Blind Test - DVD-RB vs other one clicks

[Fishman0919]

I've been working on this for about 2 1/2 weeks and I've finished the first half of this two part test. The first half of this test was to compare DVD-RB .74 w/ CCE Basic to DVDShink 3.2 and InstantCopy 8, all 3 newest versions at the time.
I started by ripping 10 movies to the HD with DVD Decrypter 3.5.2.0. Then on 3 identical machines, one with each program, started to backup all 10 movies, removing nothing... keeping all audio tracks and subtitles. After all the movies were finished I cut out a 10 - 20 sec clip from each movies, done from each program. Then I created a dvd with all the clips from each movies from each program.

Settings:

DVD-RB .74 - One Click, VBR Bias=0, Quality Prec=16, rest default.
DVDShrink 3.2 - Default, with deep analysis and AEC Sharp(default).
Instantcopy 8 - Copy Method=Auto, Reduction Method=HQ, rest default.

Movies:

The Matrix - 1:42:07 - 1:42:19
Matrix Revolutions - 1:41:48 - 1:42:08
DareDevil - 1:29:16 - 1:29:32
LOTR - Two Towers - 1:35 - 1:55
Saving PVT. Ryan - 4:40 - 4:52
Gladiator - 46:00 - 46:15
Star Wars - Phantom Menace - 2:01:14 - 2:01:30
Titan AE - 5:52 - 6:11
Shreh 2 - 2:58 - 3:17
Jurassic Park Collector's Edition - 5:20 - 5:30

Viewing Equipment:

TV - Pioneer PRO1410HD Elite 61" Plasma TV
DVD - Sony DVPS9000ES DVD Player
Cables - Monster Cable M1000CV - 1m

Test:

I work for a Mid to Highend Audio/Video store and talk to people all day about DVD's, DVD equipment, DVD coping software.... So I got the idea of making a DVD with clips from different movies done from different software and show them to people and ask them to pick the best one. All the people I ask had no idea what software was used (or really what was going on), the only thing they were asked/told was to choose which one looks the best to them. This took place over 2 weeks and 37 people took the test. First I showed the clips from one movie in order (Shrink, DVD-RB, IC 8) and pause and ask them to pick the best one to them or if they wanted to see the clips again. Then I repeated for each movie.

Results:

The Matrix - DVD-RB 27, DVDShrink 7, IC8 3
Matrix Revolution DVD-RB 21, IC8 8, DVDShrink 8
DareDevil - DVD-RB 17, DVDShrink 12, IC8 8
LOTR 2 Towers - DVD-RB 28, IC8 5, DVDShrink 4
Saving PVT Ryan - DVD-RB 16, IC8 12, DVDShrink 9
Gladiator - DVD-RB 22. DVDShrink 9, IC8 6
Star Wars PM - DVD-RB 31, IC8 4, DVDShrink 2
Titan AE - DVD-RB 14, IC8 13, DVDShrink 10
Shrek 2 - DVD-RB 20, IC8 12, DVDShrink 5
Jurassic Park - DVD-RB 23, DVDShrink 8, IC8 6

My Thoughts:

Not what I expected, kinda at a lose of words, but clearly DVD-RB w/ CCE Basic looked better then both DVDShrink and IC8 to the 37 people I showed them to.

[NaN]

Wow. I'm impressed. Thanks for the nice compare!

Cheers, NaN

[richarddd]

Great way to test. Actually watching the movies is a much better testing methodology than the various statistal analysis testing that is so often posted.

[artoor]

Bravo! In my opinion it is very good test. First of all completely objective :) I always knew that DVD-RB + CCE is the best way to do backup but You (and another people) proved that one more time. Thank You for Your good work :)

Regards

[jdobbs]

This is what I love to see -- real tests using people who don't have a dog in the fight. I also think it's interesting that IC8 comes in second also. Before I worked on DVD-RB I did all my backups with IC7 and IC8 (when I didn't have time go through all the steps to get a CCE encode done) and I was always of the opinion that it looked better than the transcoders.

Let's give credit where credit is due, Cinemacraft Encoder is the bomb. DVD-RB just makes it usable for DVD copies. Just my opinion, but I'd bet that QuEnc or HC would do well also.

[Fishman0919]

Funny you say that, the second part of this test is going to be the same movie clips done with different encoders at different bitrates: CCE Basic, QuEnc, HC Encoder, TMPGEnc and MainConcept.

