> CRF 23 uhhhhh you could've just used Beatrice's video btw

Finally managed to make an account! Keep in mind that CRF 28 of libx265 is equivalent to CRF 23 of libx264. CRF 23 of libx265 should be near-lossless. They don't call the codec high-efficiency for nothing! I don't have Beatrice's video downloaded, when I wrote the comparison for it I just used AnimeTosho to view the attachments. Plus it's 12GiB and so I would've wanted to transcode it anyway, and it's less lossy to transcode from the original BDREMUX. I'm happy with how I've done things.

The actual movie file of Beatrice is just 8.4gb, the rest are extras. And no CRF 23 of anything isn't anywhere near lossless, Beatrice uses HEVC too and their encode is CRF 15.

he is right google the difference between CRF values of x264 compared to CRF values of x265 and you will know you can get away with higher CRF values on x265 thanks for this release sir!

oh no they don't know what they're doing

PromareDayOne, do you have discord?

I don't have a Discord. The best way to get in touch with me is via Reddit. I don't check my PMs very often, but if you leave a comment I'll see it. As for CRF, I've been using ffmpeg professionally (as in, I get paid for it, it's part of my day job) for several years, I know what I'm doing. 8GiB for a movie is pure placebo, you would need superhuman eyesight to notice any flaws with my CRF 23 encode and it's less than a quarter of the size. If you pause the movie it might be possible to spot something, but I've watched it twice through now at CRF 23 and I haven't been able to see a single artifact.

ara ara~

I doubt if camera-like effect at beginning of movie is well encoded with crf23. You probably put a zone with enough bitrate scaling for it, if your encode is that good as you describe. Usually it's easier to put others into bad light rather than justify your decisions.

Hey thanks for uploading it here :)

>tfw a decent x264 release with decent subs is only half possible in this day and age

>As for CRF, I’ve been using ffmpeg professionally (as in, I get paid for it, it’s part of my day job) for several years, I know what I’m doing. 8GiB for a movie is pure placebo, you would need superhuman eyesight to notice any flaws with my CRF 23 encode and it’s less than a quarter of the size. If you pause the movie it might be possible to spot something, but I’ve watched it twice through now at CRF 23 and I haven’t been able to see a single artifact. lmao

Hi @DJATOM, it certainly wasn't my intention to disrespect your release. Aside from my own (and there I'm biased) yours is easily the best. I was just trying to defend my decision to use crf23. But I think you're underestimating the power of x265. Here is a comparison of two different frames between the Blu-ray copy and my encode: https://imgur.com/a/YUxf2QA With the second frame, I chose it at random, it's not cherry picked. The first frame is a keyframe in both BDREMUX and my encode. In the second frame you can see some slight differences, but I assure you they aren't perceptible when the video is in motion. In the keyframe there are no differences whatsoever. (Yes, Imgur compresses images. I did the comparisons with the uncompressed screenshots and if anything Imgur is less favourable to my encode) The section with the scanline effect and visual noise runs at 6404kbps, while the movie as a whole runs at 1688kbps. Remember that CRF's purpose is to keep the perceived quality the same at all times. I didn't put any limits on the bitrate. Btw, a CRF of 15 falls outside of ffmpeg's "subjectively-sane range", that's why I described it as placebo. My intention really isn't to disrespect anyone. I'm just confident in my encode and I don't want people to cast aspersions on its quality when they haven't looked at it themselves. Since I've had quite a few rude comments about my subs that I spent hours and hours on I'm getting a bit touchy.

"but so far every release group has fucked it up" Thanks.

You shouldn't use imgur for comparisons, use https://slow.pics/ or something like this.

