## General Information: This release of Elfen lied includes all episodes and multiple extras. Video tracks were encoded in AV1, Audio in FLAC and subtitles are ASS (zlib). **Source:** Video + Audio + Subtitles -> JP BDMV upload by anon; JPBD Remux by [izu]. Subtitles for NCED and NCOP were taken from [Lulu]. Subtitles are [Doki] and [Doki modified], we have not made changes to these. Manga is from [1r0n]. **Audio details:** Episodes include Japanese 2.0ch and Japanese 5.1ch; English 5.1ch tracks are in a seperate folder using the .mka container. ## Encoding and Filtering specifics: > **For episode:** ***Filtering*** * `rektlvls` + `edgefixer.Continuity`: Trying to fix the borders * Very strong `vsTAAmbk.TAAmbk`: Trying to fix the terrible aliasing in the source. * `vsmasktools.FreyChen` & `vsscale.descale_error_mask`: Trying to protect background texture and other details during AA. * `vsdenoise.DFTTest` denoising before AA and adding the noise back after AA: Trying to protect the noisy texture * `vsdehalo.dehalo_alpha` & `vsdehalo.fine_dehalo.mask`: Trying to remove halo. * `vsdenoise.mc_degrain` + `vsdenoise.bm3d`: Remove dynamic grain to prepare for `--psy-rd` based AV1 encode. * `vsrgtools.limit_filter`: Trying to protect dynamic details like rain in temporal denoise. Some scenes have significant banding. In additional to everything above, * Very strong `vsdeband.f3k_deband`: Trying to deband. * `vsrgtools.limit_filter`: Trying to protect details. Some scenes have really terrible banding, but luckily without any details. Before `vsdeband.f3k_deband`, * `msmoosh.MSmooth`: Smooth everything. * `vsdeband.deband_detail_mask`: Trying to protect weaker linearts. ***Encoding*** * Using 5fish/SVT-AV1-PSY v2.3.0-C `--tune 3` encoding at `--preset -1`. * Using a `--psy-rd` based approach to preserve all the grain and details in the show. * Using `--psy-rd 2.0 --spy-rd 1 --qindex-offsets [-16,-12,-12,-10,0,0]` in the most difficult scenes such as scenes with rain. * Using `--psy-rd 2.0 --spy-rd 1` in slightly milder scenes that are still boosted to very good `--crf`. * Using `--psy-rd 2.0 --spy-rd 2` in easier to encode scenes. * Using `--hbd-mds 1 --max-32-tx-size 1 --luminance-qp-bias 10 --qp-scale-compress-strength 3 --variance-boost-strength 3 --variance-octile 3 --enable-alt-curve 1 --qm-min 9 --chroma-qm-min 12 --noise-norm-strength 3` to keep as many details as possible. * Using `--key-frame-chroma-qindex-offset -20 --chroma-qindex-offsets [-8,-5,-5,-5,-5,-5]` to try to preserve more grain in flat areas. * Using Progression Boost Preset-Character-Boost-Butteraugli-Max to manage bitrate allocation. * Using av1an `--photon-noise [5-8]` to generate a layer of dynamic grain after encoding depending on the luminance of the scene. > **For OP:** ***Filtering*** * `awsmfunc.bbmod`: Trying to fix the borders. * Very strong `vsTAAmbk.TAAmbk`: Trying to fix the terrible aliasing in the source. * `vsmasktools.FreyChen` & `vsscale.descale_error_mask`: Trying to protect background texture and other details during AA. * `vsdenoise.DFTTest` denoising before AA and adding the noise back after AA: Trying to protect the noisy texture > **For ED:** ***Filtering*** * `awsmfunc.bbmod`: Trying to fix the borders. * `vsdehalo.dehalo_alpha` & `vsdehalo.fine_dehalo.mask`: Trying to remove halo. > **For OP and ED:** ***Encoding*** * Using 5fish/SVT-AV1-PSY v2.3.0-C `--tune 3` encoding at `--preset -1`. * Using `--film-grain 13` to synthesize the original film grain. * Using `--hbd-mds 1 --max-32-tx-size 1 --luminance-qp-bias 10 --qp-scale-compress-strength 3 --variance-boost-strength 3 --variance-octile 3 --key-frame-chroma-qindex-offset -6 --chroma-qindex-offsets [-4,-2,-2,-2,-2,-2] --qm-min 9 --chroma-qm-min 12 --noise-norm-strength 3 --psy-rd 2.7 --spy-rd 2` and a fixed `--crf 17.50`. ## Note: This release was made possible by the source providers and the VapourSynth contributors whose tools it depends on. We are grateful to the helpful members of this community who have taught us, and continue to teach us, the art of encoding and filtering. Hopefully this release will be to your liking! Feel free to join us here: https://discord.gg/83dRFDFDp7