This won't be for awhile (a week or two) but it's going to be pretty much the same test and under the same conditions.

[archaeo]

Fishman,

Are they able to look at these comparisons in the store side by side simultaneously rather than one following another?

I'm looking forward to the HC results, because what I've seen so far in captured stills is very hard to tell apart from CCE.

[Fishman0919]

I could do that, the 1st test I did that started this was having two of the same TV's side by side tune the same with the "Video Essentials" DVD... playing the same movie at the same point in time. One TV with a movie done with DVD-RB w/cce and one done with DVDShrink. I did this for awhile with different movie and I can say 8 to 9 out of 10 people would say that the one done with DVD-RB looked better. Some people would think it was different cables or different players but the bottom line was it was just different software.

[archaeo]

That's the best kind of testing - diverse sample, side by side, same clips, same hardware. Nice ;)

Look forward to your next tests w/HC. Should be interesting.

[robot1]

Originally posted by Fishman0919
Funny you say that, the second part of this test is going to be the same movie clips done with different encoders at different bitrates: CCE Basic, QuEnc, HC Encoder, TMPGEnc and MainConcept.

This won't be for awhile (a week or two) but it's going to be pretty much the same test and under the same conditions. If you have ProCoder, please could you add it to your test?
Thanks.

[Fishman0919]

I do have Procoder 1.5 and 2, I haven't use them in so long I kinda forgot about them. I should really put procoder in with the bunch but 6 encoders in the test might be abit much.

[Trahald]

nice going Fishman.. the 61" TV is a nice touch where as you might miss the impact of an encoder over a transcoder on a smaller set.

[dragongodz]

nice test. the only thing i would have liked is a little more info, such as compression ratios for the movies etc.

will be interesting to see the results of the encoders test aswell. :)

[jdobbs]

I'll be interested in seeing that too. While I am very confident that encoders will outperform transcoders pretty much every time (there are always those low-reduction exceptions), I have this nagging doubt in the back of my mind as to whether CCE is as far ahead of QuEnc and HC as I sometimes assume... that's one of those questions that can be answered by a blind test like this.

This is exciting stuff -- great job Fishman0919.

[dragongodz]

I have this nagging doubt in the back of my mind as to whether CCE is as far ahead of QuEnc and HC as I sometimes assume... that's one of those questions that can be answered by a blind test like this.
yes maybe Fishman0919 could ask if the choice they make is
a. a lot better than the others
b. slightly better
c. really hard to tell any difference betweens X and Y encoders so i picked the one which i think may be just a tiny bit better. :D

that would also say a lot about any difference gap. ;)

[DMagic1]

I too would like to see Procoder in the mix.

[Vanderlow]

Originally posted by DMagic1
I too would like to see Procoder in the mix.

Me too. We have a score to settle with the anti-procoder crowd over at videohelp. I want to stir up a hornet's nest!

[SkVid]

I've done a similar test twice with Procoder 2 and CCE. One was a cartoon for my kids, the other one was Matrix.
I was impressed about the good quality at low bitrates (about 3000 kbps) with the Procoder 2 (mastering quality). But CCE was by far the faster one. And the higher the bitrate the lower the differences. So it depends on the movie which encoder I use.
I'm looking forward for that new test!

[jdobbs]

I've noticed CCE doesn't do as well on (non CG) animated sequences as it might. Interestingly, if you look at the results of this test the only animated movie used was Titan AE and it was the one in which DVD-RB/CCE came out pretty even (14/13/10).

[artoor]

Originally posted by jdobbs
I've noticed CCE doesn't do as well on (non CG) animated sequences as it might. Interestingly, if you look at the results of this test the only animated movie used was Titan AE and it was the one in which DVD-RB/CCE came out pretty even (14/13/10).
In fact... there is just one difference of opinion between DVD-RB + CCE and IC8. One way or another DVD-RB seems to be winner with regard to arithmetical average ;)
Anyway the contrast between DVD-RB + CCE and (e.g.) Shrink in case of Star Wars and LOTR look better than in case of Titan AE. That's true :)

[monomer]

An excellent experiment... 37 clueless viewers makes the results even more reliable I think. However, I've just got a suggestion... how about giving the viewers a form and asking them to rate each clip from 1 to 10. This would turn your qualitative data into quantitive data and you could now perform some statistical analysis on it. Anyway, its just a thought...