Man, I'm really putting my foot in my mouth, aren't I? All I wanted was for a release to include my subs, unmodified, with the fonts embedded. When I said other RGs were fucking things up it was intended in a lighthearted manner. While some RGs definitely have made mistakes (and I describe exactly what these are in the Release Comparison section of my Reddit post) in other cases it's just a difference of opinion. I'm sorry for offending anyone. I've been in a bad mood since I released these subs because I spent hours on them and yet almosr all I've received are complaints and insults. I love fansubs and I thought it would be really cool to release fansubs on the same day that the movie came out so people could watch it straight away. Keep in mind that all fansubs released so far for Promare are to some extent based on mine, so if I hadn't released mine there might still not be any subtitles for this movie. The good news for everyone is that I never intend to release subs ever again.

@PromareWeekOne Come on bro.. You can't let a few people stop you from doing things that you want to do. So far I can see that 3 people are unhappy from your encode and your torrent has more than 150 downloads. It's like stopping just cause 1~2% of audience can't appreciate your work. Personally I would appreciate if you continue with your HEVC encodes cause you know what you are doing. Thanks.

welcome to nyaa

This is just more evidence that getting paid does not mean people actually know what they're doing lmao

tutorial on how to fix this dilemma. 1) head on over to https://animetosho.org/view/promaredayone-promare-english-subtitles-pdo-v1-5-bd.n1219025 2) download all attachments including the fonts 3) download the remux or your preferred encode 4) mux or play with external subs and the fonts installed on your system 5) stfu i don't think the claim that this encode is some kind of magic and somehow managed to make a x265 encode indistinguishable from a remux hold true, but it certainly is very good for its size. thanks for all of uploader's efforts

The first thing to learn about nyaa is everyone in the comments are kind of assholes (most of the time) just realize most people downloading your torrent probably like it but don't bother commenting. At the end of the day you should just make something you like and ignore most of the comments. Personally I think the encode and subs are okay and do the job though they could be improved. Thanks for sharing it.

At the end of the day they're only incompetent

>ignore most of the comments >though they could be improved so ignore the comments that are telling the uploader how to improve? not sure that adds up. also i just noticed that this is an 8-bit encode. there is literally no excuse for not using 10-bit for HEVC. it even has hardware support.

> so ignore the comments that are telling the uploader how to improve? not sure that adds up. More like criticizing him.

I have no issue with telling people how to improve I more mean people who just insult without saying what could be better and why. Meme all you like but at least explain what is wrong and how it could be improved if you want better quality stuff. That is why I say ignore MOST comments.

@herkz yes I agree, Just learn to not be easily offended on the internet and IRL, embrace the criticism and know how to take it, be with the actitude that you will never ever be omniscient. Entitlement to be respected here is naive, if you do what you like expecting others to say only nice things about it; also always there's going to be someone who is entitled enough to be offended and be a dbag in the comments of a free pirate weebsite. People are full of cognitive biases, and the exemplary here is a negativity bias. Sorry for ma bad english, thanks.

Considering the way the guy responded, I highly doubt any legitimate criticism would get through to him either way. ¯\\\_(ツ)_/¯

Thanks!

constructive criticism vs destructive criticism my constructive criticism - you should have use 10-bit - CRF=23 is fine, i agree that CRF=15 is excessive or placebo that it defeats the purpose of using advance compression like x265 or HEVC - no need to install fonts if you are gonna download this MKV release since its already embedded in the MKV - the subs should have been set to default when playing (but i just set it to default now when i re-muxed it with MKVToolNix)

Hey, I'm definitely willing to hear constructive criticism. I've read everything people have written and done comparisons and research to see if there's merit to it. There's not, you people just don't know what you're talking about. With regards to 10bit, the Blu-ray only has 8bit colour. So there's no reason to use 10bit, ffmpeg can't pull extra colour information from nowhere. @heg I actually wasn't aware that you could set subs to default, I'll keep that in mind in case I do an updated release.

Sure, but filtering is typically done in 16bit for the additional precision . Surely you do believe that the banding and aliasing in the source have to be addressed and fixed up, correct? If so, why not do that from the start? You know so much better than us, so why didn't you do the obvious?

