mpeg2 quality (> or <)

[Premjit V.P.]

If I encode a file in mpeg2 using a good software (like cinemacraft) with all the parameter as per the mpeg2 specifications and giving maximum work for the encoder (in terms of quality), will I get a result that can be compared to that of broadcast quality. Do I really require a hardware encoder for that? The source footage is assumed to be good and encoding time is also considered not to be an important criteria.

Thanks in advance for your consideration and reply.

[vhelp]

At best, you'll get "close" to the original source, but never exact.
At least not w/ any software (or hardware) encoders I have come
across and use (mainConcept; panasonic; cce; tmpg; and a few others)

It all boils down to what you will "part with" in terms of quality.
You'll have to comprimise on a few points.

* dct and macro blocks already present (before you capture it)
* noise
* capture device
* your skills level
* ect. etc.

I throw in skills, for a very good reason. Because what one person
obtains in quality (good quality) from say, mainconcepts (MC) encoder
may not be the same results obtain if another person used that same
encoder MC. Too many variables. Mostly due to knowledge.., or lack
there of. Also, has to do with the "source" they are encoding from.
ie, ..what is present alrady, in the source. Is it clean. It it full
of macro blocks. etc. etc.

To be honest, you wont obtain "broadcast" quality if you encode with a
lower bitrate than that which was used by *them* when they aired it.
When you capture it (analog capture) you are capturing everything.
That means, their highest bitrate they used. An analog capture is the
highest form of quality you can get. And DV is second to that (imo)
However, you could look at a *good* hardware MPEG-2 capture device as
a second too. Just so long as you don't re-encode from it.
To obtain the highest quality (or close to broadcast) you have to be
pretty skillfull at the process. You have to know how to look for things
not normally realized, etc. It's easy for someone to just give you a
"template", but even that is not enough. This forum is FLOODED with
hundreds of "same-old" posts asking the "same-old" questions, over and
over and over again. My argument is valid. The bottom line, is skills.
Knolwedge. And knowing how to master your tools. Be if cce or TMPG etc.
Put cce in the hands of a non-skillfull person, and they'll turn up crap.
But put it in the hands of a craftman..


Here is some of my wisdom ...

The recipe is:

* source quality
* tools
* knowledge

The rest (or in between) is up to you (the given person behind the goal)
It's taken me about four years to achieve what I have in terms of quality
today. And I didn't do it with a template. I did it through a series of
trial n error scenarios.

If you want to *try* and get close to broadcase quality, you'll have to at least
know (remember) a few simple notes.

* Know how to *see* what a given source is outputing. Is it Film or Interlace.

Interlace is 29.970 fps and every frame is interlaced. Nothing is Progressive.

or..

Film is 24 fps. But it's Telecined to 23.976 with pulldown to make it 29.970 fps.
You also have to know, if the telecined was done correctly or illregulary.
That is, is it badly done that you can not properly IVTC it back to 23.976 or
is it a clean process straight through. The reverse of a telecined 29.970 fps is
IVTC.. as I just stated.
.
If the broadcaster aired something that was Film, truely shot on film and they
Telecine it (whom ever *they* are, before the broadcasters get their hands on it)
..and the film was done properly w/ no issues, then a simple IVTC step is all
that it takes to get *that much* closer to your goal (broadcast quality)
But, if the Telecine process was done poorly or illregulary, your goal of IVTC
(hence, broadcast quality) will not be met or near. That is my OP, and also
fules my comments on the reason why I said that this forum is flooded with
the same old questions, ..why.
Only knowledge will finally answer your question(s) to issue/goal.
But, by then, you'll fianlly realize that you can't obtain the same level of
quality as the broadcasters, because you'll have finaly learned that there are
other factors that strenthen my argument. Such as color space. Encoding will
reduce the color space. That's just an example. I didn't have to include it
Well, anyways.

As far as Interlace goes ...

Because Interlace sources are 29.970 fps, there is much more pixel information
to encode. Add to that, the frames are loaded with Interlace lines that have
to be encoded. this is seen as Noise (if you wish, else think of it as noise)
When you encode it, you are not employing any benfits (such as IVTC) so your
milage will vary. That is, ..will depend largely on your skills level, and the
processes you use to obtain your final goal.. broadcast capture -to- encode -to-
author, etc.

That's my two cents worth.
-vhelp

[Premjit V.P.]

You have slightly misunderstood my question but that seems to be worth as you have taken lot of pain in answering and the contents is good enough to be kept as a reference material.

My question is only to know the quality difference between a Hardware encoder and a software encoder for mpeg2. There is no question of capturing, instead the source is assumed to be digitally recorded using a professional DV camera. It is also assumed that the nature of the source is well understood (in terms of knowledge and tools).

[lordsmurf]

DV camera.