Good work though and very interesting stuff I think...

[tigerman8u]

Good test useing other ppl for the results. Am anxious to see the next results as I use Mainconcept alot. Thnx

[quantum]

Originally posted by Fishman0919
backup all 10 movies, removing nothing... keeping all audio tracks and subtitles. After all the movies were finished I cut out a 10 - 20
This test obviously favors encoders. With most dvd9's, you can gain critical space by removing foreign language audio and any obvious junk like advertisements. By not doing this, you're creating the most extreme encoding job possible.

It's not a surprise to me that CCE gives better results under these conditions. But in the real world does anybody do it this way? I'd also like to see how far each job is getting from the original. So why not include the original disk in your test? If 100% of the viewers pick the original, you've obviously compressed too much and they're simply picking the best of the worst.

Data like this is often misinterpreted. In this case one might assume that CCE gives better results in all cases. But the reality could be that DVDShrink gives better results in moderate to light encoding jobs, but this wouldn't be revealed given the test conditions.

Why not try a few films reduced more moderately, say 15 to 35%, which can almost always be obtained by stripping some audio tracks. Also consider using the DVDShrink max sharp option which IMO gives the best results.

[DK]

I was impressed about the good quality at low bitrates (about 3000 kbps) with the Procoder 2 (mastering quality).

me too, i actually prefer procoder2 over cce when it comes to long titles/higher reduction levels

especially when reducing extra/bonus by some 25% or so procoder2 gives the better output for my eyes

cce is sharper, that's for sure but way more likely to produce artefacts as well then procoder2

the visual impression on my 100hz panasonic is better with procoder2 for many titles

[Fishman0919]

I could do that, the 1st test I did that started this was having two of the same TV's side by side tune the same with the "Video Essentials" DVD... playing the same movie at the same point in time. One TV with a movie done with DVD-RB w/cce and one done with DVDShrink. I did this for awhile with different movie and I can say 8 to 9 out of 10 people would say that the one done with DVD-RB looked better. Some people would think it was different cables or different players but the bottom line was it was just different software.


This test was just the main movie and one audio track ranging from around 90% to 70% compression.

Edit: I'm not saying DVD-RB w/CCE or just CCE is the SEE ALL, ALL MIGHTY of encoding. But to me and most of the people I have shown, movies done with CCE vs a transcoder when played on 2 of the same big screen tv's side by side.... cce wins

[quantum]

So now I'm confused. The first post said you kept everything:

Originally posted by Fishman0919
backup all 10 movies, removing nothing... keeping all audio tracks and subtitles. After all the movies were finished I cut out a 10 - 20
And now this?
Originally posted by Fishman0919
This test was just the main movie and one audio track ranging from around 90% to 70% compression.

[dragongodz]

But to me and most of the people I have shown, movies done with CCE vs a transcoder when played on 2 of the same big screen tv's side by side.... cce wins
yes but what people are wanting to know is, by how much ? at what compressions were the biggest differences ? etc etc etc.
so thats why the things like compression amount and a rating or some comments from the people is needed. that way we could gain more insight as to what the votes really mean.

[EDIT]
infact such information for an encoder test may be even more important to help tell how much difference there is between them.

[Fishman0919]

@quantum:
This test was just the main movie and one audio track ranging from around 90% to 70% compression.

This was a test I did a few months ago around Xmas time where there would be plenty people in the store, everyone was standing around look at TV's. It made it real easy to ask "Which picture looks better to you?", this one was more for my own personal curiosity.

@dragongodz:

My plans for the second test might have to change. There are just too many enocders to do: CCE, QuEnc, HC, Procoder 1.5/2, TMPGEnc, MainConcept. I myself would like to know by how much and etc... but to ask a person to sitdown and view several clips going back and forth again and again, then ask them to tell me which one was the best and by how much was it better then the next. I really don't think that it gonna fly, to much to ask.
Maybe something more simple like my first "trial" test, two TV's side by side... encoder A vs encoder B... which one looks better and by how much.

[jdobbs]

@Fishman0919

You must be wondering what you've gotten yourself into... next we'll all have you motoring around the countryside doing door-to-door polls. You don't have a "plasma screen on wheels" do you?

You did a test using real people that was needed to clear the air and I, for one, really appreciate it. If you do no more tests at all -- I still appreciate it.

