Help! Bit rate question for DV > DVD conversion

[BreakingTheSilence]

Hi!
First time posting... I've been working on converting old VHS home movies to DVD. I'm using a DV camcorder and a firewire card and the software gives me several options for bit rate settings for mpg2 compression. The default is 6000 bits in one program, 8000 in another, and I can go all the way up to 9000. I'm assuming that 9000 bits is the best, but will I notice a difference? Is the space/quality trade-off worth it? I want to preserve as high quality as necessary for a good viewing experience, but I want to maximize the space utilization on the DVDs so I can put several segments on one DVD. All and any suggestions would be welcome and appreciated. I'm using Pinnacle Studio and TMPEnc.

Thanks!

[LeTrout]

I haven't tried this yet, still on my to-do list. But I'll throw in my $0.02 (probably less) anyway. First, I would suggest using half D1 (352x480 iirc) instead of full D1 (720x480) for your resolution. I doubt a VHS source could make use of the full D1 res, so you'll probably get better quality (for a given bitrate) at half res.

If you go half D1 res, I doubt you will need bitrates as high as 6 Mbps. I typically do DV->MPEG2 conversions now (full D1 res), and I find that 6 Mpbs gives me very good results. If you use half that resolution, 3 Mbps would be an equivalent "high quality" bitrate. (note - these are avg bitrates, I always use 2-pass VBR with max bitrate ~9Mbps). Again, with VHS source (not so great quality going in) you may be able to push the bitrate considerably lower without noticeable loss in quality. This may be less true for bitrate than it is for resolution though. Also depends on things like the quality of your VHS original, amount of motion, etc. This is all pure speculation by me since I haven't done it, so I would recommend trying a sample clip with high motion and detailed areas if possible at several bitrates (say 2, 3, 4 Mbps) and see if you can tell the difference.

If you're interested in saving space, you'll also want to compress your audio stream. Going from uncompressed to 224 kbps AC3 saves over 1 Mbps, which can be used to up the video bitrate and/or put more time on the DVD. Look for guides elsewhere on this site.

You may also want to try some filtering or other restoration tricks to try to clean up the video before converting it to MPEG2. I know even less about this, so won't comment. There is an article on Tom's Hardware that goes into some of this, and I'm sure there's other good info on this site.

[SoupStain]

I completely agree with LeTrout. For VHS conversions I always use half-D1 resolution with 2-pass encoding using Tmpgenc.

If there's not much movement (eg some old studio-based TV show or theatre performance) I can usually get away with an average video bitrate of only 2000, with a maximum setting of only 4000.

For home videos, where there tends to be more movement (shaky camerawork in my case), I often use average bitrate of 3000 with a maximum of 6000 - again, two-pass, half-D1. At this setting I rarely see blocky artifacts (if I do, I just re-encode at 4000 and 8000).

My rule of thumb is to set the maximum bitrate to twice the average.

[donpedro]

I wouldn't use 9000 since some players might have problem to play it. Try to find more info about my steps in link in my signature.

[DJRumpy]

All dvd playes are required to support a maximum bitrate of 9.8Mb/s for MPEG-2, meaning you can set the Kbps to as high as 10,035 without issue. They are required by the specification to handle bitrates as high as these. If your encoder does not comply with your MAX bitrate setting, I'd suggest you use a different encoder. CCE follows the max very well. No idea how TMPGenc stacks up. I've never analized an encoded stream from that encoder.

[donpedro]

All dvd playes are required to support a maximum bitrate of 9.8Mb/s for MPEG-2, meaning you can set the Kbps to as high as 10,035 without issue. They are required by the specification to handle bitrates as high as these.

Nice theory... You should talk to my cheap Mintek But I doubt it would listen to you..

[DJRumpy]

Not theory. It's the standard. It would be rather foolish of a manufacturer to sell a player that couldn't play any standard DVD. If your player truely doesn't support it, I'd suggest a different brand was in order.

Rather than blaming your player, I would suspect a bad (read spikey) encode, before I suspected your player didn't support it.

I've read that TMPGenc is very bad about honering MAX bitrate settings, although I have never verified myself (I use CCE). CCE stays within the limits I set. You can verify your MPEG output with DVTool.

[donpedro]

Rather than blaming your player, I would suspect a bad (read spikey) encode, before I suspected your player didn't support it.

Recording was done with Panasonic DMR-E30

You can verify your MPEG output with DVTool.

I need help with that. How do I get some graf with v0.53 that is available for download ?

[DJRumpy]

Senior Moment...{*Doh!*}

It's Bitrate Viewer, not DV Tool...

[donpedro]

It's Bitrate Viewer

Any help with that ? I have version 1.5.054. What should I setup to see graf correctly ?

[DJRumpy]

Just open your MPEG, and allow the software to parse completely through the mpeg (may take a while). Note the PEAK value used:

[LeTrout]

Interesting, thanks DJRumpy I'll have to give that a try. I use TMPGEnc and though I haven't had a problem yet, I should check to see if I'm pushing things with my 9Mbps max setting (with compressed audio of course). Suppose I should just turn it down to be safe, though some of my (DV) video is fairly high motion and I'd like to max the quality.

[vhelp]

.
.
Some $0.02 cents worth on DV CAM's. . .

on the talk of CAM'ing around. Might I suggest that those of you who CAM,
look into a tripod, and learn how to use it well. I mean, I've ben experimenting
w/ several different makes and what I have found was curious to my eyes
after severl exploritory test encodes.

Movements...
* All those panning around (by hand) makes for ingrediants for plenty of
...blocks or Higher bitrate requirements.
* built-in stabalizers are ok as long as you don't use any zooming.. but once
...you go zooming, that it becomes pretty useless. You'll notice it when you
...encode your video (see your bitrate sky-rocket) (or, see more of your blocks)
* Leran the art of Tripoding. I'm still learning, but it not just "place your cam
...on a tripd, and all is well" bit. You also have to move (pan) left/right
...up/down your cam on this tripod, and you have to do it "fluidly". Any
...suddne jerks of the tripod arm, and its just as bad at times.
* Best tripod are fluid-filled, but these are VERY expensive, I found out.

None-the-less, tripoding, over time well spend, becomes an art (or close to
it) and once you get the "feel" for it, you'll see the results (benefits) in your
encodes (ie, lower bitrates/higher quality etc)

Ever since I started using a tripd in most of my footage, I've noticed a dramatic
increase in quality (or bitrate distribution) And, the more I spend time w/
learning the panning movements, the better my results get.

An Invention in process. . .
I am currently working on a gizmo (home-made tripod system) that will be
somewhat fluid-like. I've sketched a few drawings of various models and
designs, but I'm still making changes to the drawing I guess, I can't
make up my mind.. anyways..
Someday, I'll post a pic of it here.. if I ever get it completed
So, do stay tuned !!

That's about it for now,
-vhelp

[DJRumpy]

vhelp, check out something called a 'steady cam' (at least I think that's what it's called. Search Google...). The movie industry uses it. It allows a person to carry the camera, but they get no shakes. Of course a movie camera is much heavier than a home cam, but the principle should be the same.

[LeTrout]

I don't have the steadiest hand in the world and I try to use a tripod or monopod where possible, but my bitrate needs come mainly from my subject moving, not me

[DJRumpy]

That's where MPEG has it's strength. If it can properly detect that movement from frame to frame, your bitrate needs shouldn't increase dramatically. You would benefit most from multipass VBR.

[vhelp]

@ DJ,

yeah, I agree w/ you. But..
In all my many studies w/ DV encoding, I have found that the main cause
of constant bitrate spike (or very high bitrates) is because people are encoding
w/ some form of de-interlace (of any sort) And, because of this, the temporal
is lost or too much of it is, and the first thing that suffers is bitrate.

So, in my DV encoding expeirence thus far, encoding w/ the Interlace option
checked yields lesser requirements on bitrates.

I'm still playing around w/ converting DV down to:
* 23.976 Film - Progressive 'ness
* 16:9 output

..but it's proving to be a pain in the A-Butt. I can manage to get it to not
jidder much, if not smooth during 95 percent of the time, but this still needs
work or fine-tuning. I think I'm getting their, but in the end, I know it wont
be as good as Film (or even smoothness) of which I understand the reasons
why
The fealing is great when you can take the same clip that took 23mb and
encode it as such, and it now only occopies -/+ 12mb.
This can only be achieved by 16:9'ing the source. However. I found out
that because of the way these cams 16:9 (or strunch) the source, it leaves
behind residues or artifacts, which also add to the bitrate spikes. So, I
learned a new method of 16:9 process - - film at 4:3, then:
* crop by 60/60
* resize to 720 x 480
* encode 16:9

..if all done correctly, you shold have the same ratio (4:3 or 16:9) when
you play them on your pc, using the right sw player to play 16:9, of which
powerdvd is the only one you can test this theory with, and work. Anywasy..

I can meat this space saving at least halfway their w/ good results, but that
would require the source to still be pure Interlaced and smooth motion.. but
I'm making some headway, I think.

