What's the reason of encoding a visually poor digital anime with crf 13? And even more questions about saving space on the sound track (lossy Opus instead of lossless FLAC) in 2.3 Gb/ep. encode :\

I'm aiming for [transparent quality encodes](https://en.wikipedia.org/wiki/Transparency_\(data_compression\)), fixing source issues/artifacts, and allocating as much bitrate as necessary to achieve that. For video, you need a lot of bitrate to achieve that, especially if the content has grain/noise that is being preserved therefore making the video much "busier". The source BD's video averages at 36 Mbps.. my video is only a third of that in size. I'd argue it's compressed excellently. I don't see the point in doing lossless audio when the source video is encoded by me as well. I just tested, and using FLAC for the audio tracks would average around 1900 Kbps. In episode 1, only 608 Kbps of average bitrate is being allocated to audio. Once again, 3x savings. I'd rather allocate the precious 1.3 Mbps that were saved by the nature of lossy audio compression, on the video. The trained eye can distinguish between bit-starved video and a transparent one. However, to my understanding, no one actually managed to pass an ABX test anywhere near the audio bitrates I'm targeting, making it quite literally a waste of bandwidth & disk space for a difference that isn't distinguishable by a single person.. therefore I don't see a reason to use FLAC (or any lossless codec) for the audio tracks. If you're able to successfully pass an ABX test consisting of: lossless audio vs smol's audio encodes (192 Kbps Opus for 2.0); I will gladly release v2s of my torrents with lossless audio. I just find it odd you're complaining about supposedly bloating the video (I'm really not) yet you're suggesting to bloat the audio instead.

smol

Thanks smol! I do appreciate you encoding a whole series, and not cucking us with just a season.

@smol, > I just find it odd you’re complaining about supposedly bloating the video (I’m really not) yet you’re suggesting to bloat the audio instead. In the vast majority of cases, transparency is achieved at 15-16 crf with good settings, so your release is really bloated. You may claim otherwise and proof with something about noise, but all you're doing is encoding digital noise artificially added during the 720--->1080 upconversion for release on bluray in a slice with a small amount of dynamics. Such extortion... As a fan of old analog anime, I will never understand why fans of modern anime do this. Especially in the present time, when digital noise can be generated over clean video by an Avisynth script during viewing. Perhaps it's time for those who like preserving noise to switch to AV1, which also allows you to overlay it without wasting bitrate? Regarding the sound, I prefer Opus, it's great. And I think that the hearing and equipment of most users will not allow them to feel the difference between lossy opus and lossless Flac (but also I prefer balanced size/quality video, nor transparent). The only thing that surprises me and what is my first comment about -- the funny situation in which a fan of bloated releases saves space on sound (around ~150mb/per. ep.). It is not usual situation because If I understand correctly the thinking of the average “transparency” lover, preserving more materials in their original form is one of his priorities. So your point about “ABX test” or my about “user’s equipment” doesn’t make sense. btw even such "transparency" lovers like Beatrice-Raws made a lighter release, keeping original TrueHD. P.S. Nothing personal. Please don't take my complaints seriously. Current hardware power, diversity of technologies and huge disk spaces allow users to have fun in different ways. So you and I are just on different sides.

av1 tards at it again

NOOO 28 GIGS TOO BIG

@Kagemusha The fact you think CRF is the only factor in transparency says a lot about you. Nearly everything I upload is encoded with CRF 13 up to 13.7 yet many of my encodes are smaller in size compared to other CRF 14-15 encodes. You don't have any clue about video encoding, why even comment? Noise exists basically as dithering on top of the video to prevent banding and blocking. Not an issue with my encoding setup but that's the look they go for. Download AV1 mini encodes without grain synthesis if you prefer the grain-less anesthetic. Or Beatrice-Raws if you prefer useless static grain that isn't true to the source. You claim Beatrice kept the original TrueHD which is absolutely funny considering the source has PCM audio. They encoded from the JPBD, not the USBD. You claim it's transparent and smaller.. they didn't fix the dirty lines, they overlayed static grain on the video, they didn't fix the shimmering in the ED, they don't preserve grain in flashback scenes anywhere near as they should, their OP encode is lacking in fidelity due to how busy it gets and the fact they starved it of its necessary bitrate, they applied a bandaid solution for the artifacts caused by the poor upscaling done in the studio's mastering process... by all means download their release if you prefer it; but your yapping here is unreasonable and you're clearly not the target audience for this release. And no I won't be encoding in AV1 any soon. It's not widely supported enough and grain synthesis doesn't offer the control I want.