[Fishman0919]

@jdobbs:

THX :)

[dragongodz]

next we'll all have you motoring around the countryside doing door-to-door polls. You don't have a "plasma screen on wheels" do you?
HAHAHA he will need more than one though. :D

You did a test using real people that was needed to clear the air and I, for one, really appreciate it. If you do no more tests at all -- I still appreciate it.
i think its pretty obvious from peoples comments that people appreciate it of course. that doesnt mean we wouldnt like to see more of the same or even better with more information. :D

to ask a person to sitdown and view several clips going back and forth again and again, then ask them to tell me which one was the best and by how much was it better then the next. I really don't think that it gonna fly, to much to ask.
so you dont have half a dozen tvs you can playing all at once ? (joke) ;)

Maybe something more simple like my first "trial" test, two TV's side by side... encoder A vs encoder B... which one looks better and by how much.
yes that would be atleast some extra detail to more quantify the results and differences. its more like what i suggested on the first page of this thread.

[pg55555]

DK me too, i actually prefer procoder2 over cce when it comes to long titles/higher reduction levels especially when reducing extra/bonus by some 25% or so procoder2 gives the better output for my eyes cce is sharper, that's for sure but way more likely to produce artefacts as well then procoder2 the visual impression on my 100hz panasonic is better with procoder2 for many titles
But it isn't to control this what the Quality Prec setting in CCE is for? From the Settings Sticky:

"Quality_prec
...This setting controls whether CCE gives priority to the fine details of the image, or evenly colored flat areas. A low setting will give priority to details, but could result in blockiness or color banding. A high setting will favor flat areas, but could result in edge artifacts."

If I understand this correctly, it would mean a higher Quallity Prec setting would produce a softer image, with less macroblocks, similar to the output you like from Procoder

[jdobbs]

What a lot of encoders do is soften the image a little. Soft images encode easier in MPEG than sharp contrasts. It's a trade off -- sharper images with more detail and more artifacts, or softened images that lose detail but have fewer artifacts. You can actually soften the image slightly with a filter in AVISYNTH and get a similar effect in CCE.

[MackemX]

Originally posted by jdobbs
What a lot of encoders do is soften the image a little. Soft images encode easier in MPEG than sharp contrasts. It's a trade off -- sharper images with more detail and more artifacts, or softened images that lose detail but have fewer artifacts. You can actually soften the image slightly with a filter in AVISYNTH and get a similar effect in CCE.

some people like smooth and some people like sharp. I prefer sharper images myself especially when you get closeups as you don't lose the intricate details :)

[Methanoid]

This kind of test is just what is needed. Real world, blind and no hidden agendas.

I'd be interested personally in a comparison that looked at say 2 pass, 5 pass and 8 pass CCE. I have friends who encode with 8-10 passes which to me seems excessive. I wonder if they can tell the difference.

Also would be intrested in a comparison of encoders but perhaps not such a large range as I appreciate it is a huge investment in your time.

[Sir Didymus]

Hi Methanoid. :)

Did few testing, some months ago, trying to compare OPV and multipasson encodings...

So, I have already available a wide collection of cells, encoded form a single movie, at different bitrates.

If you are interested, I can author three DVDs, based on a given cell encoded with 2, 5 and 8 passes and post the links for downloading, without stating on how many passes they are made.

At what bitrate/CCE matrix are you interested in ?
- possibities with Std Matrix: 4250, 3750, 3250, 2750, 2250;
- possiblities with Very Low BR Matrix: 3750, 3250, 2750, 2250, 1750;
- possiblities with Ultra Low BR Matrix: 3250, 2750, 2250, 1750, 1250;
- possiblities with KVCD Notch Matrix: 3750, 3250, 2750, 2250, 1750...

Cheers,
SD

[Methanoid]

Originally posted by Sir Didymus
Hi Methanoid. :)

Did few testing, some months ago, trying to compare OPV and multipasson encodings...

So, I have already available a wide collection of cells, encoded form a single movie, at different bitrates.

If you are interested, I can author three DVDs, based on a given cell encoded with 2, 5 and 8 passes and post the links for downloading, without stating on how many passes they are made.

At what bitrate/CCE matrix are you interested in ?
- possibities with Std Matrix: 4250, 3750, 3250, 2750, 2250;
- possiblities with Very Low BR Matrix: 3750, 3250, 2750, 2250, 1750;
- possiblities with Ultra Low BR Matrix: 3250, 2750, 2250, 1750, 1250;
- possiblities with KVCD Notch Matrix: 3750, 3250, 2750, 2250, 1750...

