
> HDR images are required to be 10 bits or more per component. This means that some formats, like HEIF, can encode HDR, but some others, like traditional JPEG, are not able to be 22028-5 compliant, as they only support 8 bits per component.
PNG only supports 8-bit or 16-bit, making it quite inefficient for 10- and 12-bit HDR. However, JPEG-XL seems to support 10 bits or more per component. This may provide supporting context for why Apple began supporting a new image format this fall.
> for required metadata, both traditional ICC profiles and CICP tags are valid.
https://stage.color.org/hdr/07-Nicolas_Bonnier.pdf (slide “Motivations”):
> Several attempts to use PQ/HLG and BT.2020 for still photography are surfacing
> But so far, the digital still imaging industry has not settled on a reference HDR/WCG image encoding for consumers
> The purpose of TS 22028-5 is to provide requirements and guidelines for HDR/WCG colour encoding of still images
And a bit further down, referring specifically to BT.2100:
> Shall use ITU-R BT.2020/2100-2 colour primaries
Especially with monitors. I am so sick of every monitor with Vesa HDR-400 claiming they're an "HDR" monitor. I'm sorry, but that standard is garbage. It's so bad it's meaningless.
Not only is that a pathetic amount of brightness, but the local dimming required is just awful. You can have a screen with a whopping 4 zones that react terribly and slap an HDR label on it.
HDR-600 is barely any better, frankly.
Trying to play HDR content on a regular SDR screen is a mess.
If you rip an HDR version of a movie:
- VLC does nothing. It shows it as so dark it's unwatchable
- IINA shows it as brightness comparable to an SDR version of the movie, but with a weird green tint
- QuickTime is weirdly in-between, brighter than VLC but not as bright as an SDR version
- Infuse got fixed this year so that it plays HDR content to visually match the SDR version of the same content on a SDR screen (the only player I've found that can on Macs)
And if you want to convert your file in ffmpeg, good luck. I still haven't found the magic settings that reverses whatever entertainment studios are doing to translate their SDR into HDR. (The magic settings that Infuse has somehow figured out.) There's no flag or setting for it -- you have to input a bunch of custom values.
ffmpeg -hide_banner -y -probesize 100M -i "$file" -map 0:v:0 -c:v h264_nvenc -x264-params b_pyramid=2:ref=4:vbv-maxrate=30000:vbv-bufsize=15000 -refs 4
-crf 17 -preset:v fast -profile:v high -b_strategy 2 -rc-lookahead 900 -level 4.1 -pix_fmt yuv420p -color_range 1 -colorspace 1 -color_primaries bt709 -color_trc bt709 -subq 10 -vf zscale=tin=smpte2084:min=bt2020nc:pin=bt2020:rin=tv:t=smpte2084:m=bt2020nc:p=bt2020:r=tv,zscale=t=linear,format=gbrpf32le,zscale=p=bt709,tonemap=tonemap=hable:desat=0,zscale=t=bt709:m=bt709:r=tv,format=yuv420p,zscale=s=1280x720 -c:a copy output.mkvI use an LG OLED TV as a monitor for work (and a game here and there), 4k 120Hz and I'm more than happy with the brightness in HDR content. In fact, it's kind of just at the edge of being too bright for me... given the distance.
That's a huge improvement compared to a lot of the garbage HDR 400 rated LCD monitors.
If you want to say what you think is important about an article, that's fine, but do it by adding a comment to the thread. Then your view will be on a level playing field with everyone else's: https://hn.algolia.com/?dateRange=all&page=0&prefix=false&so...
What you can't do is use someone else's stuff to make your commentary more important.
It's definitely less 'effective' at dominating the discussion with a single comment but that's more or less the point.
Things get called blogspam all the time, and a three sentence post linking to a video will definitely get called blogspam.
And the goal isn't to "dominate the discussion". Nobody else wants to submit this video. Nobody is being shouted over. The goal is to start a discussion about a specific thing, and making a comment on an otherwise boring link is not actually very good at starting a discussion.
In general my complaint is that it's hard to link to a specific section or piece of information inside a larger article.
So letting comments become stories willy-nilly is not my suggestion. This is about having an external story, but no good way to point at it.
And a submission like this can't just be a comment because there's no story for it to be a comment on. The video would not otherwise have been submitted.
