Why does so many think the less ways something can play, the better it is? I'm sure most are repeating what they see. That much is obvious because anyone that knows better wont encode this way. They stick with what allows files to play on as many different types of media players as possible. (AVC8bit AAC/AC3) It also makes sure their encodes are shared longer than what can only play on computer assisted media players. The more ways something can be played the higher it's value and the easier it is to decide what type is better to have. If I want a better picture increasing the frame rates is all it takes without sacrificing play ability.
Thanks for sharing. Some changes need to made like re-encoding back to avc8bit then can be put on a thumb drive and played on a smart TV.
So, it's apparent you're brand new to the encoding scene. Welcome!
Fun fact: There are other encoders other than me? Want AV1/OPUS/10bit? Here it is!
Want some other combination? Use the search function and remux the types of tracks you want (mkvtoolnix-gui)!
Any reason to not use the recommended bitrate of 128Kbps (https://wiki.xiph.org/index.php?title=Opus_Recommended_Settings&mobileaction=toggle_view_desktop#Recommended_Bitrates) for encoding stereo audio using Opus? They say “Opus at 128 KB/s (VBR) is pretty much transparent”. Is there a massive space gain in going from 128 to 96Kbps?
Subjective listening tests, and the fact that audio doesn't quite work like that.
If it was a song with a bunch of instruments going in the foreground, normally you'd compare high entropy noises like cymbals and certain electronic synth noises during the course of a song to determine relative performance of codecs.
None of that applies in most anime, as it's either someone talking, or some faint BGM.
You can try comparing libopus 96kbit to 128kbit for an anime audio track.
This is, however, somewhat specific to anime that just have "filler" BGM, like LWA.
If there is intense FOREGROUND music, like AKIRA, Cowboy Bebop, FLCL, Eureka 7, etc. Then if I'm using my SR-009s, I can hear a tiny difference.
This brings me to my main point. No one can tell between 96 and 128kbit for LWA.
The people who CAN tell (assuming there's certain types of foreground music, which are NOT present in LWA), are relegated to Audiophile groups with expensive equipment and trained ears. And they'd have to be both listening, and A/B testing between a 96kbit opus track and a 128kbit opus track.
TL;DR There's no perceptible difference and no one can or will notice a difference. Observe: http://listening-test.coresv.net/results.htm
@2ndfire I got your point. My line thinking was that if in case some audio segment in Little Witch Academia (or any show/movie for that matter) required some additional bits it could benefit from the 128KBps upper limit and since it's VBR the size is optimized anyway. Which is why I was curious if the there were significant space savings in encoding at 96Kbps as opposed to 128 Kbps. Also it would be useful if you could provide some screenshot comparisons with the BD source and see how AV1 fares at such low sizes
The best encodes in my personal opinion are those that have the minimum size but at the same time are (visually or audibly) transparent to the source according to human senses assuming he/she has a 4k TV and a 7.1 surround sound system. Basically archival level stuff. I'm not complaining as I understand that this may not be feasible in every scenario and that not everyone puts in the effort/gives a shit about these things but it would be nice if most of the encoders in the anime scene followed a few quality standards to get the best out of their encodes especially with the new wave of upcoming AV1 encodes now that it only takes 2x slower to encode compared to x265 (https://streaminglearningcenter.com/blogs/av1-now-only-2x-slower-than-x265.html)
@2ndfire also what's your opinion on AC3 to Opus encoding? I mean the space saving is great since it can reduce a 100mb audio file to 10mb. But does this lossy to lossy format transcoding introduce many artifacts or is AC3 a sufficiently high quality source to encode from and is the loss negligible? I couldn't really find relevant test after searching on Google for this specific case
I'm curious as to why there's no ML/AI based algorithm which can automatically tune the parameters (general ones like CRF or bitrate) of the encoder after picking a sample of frames from the audio/video yet
> The best encodes in my personal opinion are those that have the minimum size but at the same time are (visually or audibly) transparent to the source according to human senses assuming he/she has a 4k TV and a 7.1 surround sound system.
This is transparent to that setup, as stated earlier. Also speaker systems are WAAAAAAAY worse for trying to analyze audio than high end headphones.
If someone with your hypothetical setup tried to A/B it they would fail.
> also what’s your opinion on AC3 to Opus encoding? I mean the space saving is great since it can reduce a 100mb audio file to 10mb. But does this lossy to lossy format transcoding introduce many artifacts or is AC3 a sufficiently high quality source to encode from and is the loss negligible? I couldn’t really find relevant test after searching on Google for this specific case
AC3 can have several different types of bitrates. It's the same as any other encoding, it's just not as good a quality source as DTS-HD or PCM.
> I’m curious as to why there’s no ML/AI based algorithm which can automatically tune the parameters (general ones like CRF or bitrate) of the encoder after picking a sample of frames from the audio/video yet
You can figure it out from tables/graphs, generally. Also people don't tend to aim for a VMAF... because it's an extra step. If you find a CRF that works for anime, it will generally look good for all of them, though it can do weird things with sizes.
@2nfire FYI in case it can be of utility
By disabling CDEF filter, 10bit decoding speed increases by 2.3 times. Disabling CDEF benefits animation and damage film encoding
https://www.reddit.com/r/AV1/comments/i3zv7c/by_disabling_cdef_filter_10bit_decoding_speed/
Woah i compared a crf15 x264 encode to av1 and i see tiny change . From 2gb to 200mb thats awesome ! We need more release on this codec for great oldies.
Ps my i7 2600k is probably dying.
Comments - 12
JustSayNo2Funi
2ndfire (uploader)
sethazazel
Rohanroms
2ndfire (uploader)
Rohanroms
Rohanroms
Rohanroms
2ndfire (uploader)
hourwatch
2ndfire (uploader)
Makoto_shin1