Just like [2ndfire's test encode](https://nyaa.si/view/1769249), this is an "AV2" encode where the bitstream codec ID was changed to AV2 **manually**, because AVM, the development ground for new aom codecs and thus AV2, still uses the AV1 codec ID to this day cause the specs aren't finalized. So I wouldn't actually call this legit AV2 because it might not end up being spec compliant in the end, meaning it may never be possible to play the file in a normal media player... I still wanted to make a *better* test encode than what he did, one that better *represents* the *actual* state of AVM. No media player can decode that file natively. You have to use AVM's aomdec reference decoder for that. *** If you want to playback the video: - Build the [AVM master branch](https://gitlab.com/AOMediaCodec/avm) - Change back the “AV2” codec ID to “AV1” with: ``` mkvpropedit "[Trix] Babylon - OP (Creditless) [BDRip 1080p AV2 Opus].mkv" --edit track:1 --set codec-id="V_AV1" ``` - Extract the .ivf file with: ``` mkvextract "[Trix] Babylon - OP (Creditless) [BDRip 1080p AV2 Opus].mkv" tracks 0:av2.ivf ``` - Then decode it: ``` ./aomdec "av2.ivf" -o - | mpv - ``` *** Encoding parameters (This was encoded using the [AVM master branch [be89819b]](https://gitlab.com/AOMediaCodec/avm/-/tree/be89819ba706ada289a613d1d6b782e7cbb1658f)): ``` vspipe "00006.vpy" -y - | ./aomenc --threads=12 --cpu-used=0 --end-usage=q --qp=144 --bit-depth=10 -o "00006_avm.ivf" - ``` My reasoning for using stock settings is that we have no idea (and nobody is gonna bother testing for now) what are the "best AV2 settings", so it's better to leave everything default, as the devs intended it, in order to have all "AV2" tools enabled. Knowledge from AV1 may not necessarily translate 1:1 to AV2 knowledge. This is why using CPU-0 is just as important: without it, we are not tapping into the entire AV2 potential because tools will inevitably be disabled or less effective at higher presets. It's also more appropriate to use 10bits since that's what everyone will be using in the future. We also know for sure it gives an efficiency boost and improves gradiants. QP144 gave a reasonable bitrate of 890kbps, close to where AV1 presents its best efficiency in its efficiency curve for most content. But it's also easier to share and make it accessible. Dunno what caused such blocking in some places though? Anyways, be sure that I will be making an *actual* AV2 release as soon as the specs are finalized and we don't need to hack our way to make an "AV2-like" encode anymore. VVC better be ready kek *** > Further guidance and help can be sought in the [AV1 Weeb edition](https://discord.gg/83dRFDFDp7) discord server