Hacker News new | past | comments | ask | show | jobs | submit login

Ever wondered about the toolset of pirate movie groups?

Typically it involves demuxing the source with eac3to, decoding and indexing the video stream for avisynth (often with ffms2 for Blurays), filtering (such as resizing with a spline kernel, debanding, fixing dirty lines, fixing double range compression, correcting aliasing, removing edge enhancement artifacts, deinterlace/ivtc for DVDs etc.), followed by piping it out through avs2yuv to the x264 encoder. x264 parameters are often hand-tuned over a series of short test encodes at a fixed bitrate. Audio streams are left as is, stripped to the core (DTS-HD MA -> DTS) or transcoded to AC3 (Sony Soundforge), DTS (DTS-HD Master Audio Suite), AAC (qaac/fdkaac) or FLAC. Bitmap subtitles are ripped to text .srt files with subextractor/suprip/SubtitleEdit/ and spell-checked. Finally everything is muxed back together with mkvmerge.




Some of these groups really mess up the sound volume of their releases in the conversion process... you have to set the volume to 150% to get a decent volume level and not all programs allow you to do that (and they probably lost more sound quality then they needed to somewhere in the process).


The encodes are made to replicate the source as faithfully as possible, generally that means not touching the audio and disabling all forms of dynamic gain (which both affects the quality, and reduces the quietness of the quiet moments and reduces the loudness of the loud moments).

I am quite certain that no audio quality is lost, as the explicit goal is to keep the audio at maximum fidelity.

Not sure what setup you are using, but most cinematic players have a way to do auto-dynamic gain. Or, if you do want a more faithful overall experience, all you really need is a better amp.


Probably because in theatre the audio often has a big dynamic range, and the groups are aiming for a "cinematic" result. (There might be only one or two loud points in the film)


I would rather have the various tools, command lines, and settings used than the pirated content. Insert obligatory "why not both?" here but seems like the secret sauce is closely-guarded IP (is this irony?).


It's not really that much about secrecy. Most of the tools are free or open source (with the exception of audio tools which can be for the most part substituted with eac3to).

The problem is that the "body of knowledge" is rather extensive if you want to cover all the corner cases and imaginative ways in which authoring houses screw things up. To be able to filter the video correctly you need to know a bit how the h264 codec works, learn your color formats, color spaces, a little signal processing, a few tables from ITU-R recommendations and some proficiency in RPN wouldn't hurt. As such it is difficult to put it all down in one guide. To make it worse, as with every niche community there are disagreements and transient "fashions". At the end of the day though this is very much chasing rapidly diminishing returns.


Just trying to point out that a great way to learn more would be to review exactly how each release was put together. Even without the "why" this would be a useful learning opportunity.


Very interesting insights, thank you!




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: