Ok /a/ so let me get this straight:
Anime is predominatly produced in 720p, and after ~2000 it is also predominantly made digitally, so most 1080p files are really upscales of the 720p source.
BUT BDs are always 1080p, so when you get a 1080p BD rip it is "genuine" in the sense that whoever ripped it presumably did it properly, and they didn't upscale it themselves. However it doesn't mean that the source video is 1080p, since it could be an upscale of the raw 720p source. But that would mean that this upscaling would come from the raw, lossless video source, right?
TL;DR: Bottomline my question is: is 1080p BD worth it? In a purely theoretical, video encoding sense.
Also this is probably a Daiz summoning thread. Pic related, looks like shit, the 720p was probably the correct choice.
Wait, this is just making me more confused. So is it actually true that all BDs are natively 1080p? Like a BD player will only playback 1080p and up/downscale it depending on the monitor?
Anon use your brain a bit.
If the anime was made at 720p or sometimes slightly above that(but still below 1080p), why would it magically become native 1080p when it is put on BD? The answer is that it doesn't unless it was mastered in that format to begin with.
No no I read that when they're made into BDs, they're put in at 1080p. As in whoever is making and publishing these BDs is doing the upscaling themselves, in house. That's what I was told. And that's apparently true for all BD disks, they're all "natively" 1080p. But now that's starting to sound dumb and untrue because even wikipedia says that BDs are "up to" 1080p. Of course that also doesn't make, shouldn't it be by file size?
Look at the OP and you tell me. Can you really not tell the difference?
And FYI OP Haruhi was produced in SD, so that's not a 720p upscale, it's a 480p upscale. You should have gone 480p.
Go read through the archives and some of the posts Daiz has said upon the subject, but as you probably know CR and funimation run 1080p streams and these are representative of the native scale you will get on BD for the most part. Most shows are at 720p or sometimes maybe 900p. The only real exceptions you get are movies and the rare show like Nyarko.
>but as you probably know CR and funimation run 1080p streams and these are representative of the native scale you will get on BD for the most part
Yes I know that, but I'm more curious about the files on the BD. Technically they're made from the raw, uncompressed, pictures that make up the video, right? Like however big a lossless video is that's where they're from. So when THEY get put on a BD in 1080p, isn't that better than what CD or funimation does?
Well if you don't mind the file size if you are downloading the BD rips then there is no real problem with getting 1080p all the time, just you won't get much extra benefit out of it other than for movies and a few shows.
Anibin is a good reference. Uses fast Fourier transform analysis to determine native resolution of shows.
Some basic moon required to navigate the site however.
This is where I got that from. It Says JC Staff and Xebec are also starting to do shows in 1080p native.
>BUT BDs are always 1080p, so when you get a 1080p BD rip it is "genuine" in the sense that whoever ripped it presumably did it properly, and they didn't upscale it themselves.
It's better to inverse upscale the footage and upscale it back to 1080p with nnedi at any case. Why?
Because Bilinear and Bicubic are scaling algorithms that BD manufacturers use to upscaling footage to 1080p and they're pretty bad compared to nnedi.
>However it doesn't mean that the source video is 1080p, since it could be an upscale of the raw 720p source. But that would mean that this upscaling would come from the raw, lossless video source, right?
>Bottomline my question is: is 1080p BD worth it? In a purely theoretical, video encoding sense.
Yes, if the show you want to watch is produced at a resolution greater than 720p.
>Would a 720p downscale of a 1080p BD look worse?
Yes but only if the footage was produced at a resolution greater than 720p because you'll lose details.
>Even if the raw source is 720p you're adding just another step of encoding where quality can get lost.
If you're not talking about lossless encoding then it's no use to talk about quality loss because what you want is lossy encoding and not using 100go for a 12-episodes series.