@Kagemusha Comments like yours are uniquely grating. You clearly understand *something* about what you are talking about, but all that half-knowledge results in is misguided conclusions. The audio encode in this is much more transparent than the video encode. I can guarantee with 100% certainty that I can see the difference between the filtered input and the final encode in a comparison, and that you cannot do so with the audio encode. This is not about "audio equipment" either—no normal playback solution (HE-1 or otherwise) will make it audible. The point of high quality releases such as this one is to deliver the highest quality possible with the smallest compromises necessary; the question becomes how low you can get the bitrate without **any** noticeable quality degradation. The bitrates chosen for video and audio are completely separate concerns. Any thoughts such as "balance" are nonsensical. The video encode should be transparent and the audio encode should be too, at the smallest size that goal can be achieved. The concern of total size comes up in the decision to bloat the audio by going vbr 192 on opus instead of just using vbr 128. CRF does not work like that. Some sources do not require lower than CRF 18, some require CRF 10. The standard of CRF 15 for everything is from a time when every encode was worse than source, starved in motion, and full of ringing. People then would use wrong encoding techniques because they didn't understand they were doing (essentially every encode before 2020). AV1 is complete dogwater for high fidelity encodes and the grain added on playback has noticeable patterns and little control. You would know all of this if you knew what you were talking about. It is obvious to anyone who is **actually** familiar with these subject matters just how little you know, but people who know a little but not a lot like you give the impression to outsiders that your points are valid. It's not a matter of preference or sides, you are just wrong due to a lack of knowledge.

We use CRF 18, a high-quality bitrate setting that is a standard recommendation on Doom9 (the expert encoding forum). Rest assured that this video will look perfectly fine even when watched closely on a large screen. CRF 18 avoids pointlessly bloating bitrate for virtually no gain, so you get a great balance of quality and size.

 >Okay, I need to figure out how to respond to Kagemusha's comments. Let me start by understanding the context. > >Kagemusha is questioning the encoder's choices, like using CRF 13 Opus instead of FLAC. They argue that the video is bloated and that saving space on audio isn't typical for transparency-focused encoders. Others, like smol and Tsukudakobashi, have replied defending the choices, emphasizing transparency and efficient bitrate allocation. > >The response should address Kagemusha's points without being confrontational. Acknowledge varied priorities. CRF isn’t only factor—source grain affects bitrate. Opus audio saves space for video. Needs differ: bloat vs necessity. Both value quality, prioritize differently. Community evolves, diverse approaches coexist. A response to Kagemusha's comments should address the technical and philosophical points raised while maintaining objectivity and clarity. Here’s a structured approach: 1. Acknowledge Differing Priorities: Recognize that encoding practices often reflect differing goals. Some prioritize minimizing file size while retaining perceptual transparency, while others emphasize archival fidelity. 2. CRF and Source Complexity: Clarify that CRF values alone do not dictate output size or quality, as bitrate allocation depends heavily on the source’s complexity (e.g., grain, noise, motion). A higher bitrate may be necessary to preserve intentional dithering or grain structure, even in digital anime, to avoid artifacts like banding or blurring. 3. Audio Transparency vs. Bitrate Efficiency: Note that audio transparency at lower bitrates (e.g., Opus 192 Kbps) is well-documented, and reallocating saved bandwidth to video can improve visual fidelity. Lossless audio might be prioritized in archival contexts but is less critical for playback-focused encodes where perceptual transparency is achieved.

4. Grain Synthesis and Codec Choice: Address the suggestion to use AV1 for grain synthesis by pointing out its current limitations in control, compatibility, and pattern consistency compared to retaining source grain via efficient encoding. Mention that grain synthesis tools are still evolving and may not yet match the artistic intent of the original source. 5. Comparisons to Other Releases: Counter the Beatrice-Raws example by distinguishing between encoding philosophies—some groups prioritize minimal filtering to adhere to source material, while others apply fixes for mastering flaws. Neither approach is universally "correct," as they cater to different audiences. 6. Technical Misconceptions: Highlight that blanket statements about CRF targets (e.g., "CRF 15 is sufficient") overlook source-specific variables. Similarly, conflating audio and video bitrate allocation misunderstands how perceptual transparency operates across sensory modalities. 7. Community Standards and Evolution: Conclude by emphasizing that encoding practices evolve with technology and expertise. While debates about methods are natural, they should focus on measurable outcomes (e.g., ABX tests, visual comparisons) rather than subjective preferences. A balanced response would maintain respect for diverse perspectives while firmly grounding arguments in technical principles and empirical evidence.