And yes the option of "submission plus comment" exists but again that's a bad way to start a discussion when the submission has nothing obviously interesting about it. And it makes the experience even worse for someone that tries to read/watch the entire submission before looking at the comments, when most of the submission is pretty boring and the submitter only wanted to submit one portion.
when the submission has nothing obviously interesting
Again, the argument here is 'what I have to say about an uninteresting thing is interesting enough to be its own HN story'. You can see how this is tricky, right? Like, the point of the site is not uninteresting things. The trivial solution is 'find something interesting to post'.
A comment on what submission?
> Like, the point of the site is not uninteresting things. The trivial solution is 'find something interesting to post'.
Sometimes interesting things are inside larger uninteresting things.
Imagine if you could only link to a domain. Half the submissions people make would become far more difficult to make.
> Again, the argument here is 'what I have to say about an uninteresting thing is interesting enough to be its own HN story'.
It's not. The argument is "this section is interesting enough to be a submission, but it would be unreasonable to expect everyone to find it in this large article/video". The pure form of this has no commentary, just linking to a very very specific part.
This exact submission isn't the purest example since it's more of a caption, but I've seen many pure examples where someone wants to submit a specific section without even a single word of extra explanation. The kind of thing you could do with text fragment links but only in some browsers and only for text.
This submission was just a comment, is what I mean. It wasn't really an HN story.
Sometimes interesting things are inside larger uninteresting things.
Sure but those are almost never interesting enough to be HN stories. There are ways to turn them into HN stories but that's done by adding interestingness. Playing logic games about what is and isn't a comment really isn't adding interestingness, it's simply claiming one can rule-lawyer one's way to interestingness and it seems sort of obvious one can't?
It's not. The argument is "this section is interesting enough to be a submission, but it would be unreasonable to expect everyone to find it in this large article/video".
Well, it is, because those things are rarely actually interesting. Like, you have to demonstrate interestingness not just claim it. And this practice is effectively elevating a claim - my claim to interestingness in someone else's uninteresting thing is so important, it deserves its own title. It's a serious special pleading and you could make the case for it with examples but as far as I can tell, there aren't many or any.
The pure form of this has no commentary
I'm not sure what the 'pure form' is, we're talking about people making up their own titles, i.e. turning their comment into the most important comment.
Incidentally, there is a site dedicated to not-having a distinction between 'comment' and 'story' and the exchange of snippets of things - twitter. It works pretty well for that use case.
Why not? "Apple backs X standard" sounds like enough for an HN story to me.
> Sure but those are almost never interesting enough to be HN stories. There are ways to turn them into HN stories but that's done by adding interestingness.
It happens often enough. Lots of news articles talk about multiple things. Sometimes you even get multiple unrelated stories in the same article. Isolating one story shouldn't be a weird thing.
You didn't reply to my question of "what if you could only link a domain" but I think that comparison is important. Sometimes URLs aren't granular enough.
> Playing logic games
I'm not playing logic games.
> Like, you have to demonstrate interestingness not just claim it.
That's what voting is for. But first is has to be a submission, and also people are supposed to be able to get a hint of what's interesting from the title.
> my claim to interestingness in someone else's uninteresting thing is so important, it deserves its own title. It's a serious special pleading
Again, the "entire domain" thing. Finding an interesting thing inside an uninteresting thing that deserves its own title happens constantly. The difference is that most of the time there's a URL you can use, but sometimes there isn't a URL you can use.
> I'm not sure what the 'pure form' is, we're talking about people making up their own titles, i.e. turning their comment into the most important comment.
The purest form is someone using a direct quote from the article, with no words of their own. Maybe they're even using a heading directly from the article. Then behind that is someone giving a very basic description for something that doesn't have its own title. Because article sections often don't have their own titles.
In the former case they are completely objectively not making a "most important comment". In the latter case, they're doing the best they can, it's not like HN blocks submissions that don't have a preexisting title.
> Incidentally, there is a site dedicated to not-having a distinction between 'comment' and 'story' and the exchange of snippets of things - twitter. It works pretty well for that use case.
And sometimes people submit tweets to HN. HN is supposed to be for anything interesting, no matter the length.
The goal here is not to confuse comments and stories. "Here is a thing I found, here is what it is in a few words, nothing further." is something I would call a valid story.
Every story sounds interesting to someone. Again, the burden of adding interestingness is on the author/poster. Could this supposed Apple support for some obscure standard most people haven't heard of be interesting? Sure. Is it interesting to merely state? No. Seems pretty straightforward.
