Transcoder Vs Re-Encoder?

[sublimelax]

This is most definitely a newbie question, but in this thread: https://www.videohelp.com/forum/viewtopic.php?p=766474&highlight=compression#766474 it mentions that there is a such thing as an such thing as a re-encoder and a such thing as a transcoder.

I gathered from the thread (hopefully correctly) that DVD Shrink is a transcoder and that re-encoders take longer, but deliver better quality results than a transcoder.

I am interested in full backups (keeping all menus, extras, etc.) and am worried about the amount of compression sometimes needed in DVD Shrink.

What is a re-encoder? Does it work better than DVD Shrink?

Any info is greatly, greatly appreciated. Thanks.

Edit: fixed a couple mistakes.

[indolikaa]

I always understood transcode means to code from one defined standard to another. Audio, for example, can be transcoded from PCM to AC3, or MP2 to PCM, etc.

Re-encode means to code within a standard but change the variables that control the coding process. When you take an existing MPEG-2 stream and change the bitrate, you would be re-encoding the stream.

DVDShrink would be a re-encoder, since you would be taking an existing data stream and changing its size but not its format.

[sublimelax]

I realize my post may be a little confusing. Haha, that is just because I'm a little confused. I have pulled a few quotes from that other thread that may make things a little more clear.

You don't need superhuman eyes to see the difference between a transcoder and an encoder at even moderate compression. Due to the way transcoders work they create very distinct artifacts. Maybe some people just don't know what to look for, but once you see it, it is very distracting.

Its extremely closed-minded to assume that just because you don't see problems in your backups that they therefore don't exist, and that anyone who does see them must be overly scrutinizing. There is a whole world of people out there who have given transcoders a fair try but still find them lacking for most backup purposes. I use transcoders for certain purposes and I find them very convenient. But when I start to see quality loss, and I know I can avoid it using another method, then I go that route instead.

Honestly, it does not take me anymore work to do a re-encode then a transcode. Either way it involves about 5 mins of prep time and then a waiting period. Sure the re-encoding takes many more hours than the transcode, but it all happens while I sleep anyway so I don't care.

These quotes make me think there is a difference between re-encoding and transcoding. The re-encode appears to take longer, but delivers better results.

Is there a difference, or am I still confused?

[Gazorgan]

I disagree.

TMPGEnc, CCE, and Mainconcept are all encoders/re-encoders. They take any (mostly) video in and output MPEG 1/2 video. There are many many variables they deal with.

DVDShrink and Nero Recode are transcoders. They take in a very specific MPEG2 video stream and output the very same video stream at a lower bitrate. You are using highly defined algorythms to shave bits where they can. They work exceptionally well on CBR encodes (which you will never see in a commercial DVD). Having to hit a specific size in a single pass (2 pass really with deep analysis) means the quality can suffer.

Encoders/Re-encoders are always better in quality than a transcoder. Transcoders are always much faster. The question is always: Is it good enough?

It's like comparing a CBR/CQ/VBR encode via TMPGEnc: Is the extra time worth some finite improvement in video quality? As it turns out CQ is usually the best combination of quality+speed, but it's harder to hit an exact file size.

[sublimelax]

Just to make sure I am understanding you correctly, it is your opinion that using TMPGEnc would yield better results than DVD Shrink?

I really do not care about time constraints. I simply want the highest quality I can get from trimming a DVD-9 down to a DVD-5 and keeping all the menus and extras and everything else. IF TMPGEnc will give me the best quality, I'll use that.

Thank you very much for your help.

[adam]

http://forum.doom9.org/showthread.php?s=&threadid=63587

Here is a good FAQ on the differences between transcoding and re-encoding. Encoding analyzes everything about the source and attempts to store the information in the most efficient way possible, in other words it tries to remove any redundant information it can. A transcoder simply takes whatever source you give it and attempts to remove data to achieve a certain amount of compression.

A transcoding engine can attempt to remove data where it will be missed least, essentially doing its best at removing redundant information. But the process is fundamentally less efficient than encoding since it has so much less information to work with, and at the same compression level encoding will always yield higher quality.

Of course this assumes that at least marginal compression is happening. If you only compressed your source by around 5% or so then its theoretically possible that transcoding could yield superior results since you have the benefit of the quality of the original encode, ie: strategically placed I frames, custom GOP matrix etc... You'd lose this in your re-encode and that possibly might be enough to tip the scales. Otherwise, absolutely something like TMPGenc will achieve better results than something like DVD Shrink. At even moderate compression the difference should be very apparant.

[indolikaa]

...and that's why I never claim to be an expert at anything.

[Gazorgan]

I concure. Small transcodes (under 15% shrinkage) probably doesn't matter what you use. Large compression truly needs TMPGEnc/CCE to get the best quality.

If you want to reaaly make a decision, drop a 6 GB M2V file into TMPGEnc, set you precision high, and do a 2-pass VBR encode. It will take more than the 60 minutes DVDShrink does.......(24, 30, 36 hours for me?)

[sublimelax]

Thanks a lot guys. You have been very informative and helpful.