This is the capture device? You just lost quality. 4:1:1

[vhelp]

If I encode a file in mpeg2 using a good software (like cinemacraft) with all the parameter as per the mpeg2 specifications and giving maximum work for the encoder (in terms of quality), will I get a result that can be compared to that of broadcast quality.

Oh. Then the short answer is.. probably not.

The reason why I said probably not (imo) is because you and I do not know "for sure"
what *they* used to encode their source materials with.
Also, sometime to think about here.. that the materials aired (or referenced)
may have been shot with various cameras (24p; 24i; 30p; 60i etc etc, I'm guessing
here) and each movie or show aired are not shot w/ a standard camera. Using that
same camera for every tv program, etc. So, it's really difficult to answer such a
question. But we could say, in a generic sence (IOW.., from a newbie point of view)
you can.. speciall when people starting throwing names like CCE into the picture
.
Too many variables to make an accurate response. What abuot lighting; cam holding;
panning and many more areas of discussion. Just because one has a pro cam does not
mean that "by default" they are going to get broadcast quality results.. specially
every time.

I wish I had a 24p cam. I'd be happen even with a cam that Telecines to 29.970
..the reason ? Because I can easily IVTC it back Anyways.

Another reason why its hard to answer accurately is because we don't know what
you are "gauging" against, let alone see *your* source matierals that you are
having trouble with. So, I'm using my wizardry and guessing here

Cheers,
-vhelp

[FulciLives]

In most cases a software MPEG encoder will do a better job (especially CCE) than a hardware encoder (at least those kinds that we mortels can buy for a decent amount of money).

The only time I would choose a hardware encoder is if I am putting less than 60 minutes on a single DVD because then you can just max out the bitrate for the video and audio.

Even then you don't get the benefit of applying any filters to clean up your source so hardware encoding is only good if you have a very clean source or some sort of hardware filter type box ... like the kind spoken about in this thread: CLICK HERE

Good Luck !

- John "FulciLives" Coleman

[adam]

I had the occasion to test drive a few high end (well for me) hardware encoders and I didn't really see much improvement over the software encoders that I use now. Sure they were fast, but when it comes down to it mpeg quality is largely the result of the algorithms used in gathering and interpreting the raw data, and so in this regard there is no benefit to doing it in hardware versus software. There are many mainstream commercial DVDs out there in the market that have been encoded with CCE and MainConcept, and probably a few other software based encoders that you would recognize.

In my opinion, assuming all else equal, yes you will get comparable quality with a software based encoder alone.

If he doesn't find his way into this thread on his own, try pm'ing BJ_M. He works in a studio and I'm sure he has test driven some very high end hardware mpeg encoders/cards and I know he does alot of software encoding too...so he should be able to give you some first hand advice.

[FulciLives]

If he doesn't find his way into this thread on his own, try pm'ing BJ_M. He works in a studio and I'm sure he has test driven some very high end hardware mpeg encoders/cards and I know he does alot of software encoding too...so he should be able to give you some first hand advice.

That was a point I was trying to make in my earlier post in this thread.

Us normal people without deep pockets can only afford what? ... consumer grade hardware MPEG encoders like the Hauppauge WinTV PVR-250 or the ADS Instant DVD 2.0 or a stand alone DVD recorder etc.

So if your choice is something like that vs using a 2-pass or mutli-pass software MPEG encoder ... I say the later would be the better choice for ultimate quality.

- John "FulciLives" Coleman

[jimmalenko]

Unfortunately, I think the purist has shown through in a few people. No biggie, just possibly a little too "perfectionist" considering the OP's question IMO.

I think the biggest issue is whether the quality gap is noticeable. Analytically we might be able to say that you cannot replicate the source, but practically, we can say that provided you use the right methods, software, settings and possibly a couple of filters thrown in for good luck, you can IMO quite easily achieve a perceptive quality that is very close, if not equal to, and sometimes better than the source. It is certainly more apparent on low bitrate/resolution sources anyway. You can quite easily make a VHS capture look better, for example.

[Premjit V.P.]

Great responses. I did'nt have a hardware encoder but was planning for one. I feel better now and would rather go for another computer than for a hardware encoder.

Many thanks folks, and special thanks to vhelp.


Bye

[vhelp]

Thanks Premjit V.P. ..for your kind words.

@ jimmalenko

I disagree w/ the VHS comment

Using my ADVC-100 and finally converting it through TMPG (cause I
feel like a pro at this tool hehe) and a flawless IVTC process..
No matter what bitrate I use for (from these VHS sources) I just
cannot obtain the quality *closest* to the original VHS tape I
captured. I've just finished performing (yet) another test w/
a bitrate of 15k (15,000) and when I played it through PowerDVD,
I still noticed the chizels.
.
This device (though perfect in everything else) is just not good
w/ VHS sources. Period.
.
It's all about *the* capture device that captures the source.
And VHS is just not cutting it w/ the ADVC-100 box. Sorry.
It very much upsets me because when you capture through the ADVC-100
the source is as sharp as the original source (including VHS) ..and
that really pisses me off, because even though at 15k birate and
the "chizels" I see, the source output quality of sharp as the
original. I love it. But, use it (the ADVC-100) with anything else,
and you won't go wrong here. Period.