@LastBreeze, thank you. Artificial intelligence once again proved to be smarter than these leather bags.

@smol, well, I think I recommended not to take the conversation too seriously:) > The fact you think CRF is the only factor in transparency says a lot about you. Classic. If I wanted to hurt you, I would answer you the same way. HEVC has **several dozen settings** for more efficient bitrate redistribution. But you prefer to simply increase the overall bitrate -- this is the easy way to "archival quality". I saw your settings. There is a standard set of settings for quality-oriented encodes, in which you only tweaked crf, qcomp and deblock. That's why I am talking about crf only. There is no point in talking about the rest. Mentioning crf was an illustration of your approach to encoding. > You claim Beatrice kept the original TrueHD which is absolutely funny considering the source has PCM audio. A and another attempt to bite me:( Do you think I checked what sound was on Blu-ray for the sake of an exact answer? But I hope you got my idea. Or didn't? It seems my English is bad enough that I'm not understood. Then I will repeat it in more detail and directly: PCM-->FLAC allows you to reduce the size while remaining lossless -- this is the main reason why archivers use FLAC. DTS or TrueHD will not change the size, so they can be left as is. > You claim it’s(Beatrice encode) transparent and smaller… Not really. I called them "transparency lovers." All of you are trying to preserve details in titles made by Chinese animators on outsourcing for a cup of rice. So much effort for nothing. There are no details in digital anime that you want to preserve, cause they weren't drawn. There you will find only rendered backgrounds and effects from Photoshop After Effects masters. Just look at these: https://slow.pics/c/sub9sBKX @Tsukudakobashi > Any thoughts such as **“balance” are nonsensical**. The video encode **should be** transparent > **It’s not a matter of preference or sides**, you are just wrong due to a lack of knowledge. Is this some kind of subspecies of digital fascism?lol

And 2 more times: > You don’t have any clue about video encoding, why even comment? > all means download their release if you prefer it; but your yapping here These 2D girl lovers are so sensitive. Really. ___ Once again. I have nothing against "archival quality" (DeepSeek suggested a useful phrase). But if you are an archivist, then you prefer to leave the maximum amount of original material untouched and for you the size does not matter. The way Mr. smol works with videos shows that he treats videos with trepidation, he tries to correct the errors of the original source without worsening the overall quality. But at the same time he re-encodes the audio in order to save space. From my point of view, these are mutually exclusive preferences, characteristic of opposite audiences. This makes me smile. In my first comment, I share my smile. That's all. Those who care about saving space, look at @LastBreeze encodes. Those who collect archives still encode 2pass and Flac 24bit, preserving cover scans and a full set of additional materials, like in u2 internal releases.

Can y'all stop yapping

@Kagemusha Seeing how you didn't respond to anything I said with proper arguments despite your yapping I'll take it as you implicitly conceding. And it's obvious to anyone reading your comments in sequence that you are simply turning to saying different things than you did before because you realise your previous stance was indefensible. Moving on. > There are no details in digital anime that you want to preserve, cause they weren’t drawn. There you will find only rendered backgrounds and effects from Photoshop After Effects masters. What exactly are you suggesting here? Those elements, whether you find their inclusion in the production artistically worthwhile or not, require a lot of bitrate not just to preserve but to not have them start becoming noticeably distorted, more so than standard "drawn" anime backgrounds and effects usually would. You can trivially see this by comparing to CR or VCB in the comparison. > From my point of view, these are mutually exclusive preferences, characteristic of opposite audiences. There are two kinds of archival: archival of the highest quality version available (this encode), and archival of source files (BDMVs). The former is meant for watching and rewatching in the highest quality, the latter is meant for preservation. The first kind doesn't care about loss that is not noticeable (if that were the goal, the filtered video would be losslessly compressed too). Thus leading to the use of transparent quality lossy audio encoding, to save space where possible while not undermining the primary goal of transparency. > Those who collect archives still encode 2pass and Flac 24bit, preserving cover scans and a full set of additional materials, like in u2 internal releases. Those encodes are all dogshit with noticeable distortion especially for anything with film grain. They aren't good for watching, because they are both bloated and starved, and they aren't good for preservation because they are not the source files. Please just archive source files.