On Nov 10, 2018, at 9:53 AM, Carsten Kurz via DCPomatic <dcpomatic(a)carlh.net> wrote: DCP-o-matic applies EBU R128 loudness analysis (if enabled in prefs). This is a method targeted at broadcast/home listening experience. Established Cinema loudness measuring standard is LEQ(m). LEQ(m) is not available in DOM (yet). It shares some concepts with EBU R128, but you can't compute a straight relationship between the two. However, after a bit of testing, I found out some relations that make DOM's R128 usable for general cinema applications. Keep in mind, even if you'd use LEQ(m) metering, you would still need to test your DCP in a well calibrated cinema, and even then, results in other cinemas will vary due to complexities not being addressed by common calibration techniques. For all the work I do, DOM's R128 is working nicely. I wrote a longish article here, which contains some advice on usable numbers and practical implications: https://dcpomatic.com/forum/viewtopic.php?f=2&t=980&p=3576#p3576 Something that many people have a hard time to understand is that there is no such thing as an objective correct measure for 'the right' loudness in cinema. Just as there is no objective measure for the color of a sunset in a movie. The established EBU R128 or LEQ(m) numbers are 'complaint' levels, they are max allowable loudness figures. You should not level higher, or you will get complaints from your patrons/visitors, or you violate certain regulations. But these are not target levels that everyone has to adhere to. If you want your movie or ad to play softer - that is okay, and it may be right for your specific work. Just make sure you don't level higher, because that will cause trouble sooner or later. Also, keep in mind, so far, regulations are only in place for ads (max 82dB LEQm) and movie trailers (max 85dB LEQm) in cinema, not for short films or full length features. You are still advised not to create content louder than these numbers. The more your work deviates from the numbers I mention in the forum article linked to above, the more important it is to check your levels in a well calibrated cinema auditorium of appropriate size and proper facilities. - Carsten _______________________________________________ DCPomatic mailing list DCPomatic(a)carlh.net http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic

I'm slow to reply on this. Sorry - been very busy. Thanks to both Carsten and CJ - fascinating stuff. CJ do you have any more info on James Gardiner's work? I can't find anything on his blog about this. For anyone else who's interested, this appears to be the open source implementation of LEQm that CJ mentioned: https://github.com/lucat/leqm-nrt But what is "Luca"? And are you able to say anything about this mysterious "something else"? I often make packages of shorts and getting the audio level of all the films in line with each other has been a consistent bugbear. Levelling everything using EBU R128 (LUFS) gets everything in the right ballpark, but it's not perfect. Some films should be played louder than others due to their content e.g. a music video will want to be louder than a film which is mostly dialogue. But even comparing 2 films which are similar (e.g. both primarily dialogue-based films) sometimes produces surprising results with LUFS. Comparing a film with stereo sound and another with a 5.1 mix also seems to be difficult. I get very limited time to do in cinema tests, so am looking for methods to get as close as possible to a good result without access to a cinema auditorium. Would be great to hear any thoughts from anyone who's been pursuing this... Jim On 12/11/2018 16:34, CJ | Digital Test Tools via DCPomatic wrote: _______________________________________________ DCPomatic mailing list DCPomatic(a)carlh.net http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic

On Nov 26, 2018, at 5:24 AM, Carsten Kurz via DCPomatic <dcpomatic(a)carlh.net> wrote: Am 26.11.2018 um 12:11 schrieb Jim Dummett via DCPomatic: It seems to be Luca Trisciani, running this business: http://dcp-werkstatt.ch/ Would be nice if he would contribute his LEQ(m) code to ffmpeg as an alternative to LUFS/EBU R128. I don't think that LEQ(m) will give more dependable numbers than LUFS in the sense that Jim prefers (automated or 'blind' leveling), but, it just makes more sense to work with the loudness nomenclature that is actually used by SAWA and TASA and in the cinema domain in general. http://www.sawa.com/wp-content/uploads/2010/10/sound_in_cinema.pdf http://tasatrailers.org/whatis.html - Carsten _______________________________________________ DCPomatic mailing list DCPomatic(a)carlh.net http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic

On Nov 28, 2018, at 6:39 AM, Carsten Kurz via DCPomatic <dcpomatic(a)carlh.net> wrote: Hi CJ, good reading. Indeed many people think that an automated approach could solve these issues. Imho, the non-technical issues are the dominant ones. Audience expectations and habits, cinema/chains habits, audio engineers expectations and habits, staff education. I remember watching 'The pursuit of happiness' a long time ago with my aged father-in-law. Not exactly known as an excruciating action loaded ear blasting flic. Well, we couldn't avoid sitting through some advertising before the main movie started. Years later, whenever we talked about cinema in general, or that specific movie, he was still complaining how effin loud that movie was. In fact it wasn't loud at all, it was one damned 30second candy ad right at the beginning of the show that at the time wasn't leveled to any standard and that staff didn't bother to adjust level for. He couldn't remember what the movie was about or anything else, but he still knew that it was too loud to remember it as a joyful experience. - Carsten _______________________________________________ DCPomatic mailing list DCPomatic(a)carlh.net http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic

On Nov 29, 2018, at 4:56 PM, Carsten Kurz via DCPomatic <dcpomatic(a)carlh.net> wrote: Am 28.11.2018 um 18:54 schrieb cjflynn(a)digitaltesttools.com: Okay, you called for it... There really are A LOT of issues concerned here. A number like 5.5 or 7 only serves as an abstract symbol. Just ONE of those many other issues is that modern production techniques make it possible to create audio content with a much higher dynamic range, tailored to the preferences of highly skilled, 'overmotivated' (and I do not even mean that in a negative way) professional audio engineers/film sound mixers. Until not too recently, they were limited by equipment and pre-digital/pre-LPCM standards. Dolby A/SR/SRD was the limit, as well as analog sound recording equipment, mixing methods, etc. Today, they have everything under their hands and facing their ears what they ever dreamed of. Digital Sound FX can create signals with ANY imaginable spectrum, dynamics, level. No distortion added by mics, amps, converters, tape... DC offset anyone? From a very recent impression, I just sat through 'The Children Act' (twice). During both presentations, I found myself adjusting it from our standard 5.5 to 6.5 during the first half hour (I can do that from my smartphone while sitting in the auditorium, hiding it from other patrons below my seat). There is e.g. one shot where Emma Thompson is staring out a closed window, watching her husband packing up and leaving with his car. The sound level of this car is (realistically) very very low. Over a period of maybe 10s, there is no other sound at all (except maybe for some even more subtle ambient sound), and the image shows just the car vanishing. At 5.5, it was below common perception threshold. Yet they bothered putting it in at that low level. And no, this was not a mistake on their side, they wanted it like this, the mix is excellent all over (I probably could have given it a 7 just as well). Now, there are probably more important pieces of audio in a dramatic movie than the sound of a vanishing car through a window. But that is the mesmerising creative level sound recordists, foley artists, and the like are working on nowadays. These soundtracks are not limited by technology or ambition or talent, only by the time these people are granted to finish their work prior to release. And by an aging audience willing to accept it. I very often do these adjustments during presentation, because I watch many of the DCPs we show. Most of the time, I dial slightly higher than 5.5, like 5.8 or 6. There are mixes that simply need to be leveled higher, there are some that really shine leveled higher. Some very few (maybe one each year) need to be leveled lower than 5.5, because they yell at the audience, and they do not contain ANY subtlety. Yes, I do this to my personal liking, and I hope my hearing apparatus is not too far off from our audiences. So far, few (if none) complaints. Admittedly, we do play a lot of dubbed arthouse titles. You may know that the germans prefer to watch/hear dubbed movies (that changes only slowly). One key aspect of dubbed movies here is that, because so many movies are dubbed in germany since nearly a century, our dubbing studios and dubbing actors work on a very sophisticated level (means, they also have plenty of work for tv, cinema and streaming services). There are always good and bad examples, but in general, dubbing is not done cheaply in germany, trust me. Both artists and recording mixers strive to achieve the best speech/dialog intelligibility, and the major difference to location sound is close/proximity miking and booths. They can later add room, reverb, reflections, EQ, whatever is needed to match the scene, but they do all that with priority on intelligibility. If the sound recordist hears the slightest mumbling - another take. So, I think, as far as speech intelligibility is concerned around germany (and similar countries in that respect), slightly lower levels do not immediately hurt that intention of dialog intelligibility (wait, I need to put that word on a shortcut-key...done ). BUT - it still hurts emotional vocal expression. The breathing and trembling is lost. Same as if you would be watching the screen from the very last back row, losing facial expression in all but close ups (if the actor is able to show such). Switch to movies produced locally, usually with location sound, and dialog intelligibility (hah!) suffers a lot. You may remember that report/survey on sound recording mixers monitoring habits I linked to on the ISDCF mailing list a while ago (I think it was also mentioned in a meeting note). I guess many smaller budget, but probably also some higher budget films, are now mixed on smaller, 'close to' nearfield monitors. Speech intelligibility (...) is SOO different between a small, well controlled nearfield studio setup and a typical large cinema auditorium. Even if they QC the mix/movie later in a decent auditorium - most of the talent there already know all the dialog lines from hearing them over and over and following the sheets during production. They are probably still concerned about various aspects of the mix - but I think they will/can no longer concentrate on speech intelligibility with priority, except maybe for some very important parts that have been messed with too much. Okay, I said 'ONE of those many issues', let's finish here for now... - Carsten _______________________________________________ DCPomatic mailing list DCPomatic(a)carlh.net http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic

I did give an overview presentation about Loudness a couple years ago, which was Part I: Loudness In Cinema