Anime this season with a resolution greater than 720p:
https://archive.moe/a/thread/124080298/#124081146 + Pikaia (1440x810) + Anime Mirai (Kumi to Tulip is 1600x900, Ongaku Shoujo is 1552x873, Happy ComeCome is 1488x837, Aki no Kanade is 1632x918)
Just think of it as complex maths works it all out.
I don't know video encoding but I do know fourier transforms. What he says is definitely possible.
When the digital signal is sent it contains information about what pixels should do what. If you know what you're doing it should be trivial to take that information and extrapolate how many pixels the data is capable of rendering.
>It's better to inverse upscale the footage and upscale it back to 1080p with nnedi at any case. Why?
>Because Bilinear and Bicubic are scaling algorithms that BD manufacturers use to upscaling footage to 1080p and they're pretty bad compared to nnedi.
You lost me
>Bottomline my question is: is 1080p BD worth it? In a purely theoretical, video encoding sense.
>Yes, if the show you want to watch is produced at a resolution greater than 720p.
Ok so then let's take anime made between 2000 and 2010 as an example. This would be mostly 480p and 720p, correct? So you are saying that the 1080p rip is not worth it.
And in the image in the OP, that's really the 480p upscaled and I was an idiot for getting the 1080p, yes?
Yes well beyond the fancy encoding theory and math behind it, I mean I don't know how to rectify it with your second statement. So if the BD itself is 1080p, regardless of the resolution of the source, the 1080p rip will be "better" than the 720p rip of the same BD? And it is "better" because it comes STRAIGHT from the BD, with no funny business from the encoder.
This is my current understanding but it's based on the assumption that all BDs are made at 1080p, regardless of the source.
>You lost me
Basically fansubbers can perform better 1080p upscale than official guys.
>Ok so then let's take anime made between 2000 and 2010 as an example. This would be mostly 480p and 720p, correct? So you are saying that the 1080p rip is not worth it.
>And in the image in the OP, that's really the 480p upscaled and I was an idiot for getting the 1080p, yes?
And yes, OP pic is SD-tier. Pretty sure it's upscaled from 480p.
Episode part is 955p and OP/ED are 1080p.
>Basically fansubbers can perform better 1080p upscale than official guys.
Honestly I don't even question this because anime fansubbing is like the fucking cutting edge of video encoding. You've got Daiz on /a/ and then haasn on /g/ who does mpv. Anime drives innovation.
Good to know though, I'll download 720p. I have a 1440p monitor so I really should have done that in the first place for dat perfect integer upscaling.
As many already said, it depends on the show. And it’s not just limited to BDs, in many cases you should already get 1080p HS. Simulcasters don’t really do anything to limit the resolution. You get pretty good quality, almost always better than TV rips.
Different scaling algorithms. Nearest neighbour is quick but gives pretty horrible quality, while nnedi3 gives fairly good quality but is resource intensive.
He already did for the BDs of the first season. 10bit and everything making the comments on it hilarious.
Nearest neighbor while going from 720 to 1440 will make every source pixel into 2x2 square. While it seems to sound nice, that will just produce jaggies.
nnedi3 is an advanced algorithm that can only scale x2, but has great quality.
>making the comments on it hilarious.
Holy kek, this IS hilarious.
>THIS ENCODE IS CRAP (N)
>WHEN I PLAY THIS ON VLC THERE ARE BLOCKS ALL OVER THE PLACE!!!11!!!
>NOT WORTH 44,93 GB
>AVI > MP4 >>>>>>>>>>>>>>> MKV
>MKV IS SHIT
>Coalgirls recommends it
Coalgirls recommends mpv now. They say if you just want a dumb player that werks get VLC, but if you want good playback and you're willing to spend 5 minutes configuring mpv it's worth it.
I wonder how many threads we'll have before people understand that native 1080p is extremely rare and BDs are just upscales.