Cheers,
SD

WOW! Yes, well not fussed as to bitrates particularly. My thought is more that I doubt most people could tell between 5 and 8 passes.

Just choose the one you think makes most different and let us have a look???

Thanks!

[Sir Didymus]

Originally posted by Methanoid
My thought is more that I doubt most people could tell between 5 and 8 passes.

Just choose the one you think makes most different and let us have a look???


The problem is that me personally even more doubtful than you about the possibilities of visually discriminating among 2, 5 and 8 passes at whatever bitrate you like, keeping constant all of the other conditions, like encoding matrix, DC precision, etc.
:sly:

Ok. I will pick let's say 2250 kbit/s, with CCE Standard matrix...

Ah, I am from PAL land; is this a problem for you ?

Edit: sorry for the silly question; noticing now you are from UK, so it isn't...

[Sir Didymus]

@Methanoid:

Here you may dld three dvd; all have been encoded with CCE, at the same bitrate (2250 Kbit/s) and with the same settings, except the number of passes [1+1, 1+4, 1+7]...

Of course, at this bitrate, the quality is not so high, and if you put the player in still mode, you may notice evident artifacts and blockiness, but this, IMHO is almost the same for all of the three...

It would be nice if you were able to tell me what of the three dvd is encoded at 1+7...

Cheers,
and nice w/e
SD

dvda:
http://webmail.libero.it/r.php?d=libero.it&wr=3zqcxucf&ws=3zqcxucf&e=libero.it&c=so5JzAyFoOX3JPiNifxuKDVWeTwdilqX77420

link valid until 25.03.2005 at 08:05.


dvdb:
http://webmail.libero.it/r.php?d=libero.it&wr=3zqcxucf&ws=3zqcxucf&e=libero.it&c=gHjEOr0WXdl15PEYbBqtA2l8mJz0bcfm77424

link valid until 25.03.2005 at 08:10.


dvdc:
http://webmail.libero.it/r.php?d=libero.it&wr=3zqcxucf&ws=3zqcxucf&e=libero.it&c=HUGvXGtLsVao1ImJ6UD0kQxmt0TMToNl77428

link valid until 25.03.2005 at 08:15.

[Methanoid]

Originally posted by Sir Didymus
@Methanoid:

Here you may dld three dvd; all have been encoded with CCE, at the same bitrate (2250 Kbit/s) and with the same settings, except the number of passes [1+1, 1+4, 1+7]...

Of course, at this bitrate, the quality is not so high, and if you put the player in still mode, you may notice evident artifacts and blockiness, but this, IMHO is almost the same for all of the three...

It would be nice if you were able to tell me what of the three dvd is encoded at 1+7...

Cheers,
and nice w/e
SD


OK well I watched first 30 secs of each back to back and could not see any diff. So I then loaded the VOBs into Womble MPEG2VCR and looked at the frame at 30:00 and I RECKON (best guess)

A (best)
B
C (most artefact)

But lets see if anyone else wants to report???

[dragongodz]

lets see if anyone else wants to report???
or how about continuing this in a different/new post ?

[mrbass]

Reminds me when I hooked up s-video connection to my dreamcast. All the smoothness disappeared. Sure it was clearer at a higher resolution but many jaggies showed up in Soul Calibur, etc. Not sure how many of my neighbors have 61" TV's. My next door neighbor has a 55" HDTV but I'm guessing he's the exception rather than the rule. My point? Basically if you have a huge TV, say 42" or above you'd be silly not to use CCE with DVD-RB if your going to backup the whole movie or just start saving money for a HD-DVD collection. But if the same tests were done on 32" TVs the results might not be the same. Maybe I just live in the ghetto part of town with neighbors with dinky TV's, myself included. Only way I'd recommend using DVDShrink, dvd2one, dvd95copy, ic, etc. for large TV is if you do main movie only, one audio stream, and perhaps even editing out some credits, etc. so get a low compression ratio.

[jamewoong]

The accuracy of this test is poor! We're 2007 (I was not here at that time, if not...), so let me bump this for you. You test it by using people's point of view. It's not a good analyse as people that you asked for the quality is lower than 50! So this statistic is useless because if there are two videos that got a similar quality, then they will choose one of them by random.