Howerver, speaking on VHS still..

When I run the same footage VHS tape through it and my TRV22 (using
it's passthrough) (which is MV 'free) the VHS is smooth and no
glitches in the final MPEG-2 video. IMO, its night vs. day. So,
for my VHS projects, the TRV22 is *THE* device to use.

A tip here ...

Also, I use IVTC built-in to TMPG. I love it. I use the short
method, which is instantanious, and no waiting Anyways.
.
With VHS, and when I IVTC, for sharpest look (IMO) be sure to tic
the Interlace=true (settings\advance\video Source type) The reason
I suggest this is because the source is Interlace (though Telecined
to 29.970 fps ie., II PPP II PPP II PPP) and, IMO seems to my
give me sharpest or original looking VHS reproduction.
.
But, I won't say better ..just as good (closest, that is)

@ fulci ...

Don't forget the DVD Xpress as well. It's a great box, and its really
supereior with VHS too Really.
.
Also, IMO, I feel that when using the Hardware MPEG devices for your
(those that use it as such) re-encodes, one should use the highest
bitrate their hardware will allow. My DVD Xpress allows 15k. You
see, for 2nd generation encodes, you want to have the highest quality
"original" to the source as you can get. Much like an uncompressed
AVI. You want no blocks in your hardware mpeg captures. Don't
assume that a lower bitrate is enough, just because the hardware is
that.. hardware. Always use 15k, cause that's the limit on most
hardware encoders. And, should think of AVI, when doing this. If you
know what I mean

@ Premjit ..

FWIW.. those peoples that dont' have the eye for detail when
they first look at a source (broadcast) and see no issues, its fair
to assume that what you see (based on your encode project) that you
have received the same level of quality as the broadcast. Until
you receive the eye for detail etc through the many battles
of capture/encoding. I've performed many many many many many.
And based on the many's, I share my wisdom (or opinion) here

Be Good,
-vhelp

[jimmalenko]

@ jimmalenko

I disagree w/ the VHS comment

Using my ADVC-100 and finally converting it through TMPG (cause I
feel like a pro at this tool hehe) and a flawless IVTC process..
No matter what bitrate I use for (from these VHS sources) I just
cannot obtain the quality *closest* to the original VHS tape I
captured. I've just finished performing (yet) another test w/
a bitrate of 15k (15,000) and when I played it through PowerDVD,
I still noticed the chizels.
.
This device (though perfect in everything else) is just not good
w/ VHS sources. Period.
.
It's all about *the* capture device that captures the source.
And VHS is just not cutting it w/ the ADVC-100 box. Sorry.
It very much upsets me because when you capture through the ADVC-100
the source is as sharp as the original source (including VHS) ..and
that really pisses me off, because even though at 15k birate and
the "chizels" I see, the source output quality of sharp as the
original. I love it. But, use it (the ADVC-100) with anything else,
and you won't go wrong here. Period.

Howerver, speaking on VHS still..

When I run the same footage VHS tape through it and my TRV22 (using
it's passthrough) (which is MV 'free) the VHS is smooth and no
glitches in the final MPEG-2 video. IMO, its night vs. day. So,
for my VHS projects, the TRV22 is *THE* device to use.

A tip here ...

Also, I use IVTC built-in to TMPG. I love it. I use the short
method, which is instantanious, and no waiting Anyways.
.
With VHS, and when I IVTC, for sharpest look (IMO) be sure to tic
the Interlace=true (settings\advance\video Source type) The reason
I suggest this is because the source is Interlace (though Telecined
to 29.970 fps ie., II PPP II PPP II PPP) and, IMO seems to my
give me sharpest or original looking VHS reproduction.
.
But, I won't say better ..just as good (closest, that is)

That's cool, vhelp. If we all had the same opinions and perceptions, it would be a pretty f**ked-up world

I guess that I'm lucky in that I'm in PAL land and don't have to worry about IVTCs or anything like that. Like you, I have my VCR hooked up via an ADVC-100. I transfer using ScenalyzerLive into Canopus-compatible DV. I then use the Convolution3D filter in AVISynth and *always* encode to 352 * 576. I am also cheating a bit because my comments were directed towards watching the final product on a settop player, as opposed to a software player on your PC. In my experiences, I have found that regardless of the source (even DVD), PC software players always make it look worse than settop players, using default configurations.

I still stick by my opinion regarding the ability to match or improve VHS, with the customary disclaimer that individual results may vary.

This thread has developed into a very interesting discussion