I believe that the key to a successful smoothness of these lower fps's is to
limit the cam's jerks and fast pannings.. as these are just not smooth when
done in the hands of a human.
Also, I've ben doing lots of studying on the "stabilization" and "lighting"
and I have learned that even Cinema cams have problems w/ motion, of
which I did not know this.. strobbing in those higher models equipments.
And that it's worse for 24p due to the lower fps. Now I realized why they
use those strange cam arms and why they are so heavy - - so there
is no vibrations and wobbles etc etc. I've learned this myself (after some
deducements on my own) when I did some elaborate/strange camming w/
strange setups (out of curiosity mainly) But, now I have a better idea, and
I am now aiming at these factors as well

@ DJ,
I've heared of this steady cam, and there are many that claims "steady"
but I've never actually tried one yet.

However, DV caming has become an new, fun and endeavor for me. I can't
seem to put my cam down anymores. I always have it w/ me or in my bag
somwhere, and I even now have mini tripods that I can carry in w/ me, not
to mention even a tiny 3" one too, of which does a great job on a table :P
And, one at work on my desk, and can extent full hight.

-vhelp

[DJRumpy]

I agree. It should be left as interlaced. I never deinterlace a video. period. It's ugly and destructive to the video.

I can only suggest that you use the Alternate Scanning Order in your encoder to properly handle, detect, and encode interlaced signals. Other than that, your somewhat at the mercy of your encoder

[ronnylov]

I have experimented a bit with DynaPel Stedyhand (http://www.dynapel.com/com/private/sh_overview.htm) amd sometimes it can make wonders with a shaky scene.

I have also bought myself a "single legged tripod" which is a telescopic stick that I use as camcorder support. It's much easier to transport than a 3-legged tripod and it still makes the video more stable.

Regarding converting old VHS home movies I use quite high bitrate becuse in my opinion the movies are more worth than DVD-R discs. I use CBR encoding and 8 mbit/s video and uncompressed audio. I can still fit one hour per disc. Regarding resolution I found that I gained nothing by going down to halfD1. It seems that the details where lost anyway in the source VHS so the encoder could encode full resolution almost as easy as halfD1. But I suggest you try both alterntives and choose what you like most. I preferred the smaller pixels when using full resolution.

[DJRumpy]

I agree. When using DVD, it's not necessary to use Half-D1. The difference is very visible on an HDTV. Why CBR though? You would be better served by VBR, even with a high average bitrate. The extra passes would only improve your encode (assuming you used multipass).

Even CQ mode with comparible max bitrates would give you the same quality as CBR, probably reduce the space required (very unlikely that your video would require 8Mbps throughout the entire video ), and still only require a single pass.

[SMHarman]

Experiment.

Buy a DVD-RW and then you can experiment with D1 1/2 D1 and different Bit rates to get what you feel is an acceptable trade off between space utilisation and quality.

I've yet to go down this route, but other posters have noted that although 1/2D is quivelant to VHS you can see the difference. They eye percieves analogue errors much differently to digital ones.

I thought that the max bit rate was 9800 combined for audio and video. Or is this for the MP2 stream. If your audio is PCM then your video can be 9000+ MP2

Personally I max at 8000 Video and that leaves 1500 for PCM audio and a little spare overhead.

[ronnylov]

It seemed that 8000 kbit/s video bitrate was close to maximum bitrate for my DVD-player, especially when using PCM audio so I thought that doing VBR with min and max equal to the average bitrate would not gain anything. But some people say that VBR helps motion compensation but other say it is worse than CBR. I can't see the difference and it is faster with only one pass. I use VBR when I use a bitrate below maximum, like 6000 kbit/s average, 8000 kbit/s maximum and 4000 kbit/s minimum. I know there have been many discussions regarding VBR/CBR at max bitrate. I haven't done any testing yet in this matter so I actually don't know if VBR with min=max=average bitrate is better than CBR or not. I was satisfied with the results of 8Mbit/s CBR.

By the way I used Mainconcept encoder 1.4 and in my opinion it gave better results than tmpgencplus, cce or canopus procoder. So I recommend that you try this encoder and compare it with your favorite encoder.