@PromareWeekOne maybe you have not read this yet Why does 10-bit save bandwidth (even when content is 8-bit)? more here http://x264.nl/x264/10bit_02-ateme-why_does_10bit_save_bandwidth.pdf

@heg Interesting, I wasn't aware of that! That document is pretty light on details, but considering it's from ATEME then I'm sure it's accurate. Running a test now to see just how much more efficient 10bit is. @LightarrowsEXE I don't run any filters. Filters try to guess information that simply isn't there. Promare looks great without debanding or smoothing, I'm not going to risk softening the image for a miniscule improvement. Just my personal preference, not trying to imply that nobody should ever use postprocessing, it's just not something that I'm interested in.

Sure, but in PROMARE's case I can hardly see why you'd take the stance that this movie [looks perfectly fine](https://slow.pics/c/Oe8ClQK1) without any sort of debanding applied (and don't worry, if you want to find them in the flesh, you can find all the scenes that I decided to fix in my encode [in these frame ranges](https://hasteb.in/ehewecuk.py). They should be easy to spot, a lot of them are quite dynamic and super noticeable if you pay any kind of attention). That and with how you keep telling everyone else is doing it wrong and keep holding that stubborn "I know better than all of you, I get *paid to use ffmpeg*, gnargnargnar" attitude as if you could do no wrong yourself, it's no surprise at all that people are hounding you. Do everyone a favor and actually try to learn how to encode beyond pushing something through ffmpeg and thinking you've done us some great favor with a starved encode. We have plenty awful mini-encode groups that already do just that. Anime isn't live-action, you find much more issues with BDs that need proper addressing than you do with that. If you want a place to start, go join [eX's Discord Server](https://discord.gg/ZB7ZXbN) or something. People will gladly bring you up to speed and help out with what they can if you don't make yourself come over as you did here.

Sorry, I'm certainly aware that I don't know everything and that I make mistakes. I have seen the aliasing in Promare's BD release and can understand why people would want to use a filter to remove it. My intention was to make an accurate encode, not one that smoothes out issues with the original release. If Trigger wanted to anti-alias they could've easily done it themselves, my release was never intended to be a remaster. (And if Trigger had done AA it would look far better than any postprocessed AA, since it would have all the original geometry information from the vector imagery) The reason for my "I know better than all of you" approach was because people were very quick to call me incompetent. I'm not incompetent, I know how ffmpeg works. Yes, my professional experience with it is purely for live action video. But I know what a sensible crf value is, and I know that the conventional wisdom of 23 being "just okay" applies to x264 but not x265. Maybe things would've looked better at crf 19, but crf 15 is unreasonable. I got defensive because it's something that I worked hard on and people were shitting on it without being constructive. I shouldn't have written some of the things that I wrote, and I'm sorry for that. I appreciate your suggestion to join the Discord, but I do still think I'm not going to put any more time into fansubbing. It took hours and hours, it was tedious and boring, and I've met more nasty people than in any other hobby. On top of that, the legal risk is too high in my country. None of this was worth it.

I think it also just felt really unfair that people were nitpicking my video encode settings when at least I did better than most other releases. Leeson and Beatrice both had good releases, but the other releases are terrible. (One of them blindly converted ASS to SRT, most of them don't include the required fonts, one of them included the fonts I recommend but had Kusou's modified script which doesn't use those fonts, and Kusou said they made their own script from scratch, then deleted their upload when I called them out on the fact it was obviously based on mine). I clearly put more work in than them and yet the criticism here has been far harsher.

