TEST/POLL: All different DVDtoDVDR methods

[adam]

No its not possible for a compressed domain transcoder to change an I frame to a B frame or vice versa. Something must have gone astray somewhere.

Even if there is visible difference between all your sample frames and your source frame, that still doesn't address the fact that sample 1 may be compressed by %75 while sample 2 may be compressed by %10.

I've used this analogy before, but testing video quality by sampling a single frame (especially when using compressed domain transcoders) is like judging who has more overall money by looking at a single coin from their pocket. That single sample viewed in isolation is no indication at all of what the rest of the samples will be, and it has no meaningful bearing on the total final output. It simply tells you nothing at all.

[DereX888]

@adam

but testing video quality by sampling a single frame (especially when using compressed domain transcoders)

I thought I explained the selection in previous post?
Each final video was burnt onto dvd-rw and viewed on a regular 42'' TV for couple minutes by 4 people (including me). We watched about 3-4 minutes of each. few times we play it at 1/4 speed, zoomed picture up to 6x etc.
All of us agreed without questioning on 2 best samples instantly, while third sample was just arbitrarily added by me (due to too many choices of 'jurors').
Therefore all 3 sample pics are from the selected, visibly best videos.
Took me long time to find all 3 the same I-frame frames on each (and same frame on original came out not to be an I-frame - but anyway you supposed to judge which sample looks best, not the original).

Even if there is visible difference between all your sample frames and your source frame, that still doesn't address the fact that sample 1 may be compressed by %75 while sample 2 may be compressed by %10

Adam, that actually does not make much sense to me what you are addressing here. Each sample was made by different encoder/transcoder from very same source, and their task in this test was to accomplish target of fitting the source onto a single DVD-R at best quality they can. How they accomplish it - by encoding/transcoding it up to 75% or by 10% does not matter, or to explain better: if any of them could accomplish putting the source on a DVD-R without any compression - we would have obviously had an instant winner
I dont even know at what compression ratio some transcoders do their job (i.e. DVD2One won't show anything AFAIR), and whatever settings those programs "came to be set" by themselves to fit the source on a single DVD-R is what was judged.

[adam]

With all due respect, I don't see how we can judge these samples according to what you and your friends saw on your tv. You posted single frame samples and asked us to evaluate them. I explained how this is impossible to do with any accuracy.

I chose to throw in the (especially with compressed domain transcoders) because of the way transcoders work. There are basically two parts to an mpeg file, motion vector information and DCT data. Compressed domain transcoders can ONLY compress DCT data. Any given frame might be filled with DCT data or have very little at all, thus a transcoder has to pick and choose which frames it compresses. So if you set your transcoder to compress to %50 it will selectively pick frames to compress, whereas an encoder will compress all frames relatively equally. Even though the entire clip is overall compressed by the same amount, any given frame chosen from those video samples may be highly compressed or not compressed at all (beyond that of its source of course.)

It does not matter that YOU viewed these video samples, determined their relative quality, and then chose single frames to represent their quality. It does not matter that all video samples were overall compressed by the same ratio. What matters is that any given frame within that video that you chose may have been compressed a great deal or very little.

Let's assume that this single frame you chose was comprised of %90 motion vector information and %10 DCT data. That means that at the absolute most none of the transcoders would have been able to compress it by more than %9.9999. There's simply nothing left to compress at this point. Yet if you use an encoder it could compress this frame as much as it needed to.

Comparing a frame compressed by ~%10 to that same frame compressed by ~%90 tells you absolutely nothing. Jump to the next frame in the sequence and you may get the exact opposite result.

[DereX888]

Comparing a frame compressed by ~%10 to that same frame compressed by ~%90 tells you absolutely nothing. Jump to the next frame in the sequence and you may get the exact opposite result.

I got it, understood you perectly - and thx for the nice explanation.

There is only 1 transcoder sample among them 3 best, so far only 1 person thought it looks better (weird...)

[gitreel]

Derex, I am comfortable with my sexuality. I just don't spend time watching porn. I have seen porn before, but I get really uncomfortable when other men start talking about their schlongs, because I don't need to hear about it. I also do not spend time looking at mr. happys; whether it was real or not was not the point.

You had a schlong for a spaceship, which imo was bad taste considering you are a male. The bear with a penis in your avatar also would imply that you are a little fruity.

[FulciLives]

Derex, I am comfortable with my sexuality. I just don't spend time watching porn. I have seen porn before, but I get really uncomfortable when other men start talking about their schlongs, because I don't need to hear about it. I also do not spend time looking at mr. happys; whether it was real or not was not the point.

You had a schlong for a spaceship, which imo was bad taste considering you are a male. The bear with a penis in your avatar also would imply that you are a little fruity.

At first your ranting and raving about the penis stuff was funny as I thought it was tongue in cheek but now it sounds as if you are serious since you just can't seem to "let it go".

In short you sound like a homophobe.

Get over it!