You didn't reply to my question of "what if you could only link a domain"
I don't understand the question, I guess. The granularity of URLs happens to fit the granularity of HN stories. These things are connected and sometimes the granularity does get broken and the results are often iffy - e.g. posts of individual tweets, posts of random github issues, etc. Such posts are sometimes good but often bad because they don't actually form interesting stories that produce curious conversation. 'What if you could only link domains' just feels like a different version of 'what if HN was just twitter'. And, incidentally, HN does have conventions against certain types of 'domain' linking - if you link a domain that's a collection of things, it will get modded down and you'll be told to pick something from the collection and post that instead of the whole thing.
I'm not playing logic games.
I am!
That's what voting is for.
No, voting is just a part of that. There's a lot of convention/culture and there is very much a convention against letting people shape the conversation with their own gloss on other people's stuff. That's what commenting is for and what titling is not for.
Finding an interesting thing inside an uninteresting thing that deserves its own title happens constantly.
Well, you'll have to find some examples because the ones that come up are almost always of not-interesting things or interestingness special-pleadings.
In the former case they are completely objectively not making a "most important comment".
Of course they are, especially when they are selecting a non-representative part of the article which is quite a lot of the time - in fact, by the process you are describing, by definition. You don't get to tell other users what is important in an article that isn't yours from the privileged position of a title. You don't get to mega-comment, basically. This isn't some law of physics or the only way to run a messageboard but it's hardly some unreasonable mystery of HN. Lots of subreddits and other forums operate similarly, for similar reasons.
And sometimes people submit tweets to HN. HN is supposed to be for anything interesting, no matter the length.
Well, not so sure about no matter the length and people do submit tweets to HN and sometimes it works but sometimes it doesn't, I touched on this above. They're the wrong granularity, there are titling problems, dupeyness problems (there are infinite tweets about the same thing), etc. Tweets do kind of break the classic 'link aggregator' design assumptions. That's not the fault of tweets or the tweeters or the posters of tweets to HN but it's definitely a thing.
"Here is a thing I found, here is what it is in a few words, nothing further.
Link the thing, add your comment.
That's the problem. When there is no URL for the thing!
If the thing could be specifically linked, no comment would be necessary.
As the OP here said, they found a specific section of article to be interesting: https://www.lux.camera/halide-for-ios-17/
And they wanted to share that section they found, no additional words, but clearly there was some motivation to use the original source.
But neither URL goes to the thing.
There are two workarounds. Both are bad.
Making a blog post good enough to be a story requires a huge amount of effort, and is not a fair requirement as an alternative to simply submitting. (And a quick, simple blog post wouldn't be acceptable by HN standards.)
Making a submission with a URL that doesn't go the right place, with a comment that points out the thing, only works when the more generic URL is actually interesting enough on its own. And even then it might be worse off for bundling too much together.
"Halide 2.12: All The Latest iOS 17 Photography Features" is not very interesting on its own. "Support HDR images in your app" is not interesting on its own. With the way you're talking about keeping up standards, surely you don't want more bad URLs submitted.
> All these worlds are yours, except Makeupyourowntitleuropa.
I don't want to allow any more making up of titles than you normally get when the specific thing you're submitting doesn't have a title. Just a bit more flexibility on choosing the "specific thing". A rule that it has to be a direct quote would also be much better than the status quo. Anything to avoid a submission of one specific thing turning into a submission of a broader and 90% unrelated thing.
> at least some evidence
Jesus Christ. You keep acting like this never happens when there's an example right in front of us. That's at least "some" evidence, come on.
Even if you think the title was unacceptable, surely you'll agree that neither of these two URLs is good for submitting the thing that OP found and a good chunk of 50 people found interesting.
Using a post title edit to focus a submission of a broader link — such as an Apple WWDC video about HDR still images support — onto just the piece that you find interesting — that Apple supports an ISO standard — is precisely what is unacceptable to do.
That is the exact editorial behavior that HN specifically prohibits in title edits when submitting. The reasons for this have already been explained at length by dang in hundreds of comments over the years, and the guidelines are written specifically to prohibit the behavior you’re arguing in favor of. Trying to argue against that guideline will get you nowhere in the comments on a low-quality title edit submission, and I won’t be rehashing those reasons with you. If you wish to continue participating in HN, you’ll need to get used to disappointment on title edits, or take your concerns to the site moderators in email: hn@ycombinator.com.
I stand by my original point. My top-level comment on this post is intended to show what could have been submitted, as a blog post about this issue, to HN. We could have had an informative post that provided the relevant news, the supporting context, and even the 2022 presentation about why the standard exists at all.
Instead, all we get was a title edit.