Actually Anibin calculates KyoAni shows as 955.5p because of filters that blur the lines to make it blend better, but they're actually produced in 1080p.
read the top right corner
Ok yeah you didn't understand my question. When the BD is made, it's made from THE source. It's made from the thousands of pictures used to create the video, and encoded into the BD. So, my question was, since it is an upscale of the original lossless video file, will the 1080p upscale be "better" because it was done by professionals.
But >>124844629 effectively answered my question. The answer is no, not because the 1080p is an upscale, the answer is no because the people who made the BD are shit at encoding, so downscaling it to 720p is actually better quality.
I still feel like it was a legitimate question that deserved a better answer than "it's upscaled so bad".
Nigga just download the 1080p version.
Worst case scenario you lose a couple gigabytes.
Best case scenario, you get to watch everything in the highest quality available.
These are the same retards that think there's no Season 3 for Haruhi, because of Aya. There's no reason to take the risk and trust they know what they're talking about at all.
I don't use after effects but I'm fairly certain the 1920x1080 in the upper right hand corner is the resolution of the video file. I have no idea what show that is, KnK? Chuuni? It could still be the OP/ED, but it's in 1080p.
It's the software issue.
Retas works best when scan the animation paper at 144 dpi which give you 950p.
Scanning paper in higher dip than than will produce artifacts like random dots, disconnected line, wrong line color when traced with Retas which takes a lot of time to fix.
>I don't use after effects but I'm fairly certain the 1920x1080 in the upper right hand corner is the resolution of the video file
Well I use After Effects and I know it's not the footage resolution. You get the 1080p output when render but the animation footage you work with could be something that's not 1080p.
Example of the said artifacts.
Could also blame it on the bad inbetweener but you get the point.
Different studios have their own standards when it comes to the actual scan & render resolution of their animation.
Most studio out there still render at 720p.
Usually kyoani is scan/rendered 900p and 1080P for their OPs.
J.C staff at 720p to 1080p.
Frames like this led me believe just OPs/EDs are full 1080p. Nowhere else in KyoAni shows have I seen lines as sharp as Tamako's nose here. K-ON! movie also had this sharpness.
Tried to write this post in the comments section of that blog some days ago, but ultimatemegax blocked it (1/2):
>Sorry, but isn't it quite precipitated to rule RURIN work as erroneous here using no true evidence about his sampling methods? And no evidence beyond your possible explanation here?
>Because an example of only a mere asset being worked on (of others that could be composited into an image, and if non-vectors worked on at variable resolutions too) showcasing image pixel density info and during just one step of the workflow, plus your assumption that added image "filters" somehow alter heavily the spatial frequency data of an image, isn't enough to dismiss in this manner a fairly decent approximative solution.
Tried to write this post in the comments section of that blog some days ago, but ultimatemegax blocked it (2/2):
>We have to take into account that; a) Kyoto Animation and associated photography and video editing production companies use different solutions during said workflow (in the case of KyoAni, apart from After Effects we have seen them patently demonstrate an use for RETAS!PRO suite utilities like PaintMan, or Autodesk Maya 3D) that could affect their final output, and b) that the 2-D Fourier transform and the resulting spectrum (understandable approach to it here; http://en.wikibooks.org/wiki/Basic_Physics_of_Nuclear_Medicine/Computers_in_Nuclear_Medicine#The_Fourier_Transform_-_A_Pictorial_Essay) used on RURIN analysis should be able to denote picture artifacts, such as layered effects with a known fixed radius, and better yet as long as multi-sampling is used (continuous line-art isn't fundamental, but per-pixel brightness information).
>And even then I think this article doesn't ultimately demonstrate either why there is a discrepancy of image resolution analysis, that RURIN notes, on some of their series between their OP/ED and the rest of the episodes' footage.
>All in all, we can't be sure of the exact situation unless we understand clearly how the process is done (we'd have to personally ask, of course!), and have to be aware of how this kind of speculation doesn't help in creating a truly well informed report that securely refutes the exhibited arguments.
All you need to know about that person, spreading bullshit as if his fanboy word is wisdom, and not even letting others contest a thing.