You compare the DVD Shrink, Instantcopy 8 and DVD Rebuilder, but you forgot to mention the time used for the encoding! If DVD Shrink took 1/4 of time, then we should think about it. i.e. I can take up to 1 years to analyse a video by extracting and compressing a video frame-by-frame. Let's compared its quality and you'll see that mine will be the best of all! Duh...

Well, your test is just a guess as we all know that time of encoding will increase the quality...

EDIT: That's why it's Blind test... lol

[archaeo]

You test it by using people's point of view.



Yep. Quality is in the eye of the beholder. Period :)

[Video Dude]

Well, your test is just a guess as we all know that time of encoding will increase the quality...

You can't measure quality by the time it takes to encode.

[SuLyMaN]

Quality vs time is really no trade off at all :) I rather encode a DVD whole night than to watch a blocky DVD transcoded in like 15mins....

[jdobbs]

The accuracy of this test is poor! We're 2007 (I was not here at that time, if not...), so let me bump this for you. You test it by using people's point of view. It's not a good analyse as people that you asked for the quality is lower than 50! So this statistic is useless because if there are two videos that got a similar quality, then they will choose one of them by random.

You compare the DVD Shrink, Instantcopy 8 and DVD Rebuilder, but you forgot to mention the time used for the encoding! If DVD Shrink took 1/4 of time, then we should think about it. i.e. I can take up to 1 years to analyse a video by extracting and compressing a video frame-by-frame. Let's compared its quality and you'll see that mine will be the best of all! Duh...

Well, your test is just a guess as we all know that time of encoding will increase the quality...

EDIT: That's why it's Blind test... lol You are just flat wrong. The test is to measure quality. What does that have to do with time?

My suggestion: If you want to justify your use of DVD Shrink -- you don't have to do so. Just use it. It's a fine piece of software. DVD Rebuilder and DVD Shrink are just not similar at all, they work two entirely different ways, and are aimed at two different audiences.

I for one feel that the quality difference has been proven beyond any reasonable doubt, not just through this thread but also innumerable others. Let's please not start this tiresome argument again.

[Digga]

You test it by using people's point of view. It's not a good analyse as people that you asked for the quality is lower than 50!we're looking forward to your own grand scale tests. please leave an announcement when it's finished here.

So this statistic is useless because if there are two videos that got a similar quality, then they will choose one of them by random.1.quality is subjective, so not everybody will perceive video A and B in the same way. even if they are the same, people might still rate them differently (e.g. due to the placebo effect).
2. just because two items look the same doesn't mean the test is useless. why would it? it's just plain wrong.

I suggest you redo your statistic classes before you continue posting in such topics with such manners.

[jamewoong]

I suggest you redo your statistic classes before you continue posting in such topics with such manners.
Who care about people's perception. Science beats all. If there's something that can proove that the test was correct via PC analysis, then okay. But using people's perception doesn't have an accuracy of 100% (for me, okay?)!

Well, it was just my reaction... Don't know why you get mad. You are the one that should return to your statistic class! :p hehe The more people you get from a stats, the better result you'll get. And 50 people is a correct number depending where... Below that number, the stats is poor.

Between, probably I was wrong about the speed. But how come increasing the pass will get a better quality? And increasing it, it will also increase the time --> this is my point about the time. And yes, there's a limit to a max stats. i.e. After 3-5 pass, the result is the same... Duh...

Anyway, as the author said, we should stop discussing about this matter.

Let's please not start this tiresome argument again. okay? Thanks. hehe

[Fishman0919]

Who care about people's perception. Science beats all. If there's something that can proove that the test was correct via PC analysis, then okay. But using people's perception doesn't have an accuracy of 100% (for me, okay?)!


What are you talking about.... "who cares about people's perception".... mpeg2 video is basic on visual perception and what people think they see and not whats there. 29.976 FPS and you think every one of them looks as good as the next. Before you spout off about things you don't or can't understand... you might want to read up on visual perception with mpeg2 encoding, I frames, B and P frames, GOP's and how encoders differs transcoders

EDIT: A good place to start reading LINK (http://bmrc.berkeley.edu/frame/research/mpeg/mpeg2faq.html) and LINK (http://bmrc.berkeley.edu/research/mpeg/faq/mpeg2-v38/faq_v38.html)