Uncompressed AVI to MPEG2 VBR.. does shaky camera=motion?

[Rudyard]

I am still a little unsure on how to compress my uncompressed AVI file to MPEG2 for DVD burning. It is all old VHS footage and taken on super 8 and old video cameras by mum/dad/family etc resulting in general home movie type shaking and camera work.

Now my understanding is using a VBR I select an average, a minimum and a maximum bitrate which will use maximum for high motion sections and minimum for low motion sections (correct?)

Why I am unsure is if it is general home movie footage and contains some shaky camera work is that the same as motion? i.e. It needs high bitrate? And if there is a lot of motion in a particular section will VBR encode with the higher bitrate EVERYWHERE it is needed or to meet the average will it drop the bitrate in some high motion scenes if there are to many of them and thus losing quality?

I was thinking of using 4,000 min 6,000 avg 8,000 max but am not sure if that is to much or to little or just right for preserving our family movies.

Thanks for any help in clearing this up.

[guns1inger]

Any camera movement is motion. Shakey camera movement is worse than a smooth pan because the motion is unpredictable and usually happens for a much longer period - often the whole tape. The general recommendation here is to use high bitrate (approx 8400 kbps) CBR methods to encode this type of material, and restrict the footage to 60 - 70 minutes per disc

[daamon]

And if there is a lot of motion in a particular section will VBR encode with the higher bitrate EVERYWHERE it is needed or to meet the average will it drop the bitrate in some high motion scenes if there are to many of them and thus losing quality?

I was thinking of using 4,000 min 6,000 avg 8,000 max but am not sure if that is to much or to little or just right for preserving our family movies.

You're basically right in your understanding of how VBR works - high bitrate where needed, low bitrate if it can get away with it, and aiming to meet the average. But your interpretation of this is not quite right - as demonstrated by your choice of the minimum. It's far too high.

But hey, when I first started using VBR, I made a similar mistake - I assumed that a min of (coincidentally) 4,000 would be good as it's still a decent bitrate. The problem is, a figure like this has the opposite of the desired effect. Here's why...

* If a scene only needs 500kbps (totally black during a transition, for example), or needs very little at all (1,000 or 2,000kbps) it will be forced to use 4,000.

* There are going to be numerous scenes that could well need less than 4,000 - but they'll be forced to use 4,000. To meet the average, by default, scenes that need higher bitrates won't be allocated as much as they could need - resulting in lower quality, usually in the form of macroblocks.

* In short, too high a "min" will place a strangle on the max bitrate allocated.

So, now understanding this principle, I would recommend a min of 500kbps and a max of the 8,000 you've got.

I hope that helps. Good luck...

[zoobie]

you may want to use vDub's deshaker to see your other options

[Rudyard]

Thanks for all replies, I will and do do tests aswell as bugging forum members but I just like to try and understand what is going on!

Daamon if I could just clarify something: Say my average is 6,000 Min500 Max 8,000 and I have say 100 minutes for arguments sake.

70 minutes are high motion that require the max and 30 mins are slow which can handle min - average bit rates.

Will part of the 70mins that require high bitrates be reduced to meet the average or will it give all the 70minutes the max bitrate it needs and the average just works out a bit higher?

Thanks again for your help.

[daamon]

you may want to use vDub's deshaker to see your other options

It depends on how shaky the video is - In the days before I used (discovered) tripods, my home video was full of wobbly pans and bad zooms. I encoded this using CBR at 8,000 with WAV audio at 1,536 and it came out very nicely.

As the source is VHS, the source resolution is much lower than full D1 (720 x 480 / 576 for NTSC / PAL) and so a comparable resolution can be used during encoding (perhaps half-D1 - 352 x 480 / 576), which requires half the bitrate.

This means that an encode at half-D1 and 5 - 6,000kbps could actually yield better results, albeit a slightly less sharp picture than a D1 DVD. But that doesn't matter coz the picture wasn't sharp in the first place.

Something to look into Rudyard...

[Rudyard]

The first response suggest that Daamon aswell, I will try that and see. I have also seen many conflicting opinions regarding encoding at VHS resolution.

When there is no yes/no answer (depends on footage etc) I guess I will just have to keep testing and see if I can notice any differences.
Thanks for helping try to understand what exactly is going on in the whole process!

[daamon]

Daamon if I could just clarify something: Say my average is 6,000 Min500 Max 8,000 and I have say 100 minutes for arguments sake.

70 minutes are high motion that require the max and 30 mins are slow which can handle min - average bit rates.

Will part of the 70mins that require high bitrates be reduced to meet the average or will it give all the 70minutes the max bitrate it needs and the average just works out a bit higher?

Assuming those 70 minutes all require the same bitrate, then they'll all get the (same) highest bitrate they can, given the constraints the encoder is working under. It may or may not be the actual max that you set - that just limits the peak.