[DJRumpy]

VBR is the standard for DVD. Examine any VOB off of any movie DVD you buy, and you'll find that none of them use CBR. It's inefficient, except for speed. CBR offers no benefits in regards to quality, that VBR cannot meet. On the older formats like VCD, CVD, and SVCD, the benefits of VBR were somewhat limited to just a smaller encoded file, since CBR typically had the person using the MAX bitrate for the format regardless. VBR didn't shine here, simply becuase it didn't have room to stretch it's legs, so to speak. A 2524 CBR SVCD looked the same as a VBR SVCD with appropriate AVG/Min/Max settings. On DVD however, you can set VBR to the full max available to the format (10,035 kbps), while still placing hours of video on the disc. If you tried that with CBR, you'd find out very quickly that using a bitrate that high would severely limit the amount of video you could place on the disc. The other benefit of VBR has to do with bitrate allocation. When multipass is used, the encoder can also insert an I-Frame when necessary to better handle scene changes and fast motion. This helps to prevent macroblocking in problem areas. It also has the added benefit of using bit allocation at the macroblock level, rather than the frame level. CBR technology just can't match it.

If your doing CBR simply due to speed, then consider CQ mode encoding. It also does only one pass while saving space, and without sacrificing any quality. Just set your MAX to 8000 (or whatever you like, as your encoder must support the 10035kbps max to be DVD compliant..just remember to include the audio kbps in that value..use AC3 for best effect). If the video requires 8000kbps throughout the entire video, then that's what CQ mode will apply. It only drops the bitrate when the video doesn't require it. You don't gain the benefits of the macroblock bitrate allocation, or controlled i-frame insertion, but you will gain the benefit of much more space.

[ronnylov]

But 10035 kbit/s is maximum for video + audio + subtitles, isn't it? And there will also be bitrate peaks above the maximum set value, at least when encoding video with tmpgenc. I used 8000 kbit/s video + 1536 kbit/s audio. This leaves some margin of around 500 kbit/s for bitrate peaks (CBR is not exactly constant). If I encode with TMPGEnc and use PCM audio I can't use higher video bitrate than around 8000 kbit/s, otherwise the DVD authoring software complains about non compliance and then I may have to reencode the video or compress the audio. I could have chosen something like 9000 kbit/s video and compressed audio but in case I need to edit the audio later I prefer uncompressed audio. I am making some kind of mastering copy as a backup of the original tapes.

As I said I was encoding at the maximum bitrate of the DVD format just because I don't care if I need more DVD-R discs. I agree that it is definately better to use VBR when you need to fit more than one hour on one DVD-R. In this case one hour per disc is OK so I used CBR at max bitrate of the DVD format (with some margins). But I may consider a switch to VBR if it gives better quality. It's worth investigating further. Thanks for the explanation!

[DJRumpy]

TMPGenc is bad about bitrate spikes. CCE complies with the maximum set with no issue. I haven't heard much about Main Concept. Maybe someone who uses it can comment. In either case, your player should handle the occasional bitrate spike without issue. Too many within a GOP could cause a lurch. If your encoder doesn't behave, then play it safe, and stay low.

You bring up an interesting question about subtitles. I'd be curious to know how much bitrate they take up.

Regarding your audio issue. You might do what I do. Use a CD-R to store your original PCM audio. CD-R's cost pennies. If you ever need to re-edit, you have your pure source, without sacrificing all the space it takes to keep it on DVD.

Just a suggestion...

[tito13kfm]

Regarding your audio issue. You might do what I do. Use a CD-R to store your original PCM audio. CD-R's cost pennies. If you ever need to re-edit, you have your pure source, without sacrificing all the space it takes to keep it on DVD.

Just a suggestion...

Holy crap that's so simple it might just work!!!

I've been encoding to mp2 audio for 2 reasons, No dual layer burnable DVD's yet and second I can't tell the difference when playing them back on my crappy tv speakers.

But I've always been concerned with the whatif's. What if something happens to my original dvd's (dvd rot, theft, I get extremely poor and have to sell them(yes I know that would be illeagal to keep the copies, but I swear I'd buy them back eventually)) What if I get a DVD-9 burner? What if I upgrade to a 5.1 dolby surround Bose audio blow the windows out system?

Now I'll just keep the original PCM audio on a cd-r in the same box with the DVD backup I create.

Thank you again for the idea.

[ronnylov]

One subtitle stream need around 10 kbit/s in average, so it's not that much. Great idea regarding usuing a CD-R to store the audio. I think I'll try that. I can fit one our uncompressed wav 48 kHz stereo as wav file on a CD-R! Thanks! With compressed audio I can use higher video bitrate.

Regarding mainconcept encoder 1.4 I have not tested the bitrate peaks. But the more I use the encoder the more happy I am. It gives better quality than TMPGEnc in my opionion. Yesterday I tried it on Peral Harbor, 3 hours on a DVD-R with full resolution and still very good quality! The 2-pass VBR option gave a very accurate filesize, almost as accurate as CCE.