How do the "scene" releasers do it?

[theredbaron]

Hey everyone,
I've been experimenting with various compression/transcoding methods and am flabbergasted at how the scenesters do it! Is there an easy way to get a recording (recorded on High setting), convert to XVid (most popular scene format it seems), have it look sharp and awesome, but be very small on size? The epis I've downloaded in the past are half hour shows at 175MB and hour shows at ~350MB. Are there settings in Gbpvr that can be tweaked to make it an even better/sharper picture, to then compress/transcode to XVid? I've heard of several methods of transcoding like AutoGk, Mencoder, VirtualDub, etc. Which is the easiest to use, while providing the best results (in people's opinion)?

[pcostanza]

I've tried a few of the programs you've mentioned with my tv caps and can never, ever get them to look even close at 4 times the final size. I don't know what they do either but wish I knew.
For me, AutoGK was not good at all. Dr. Divx was ok and virtualdub had some audio sync problems for me. So, I gave up.
Like you, I'm interested in learning how they do it.

[-stattik-]

First, most of the releases start with HDTV captures. If a SDTV capture looks grainy, it's going to look grainy when you transcode it. If you open any of the shows with gspot, it usually tells you what application was used to do the encoding. VirtualDub(Mod), and Nandub are two of them although there are usually a few other applications used to prepare the video for transcoding. The actual settings used to make these shows seem to be closely guarded with each person having their own secret "recipe".

If you're interest in a little more streamlined method you can check out this link:

http://gbpvr.com/pmwiki/pmwiki.php/Utili...anscodeCut

It's a script I put together using mencoder. By default, a half hour show will be about 175MB(with the commercials cut out.) If you need a show to be exactly 175MB and of the highest quality, I would suggest the 2-pass profile. You can set the script in test mode if you just want a 30 second snippet to check out.

Remember, garbage in, garbage out. The cleaner the source file, the better compression and output you'll get.

[jam_zhou]

I agree with stattik. You need to start off with a good source. I've read that you can't compress noise - totally random and no interpolation. So that accounts for some of the file size. The other is the algorithm. For my TV transcodes I prefer speed over quality because I tend to have 4 to 6 hours of shows recorded a day. I choose to have a transcode that runs 2 times faster - meaning a 30 min show transcodes in 15 min. I have also transcoded DVDs that fit the description you are talking about. For those, the algorithms are different and it takes about 4 hrs for a 1.5 hr movie. The time difference shows that there's a lot of extra analysis and compression going on. If I wanted top quality my computer would be continuously transcoding - not good for CPU or HDD.

Anyway, if you want the command line parameters try out http://mewig.sourceforge.net/ which will show you all the parameters. Look for the encode.bat that it produces and use the script that stattik references on a tv show - I also recommend using the test mode (-endpos 120).

I've also heard that gui4ffmpeg will give you the command line parameters as well.

Give it a go and share your findings

[colin]

You need to look at http://avisynth.org/. This is a frame server which allows you to use filters to improve the quality of the video whilst compressing it in size. You can use this with mencoder, tmpgenc, HCencoder, mainconcept, etc. [I hear good things about HCencoder]

There is no real secret but it will take you some time to get to quality encodes. Remember, what looks good to one person, may not look good to another. This is not the forum to discuss this as it is such a huge topic and I would refer you to http://forum.doom9.org/ to learn more about this and video encoding in general.

As -stattik- says, garbage in, garbage out. If you want to so this with you tv captures, then record at a high bitrate.

Cheers,
Colin.

[pBS]

yea,most of those shows come from the dvd so they are excellent quality going in..noise is the biggest factor...it's all detail..and doesn't compress well..
also there's pre-compression...sattelites and digital cable are already compressed, so that doesn't compress very well past a certain point..
[still, sat. and digital cable are the best 'capture' sources to compress from,very little noise, but they have some artifacts from the existing compression]
it's amazing how smal you can make a dvd...then a capture of that same dvd with almost no noise [direct cable] won't compress nearly as well...
same reason you rip a cd digitally, lots of info lost/noise added in the analog conversion process..
i really like the Hank's encoder..[HC Enc] it's not the fastest but produces pretty nice quality/file sizes...but basically the more time you spend on compression, the better/smaller results will be with any compressor..[tho some waste time]

and alot of those 'scene' guys are making custom profiles for every movie...
tweaking and tweaking every scene to get those results...
[the 'magic' is actually a lot of trial and error]
this is the 'holy grail' of the video industry...if there were a simple answer, [how to compress to small file size, in a small amount of time, and still retain quality] tons of people would jump on it..
i know i would!