All good man, I just wanted it to be clear why people were dog piling you, as for fansubbing quality, there are quite a few things that could see some definite improvements. At the end of the day people trash on stuff because they want to see the best, and very few releases ever come close to that. I'd say that crf 15 (@1080p) is acceptable with decent-enough filtering, since the filesizes shouldn't end up unbearably high unless you're dealing with dynamic grain usually. Especially lineart and gradients end up much better preserved, since those tend to get lost really quickly. With live action you run into a case where most of the quality is kinda set in stone, and you can't do much in the way of fixing up any defects—the best you can usually do is just try to get it as small as possible with as few artifacting popping up as possible, which is infinitely easier with live action because due to its nature "everything" is detail. Not so much the case with anime, unfortunately. If you ever do plan on trying out some fansub work again, I'd urge you to hang around fansubbers before attempting something. They'll be able to give you good pointers, and usually they're more than open to help someone new out (assuming they don't act like a shitter or an entitled brat, which unfortunately many people do). I'm sorry to hear that the response to your work has tainted your interest in it, but there's definitely some good and fun sides to it if you ever wish to consider picking it up again. And yeah, I feel ya. A lot of people have no clue what they're doing and are too stubborn to try and improve. If it makes any of this slightly better, there's a group with a very experienced TL working on it right now to bring out a "proper" release. I'm also encoding it for them, and plan to fix ~~and unfortunately destroy in some specific scenes~~ more than the other groups have tried so far. Hopefully that release will not end up in flames (pun intended), but I have hope that it'll end up alright.

Trigger couldn't easily have AA'd considering they have no control over the mastering process, and also the way that media assets are handled the error was either created internally (because of the limitations in bitdepth in the formats used to transfer assets) or externally by the company in charge of mastering. In either case, there's no easy way in the workflow to address the issue. animation is garbage and most products that are used in the workflow have zero interest in addressing these kinds of issues.

Also who gives a fuck about being constructive you put out shit with some bullshit excuse of "I'm a pro at using ffmpeg" if you can't take being called shit for something you put out then don't put out shit.

Thanks @LightArrowsEXE, I look forward to checking out that release. Maybe it was arrogant of me to think I could produce something on my own without talking to others for advice. I didn't really know of any fansubbing communities, especially because I don't use Discord. Whenever I try talking to people on IRC nobody ever replies, probably a timezone issue. @govna Interesting to hear that, that's really surprising to hear about the tooling. Like I said before, my "I'm a pro with ffmpeg" thing was because the criticism that had been provided so far was not constructive or even valid. It was a stupid comment to make, but surely you can understand how frustrating it would be to be belittled by people who haven't given any evidence of knowing what they're talking about. If you hear enough baseless criticism then you start to tune out the legitimate criticism too.

I've since changed how I feel movie sux feel free to do a bad job with it if anyone wants to sub night is short walk on girl tho lmk

@PromareWeekOne Whenever someone tries to be a prick about your encodes just say a single line **"Do it yourself"**. Works all the time.

thanks for subbing, do you have a separate srt file for the subs that i can use for my raw file?

@LightArrowsEXE can you tell me what the name of the group that's releasing the encode you're talking about is? just so i can keep an eye out.

@aptop6 that is legitimately one of the worst comebacks you can give, and to this day it still baffles my mind that people think that's the right way to do it. If taken to some more extremities, it means you can't criticize food if you can't prepare it yourself, or have issues with an anime unless you can write a screenplay yourself. It's incredibly dumb and just leads people to ignore legitimate criticism. @Squeezi Since it's a one-off group and not an established group, we have yet to really settle on a name. The only info so far is that it *obviously* has to have a fire-related pun, and the release will end up on a Trusted account, so look out for those two signs! :)

@LightArrowsEXE Thanks for the answer and for doing it. I'll be looking forward to the release :)

I don't think most anime issues arise during the screenplay

After reading the entire trash section, I'm sure of 2 things, first THANK YOU !! and second wtf people !!!

Man, I love fansubbing drama, I should start a YouTube channel and do dramatic readings of Nyaa comments.

Started watching this today next to Leeson's and I have to say the audio is fucked on this. Media file says it's 5.1 but the mix sounds weird compared to Leeson's. It sounds almost as if it was converted to 2ch but then played back again in 5.1. What's the deal?

Please can you do english sub for Promare SIDE Galo and Promare SIDE Lio ?? https://nyaa.si/view/1219416

I encode quite a lot. Watching crf25 on a 24" screen, there is no difference between it and the original. At crf28 I can notice some graininess. I feel crf23 itself is overkill.

@Deikun, not sure what the issue is with the audio. I don't have a surround sound setup so it's hard for me to test, but I exported the channels separately and compared them in Audacity. https://imgur.com/Irt5eFF I'm only comparing 3 channels here to make it easier to see. First, there's significantly different sound on each channel so it's definitely not 2ch converted to 5.1. Second, the sound is practically identical to the BDREMUX. I know inspecting a waveform visually is not a very good way to compare audio, so you'll just have to take my word for the fact that I solo'd each channel and compared the BDREMUX to my encode and couldn't hear a single difference. I have a pretty nice audio setup too, so I'd expect to be able to notice any difference. So as far as I can tell, my audio is identical to what's on the Blu-ray. Maybe Leeson did some touch-ups/remastering? I don't have his release. The only other thing I can think of is that ffmpeg is unable to decode DTS-HD in full quality, but I rather doubt this. Also the movie sounds fine to me. @LightArrowsEXE @heg I finished my 10-bit test and the file size is reduced from 1.60GiB to 1.58GiB. But the end-result does look *slightly* nicer in a few scenes. I think the only scene where it would be noticeable during playback is the XFLAG logo at the very start. The difference isn't worth a re-release for, but I'll keep it in mind for future encoding projects because there's no reason not to do it. Of course, if I did postprocessing then I'm sure 10-bit would have a bigger difference. I'll leave that to the anime release experts though, thanks for the advice. @hanabi crf25 is good, but the difference is definitely noticeable. crf23 still achieves some impressive compression and looks great.

@PromareWeekOne then perhaps it was Plex being confused by it? not sure why though, the file size is so small that it would direct stream to the screen. all i noticed was that main dialogue parts were being back channeled to the rear speakers more than the front left/center/right. i assumed it was compression somewhere, i'll try forcing some settings and see what's up.

@PromareWeekOne  You definitely get the most benefit out of it after post-processing, due to the increased precision they work at (at least, in VapourSynth. I'm not sure if ffmpeg filtering does the same?). Interesting to learn how to do, but also very demanding and not worth if most BDs already look "good enough" for you.

@Deikun Plex always transcodes softsubs, it burns them on the fly. Plex apps can only play hardsub. @LightArrowsEXE I believe ffmpeg plugins by default work with the same pixel format as the source. But you can easily adjust the pixel format within the filter graph. If you for some reason wanted to, you could run 6 different filters with 6 different pixel formats and then output to a different one yet. So you could definitely take 8bit input, run filters in 10bit and then encode to 8bit output. But at that point you might as well just encode to 10bit.

@PromareWeekOne I have no problems with softsubs on Plex. It's just the audio mix that is wonky.

@Deikun What I'm saying is that Plex won't direct stream to the screen. Because the softsubs are present, Plex is forced to transcode. (But it's possible that it would only transcode video and stream-copy the audio, I'm not 100% sure)

@PromareWeekOne > Plex apps can only play hardsub. This is not the case for at least a few years now. A lot of modern Plex apps direct-play ssa softsubs just fine cause they're mpv-based, and what's better for playing anime than mpv nowdays? I've personally tested the old PMP and the new Plex app for Windows and Mac, iOS and tvOS. The only things that still can't do it properly now are web player and probably everything Android-based, and specific hardware players/TVs I guess. @Deikun What's your Plex player? And was it specifically direct-streamed, or direct-played? The former always mean remuxing on the go while transforming some tracks as well. Direct-play is always an unchanged direct copy.

Thank you!

@LightArrowsEXE how's the progress on your release? Can you give us a date?