And this link was not going to be submitted otherwise, so focusing on that part didn't steal the spotlight from anyone else.
I really don't agree that the guideline was written to make certain kinds of post impossible. I think it was written to avoid people stealing the spotlight, and it's a little bit too broad for that goal.
> We could have had an informative post that provided the relevant news, the supporting context, and even the 2022 presentation about why the standard exists at all.
Cool. But you shouldn't have to do that. It would be interesting if every post had to be an informative user-written blog post, but as argued earlier HN is not that site.
Or don’t: HN users and mods will continue flagging and reverting editorial title edits by whatever definition the mods are currently using that is unacceptable to you, and nothing will change.
Personally, I’d take the chance of persuading them by sending an email, over the certainty of having my opinion disregarded by not sending an email at all. Up to you, though.
It seems generally better (or at least in keeping with HN's 'DNA') to err on the side of keeping things minimal and trust users to figure them out.
I am aware of this guideline. The problem is that the original title is so... meaningless? that I decided to give the site a title that explains why it's interesting.
In no way do I disrespect the idea behind this rule, I can see how it would lead to clickbait titles and such. I also didn't want to write a blogpost with a better title that simply refers to this resource at Apple's.
For full transparency, I came across this site because this interesting blogpost referred to it: https://www.lux.camera/halide-for-ios-17/
The reference is in this sentence: > Apple wants to avoid a similar fate in still photography, so this year they put their weight behind an open standard called "ISO HDR." Ok, technically it's ISO/TS 22028-5, which doesn't quite roll off the tongue. The important thing is that with iOS 17, Halide captures HDR photos. In fact, it has for years!
I don't want to say "it's ok to break the rule in such a case for such a purpose" because one can't say that with any precision—it will just be heard as "ignore the rule and do whatever you want".
When the ambient light is given by the screen ...
So, presumably, the existence of a standard under development — and the endorsement of it by a major manufacturer of cameras and displays — was considered newsworthy by the OP.
Yes, I'm hoping that the mess that is "HDR" video standards can be prevented in the still images arena.
Hopefully, with Apple's weight behind this open standard, other producers will adopt it as well.
I think the reason you don't want to drive OLEDs as hard (so there's current limitation/ABL) is more about burn-out than burn-in.
I regret that hn doesn't have HDR support so as to properly emphasize that <em><blink><pain>THIS IS NOT A FEATURE.</pain></blink></em>
No thank you! If I’m outside enjoying the sun with sunglasses on then I have zero interest in using a laptop!
But you'll have to turn HDR on and off depending on the content you're looking at. If you're looking at SDR content, it will look wrong if the display is in HDR mode. That sucks, and I hope some day is properly fixed.
OLED is also nice because it has perfect blacks (no local dimming needed), but it struggles with brightness, which is the other half of HDR. We still haven't figured out how to get both.
I think it's terrible. The only time I even remember that HDR is a thing is when some embedded video in a web page gets much brighter than the rest of the screen and the rest of the screen looks dim as a result. It's never a "wow that sun looks amazing" style experience, it's always a "wow most of my screen just got really dark" style experience.
I think maybe HDR has a place in movies and TV shows and even maybe YouTube style videos (not that I've experienced that), but I think enabling HDR for small video embeds on a web page was a terrible decision.
It's very different from the experience I've had in Windows where if you enable the HDR setting everything looks shitty and washed out.
Right now it’s basically OLED in a dark room or don’t bother. Large TVs can get away with FALD, maybe.
If everything is dim in comparison because the HDR is so bright, is that happening only in especially bright scenes, or is that happening all the time?
I'm not talking about this in the context of watching long videos or something, but when just browsing through stuff on sites like Reddit. I'll be clicking on the "expand image" button on a series of posts (using the old reddit design) and then suddenly one of them is a HDR video which makes the rest of the screen appear to dim until I close it. So we're not talking about "especially bright scenes". And I'm sure that if I fullscreened the video and the whole screen was covered by that single piece of HDR media, it would look alright.
My guess is that those videos are filmed on an iPhone in HDR mode.
If the sun is in frame, it should be very very bright, but it shouldn't be big enough to cause a problem.
If someone embeds a video of the sun then they're trolling you.
Right. Those aren't pictures of the sun.
I'm saying Apple's code is doing the wrong thing. Apple calibrated things wrong.
I mentioned full-frame sun as a totally separate scenario just to be thorough.
Windows HDR support is a very mixed bag, just like their early support for high DPI, and I think that’s probably what leads some reviewers to dismiss it.