- John "FulciLives" Coleman

P.S.
One of my old Avatars

[stiltman]

I find it very disturbing that you want to show off your toys in a public forum





just wait till this hits and stays on google

[DereX888]

@FulciLives
I find your old avatar funny, but I wouldn't use it myself - too real looking to me.

@everyone
First: My bear avatar is a 'flasher', not just a penis as i.e. your old avatar was. You cant have a 'flasher' who's not flashing his schlong, can you?

Second: Another 'penis' you pick on is a *spaceship* in the shape of a penis. Still it is not a penis, not even as close to as your old avatar is.

Third: I like porn. Many people like porn. Porn is legally sold in most of the countries. There is nothing wrong with the porn, so whats your point? You dont like it - fine, no one is forcing you to watch it.
However the movie in question here is not a porn either. It is an 'erotic comedy', something what is shown on many public tv stations throughout the world on Fridays every week (ofcourse except mighty puritan US of A, where they rather show gut-ripping action war flicks rated PG instead of a beauty of a human body, but thats OT).

Took us few hours to prepare this test. Hopefully only some of you won't appreciate its 'twist' and won't find it funny in the same time, but if not -not a big deal, just few wasted hours. I still appreciate whatever is your input on this subject anyway, and if it is really so outrageous to most of you I'll just take it down.

[DVDDave]

alright already--enough with the ******* penis. i can not believe someone would use a teddy bear with a penis as an avatar. that pic should be removed ASAP

[DereX888]

You can pry it out of my posts only over my dead cold computer!
My avatar stays.

[ViRaL1]

Never really had the urge to look that close.

[GKar]

You can pry it out of my posts only over my dead cold computer!
My avatar stays.

DereX888, I have to admit you are a cool dude (a dude I assume) as you maintained a very good sense of humor throughout the assault. Kudos man. Curses and other expletives would have been flying from a lesser person.

[petar]

Not crazy about looking at penises myself but Derex888 is OK.

I agree with Adam that looking at isolated pictures you won't get an objective result. Try posting several consecutive pictures from the same GOP (not sure if Baldrick would be happy about it though). Also posting parts of the picture may not be the best choice. The actual motion scene may be a tough one (as you described it, though one can't tell from a single frame) but you are then comparing only parts of that picture. That's not very objective either. MPEG compression splits the picture in slices and then in macroblocks (and then in blocks). A single picture may have macroblocks which are very well reconstructed and others not. So even though the "penis rocket" may look great, other parts of the picture may not.

Also just one moving object on a static background, no matter how fast is going and even though there may be occasional stars flashing, is actually not such a difficult scene. As I mentioned, because the picture is split in MBs compressing this is not such a big deal.

[gitreel]

I too think Derex is ok.

[jimmalenko]

Each final video was burnt onto dvd-rw and viewed on a regular 42'' TV for couple minutes by 4 people (including me). We watched about 3-4 minutes of each. few times we play it at 1/4 speed, zoomed picture up to 6x etc.
All of us agreed without questioning on 2 best samples instantly, while third sample was just arbitrarily added by me (due to too many choices of 'jurors').
Therefore all 3 sample pics are from the selected, visibly best videos.

Having had the benefit of three other people watching samples of 3-4 minutes duration and collating their opinions, you then run a poll and post miniature still frames on a public forum of so-called "experts" who know their shit, and wonder why people are jerking your chain ?

Can I add that I like porn too ?

I voted for sample 3, BTW

[DereX888]

[...] removed just to 'get over' and don't invoke any further useless p***s discussions [...]


@jimmalenko

...and wonder why people are jerking your chain ?

I won't lie saying I wasn't expecting it at some point

Can I add that I like porn too ?

I voted for sample 3, BTW

cool!

I could have ask have you voted for the best looking sample of a 'frame', or was it for the best looking sample of a 'spaceship' but it is only a rhethorical question that just came to my twisted mind


@gitreel @GKar
TY.
I too think most of people using this board are OK!
(except for the puritans, politically correct a*holes, RIAA & MPAA supporters, rascist white/yellow/red/brown/black-skin pigs, illegal aliens stealing jobs from righteous citizens, greedy corporate bastards moving everything to Asia and 3rd World, damn porn watchers and penis-like spaceship supporters - oops, wait, that is, umm, me never mind )


@petar

just one moving object on a static background, no matter how fast is going and even though there may be occasional stars flashing, is actually not such a difficult scene

I also would say so, but actually it was difficult - at least for one of the most popular encoders here (judging by the number of guides including it), and thats why there is no sample made by TMPGEnc in this test.
With its max 2-pass VBR encoding the sample of the begining of the movie was really very bad quality, them flashing stars had so much blockiness.
Again - I agree with you, Adam, and everyone eles questioning credibility of such tests not using video clips.


Probably most of you guessed already from what programs 'best looking' sample came from, but give it few more 'curious' people to vote (to have some round number of votes) and I'll end this poll in next post with info on what made samples 1, 2 & 3.