This, of course, isn't going to be the case. But it's a good point to start from... Each part of the video will require different amounts - it's the relative requirements that are met, not the actual.

So let's say one scene needs 8,400, and another needs 8,000 (numbers picked for ease of use), then the former needs 5% more than the latter. So, if the former is allocated (say) 7,350 then the latter will be given (around - it's not an exact science) 7,000.

The average is the overall driver - it won't be met exactly, but it'll be pretty close. The variance between the lows and the highs (limited by the mins and maxs) is what will change more, in line with what I've just described.

Clear as mud? I hope not... This link may help - scroll down about a 1/4 of the way, and look in the bit describing "Rate control mode 2-pass VBR".

I hope that helps - if not, feel free to ask more...

[daamon]

The first response suggest that Daamon aswell, I will try that and see. I have also seen many conflicting opinions regarding encoding at VHS resolution.

The 5 - 6,000 I mentioned was thinking CBR - it depends on the length of your footage. Use the VideoHelp Bitrate Calculator and this link to help you figure things out.

When there is no yes/no answer (depends on footage etc) I guess I will just have to keep testing and see if I can notice any differences.

Welcome to the unspecific and vague dark art of video...

Thanks for helping try to understand what exactly is going on in the whole process!

No problem - it's a pleasure and, sick I know, fun...

[guns1inger]

Something else to consider when using VBR. The encoder will try to give you a video with an average bitrate near or on the bitrate you set for the average. Therefore it must, ultimately, balance high and low bitrates to get there. This is just simple math. The closer the average bitrate is to the max bitrate, the less opportunity there is for the encoder to use the lower bitrates. An ideal setting has the minimum and maximum evenly distances above and below the average. This sets sensible bounds for the encoder. You can use 500/6500/8000, but the encoder wont be able to use much below about 4500 - 5000 very often, or the distribution will be thrown off. That is not to say you can't use wildy open min and max figures, but they may not be used. This is also another arguement for 2 or more pass VBR over single pass VBR. Given free reign with the lower numbers for a single pass encode, the encoder may well under allocate, giving smaller file sizes and lower quality. The 2 second pass allows for a better balance of distribution of numbers.

[daamon]

Sorry, Rudyard. guns1inger makes a very good point, and it is worth stating that, all along, when I've been talking about VBR I was meaning 2-pass VBR - although I never actually said that... D'Oh!

[FulciLives]

I almost neve set the MIN below 2000kbps when doing a 2-pass VBR encode and in the case of using an AVG of 6000kbps - 6500kbps I would probably use a MIN of 2500kbps

I don't like setting the MIN to a very low level as advocated in this thread.

- John "FulciLives" Coleman

[daamon]

I don't like setting the MIN to a very low level as advocated in this thread.

I'm not saying you're wrong, or that I disagree, I'm just curious to hear your reasoning.

My reasoning on going as low as 500 is solely logical, and not based on anything technical. It's also supported by my results - but each person's circumstances (source files, different encoder etc.) are different.

In the end, it's whatever works for you - then that's the best thing.

[FulciLives]

The 2000kbps "lower limit" is something that people have used for a long time for a variety of reasons.

One reason given is that most encoders don't do well at a lower bitrate with Full D1 MPEG-2.

One reason is that encoders are "set up" this way. For instance TMPGEnc Plus auto sets the MIN and MAX to 2000kbps and 8000kbps and one would think that was done for a reason.

Another reason is that you would never want the encoder to go lower than that anyways because it just wouldn't look good ... I don't care if it is a solid black screen or not LOL

Perhaps it is a bit subjective but I've never seen a commercial DVD use bitrates as low as 2000kbps

- John "FulciLives" Coleman

[daamon]

That all sounds reasonable to me... Cheers for the info.

[tarrickb]

So all this talk about min bitrates and we haven't touched on max bitrates!
if your average can approach something like 7500, is it recommended that the max be upped to 9800 from 8000?

At what point should the max be increased?

[tarrickb]

another point of possible intrest is that Sony Vegas's default DVD template uses 192,000bps/6000avg/8000max.........so the min differs quite significantly from TMPEnc

[FulciLives]

So all this talk about min bitrates and we haven't touched on max bitrates!
if your average can approach something like 7500, is it recommended that the max be upped to 9800 from 8000?

At what point should the max be increased?

No point in doing a 2-pass VBR if the AVG is going to be 7500kbps ... might as well just do a CBR encode then.

- John "FulciLives" Coleman

[tarrickb]

any disadvantages to using VBR in this scenario, other than wasting time...

[tarrickb]

do you guys think i should just maintain the ratio...
So if in Vegas i wanted to up the bitrate by 1500kbps
I bring all the values (min/avg/max) up by 1500...???

[FulciLives]

do you guys think i should just maintain the ratio...
So if in Vegas i wanted to up the bitrate by 1500kbps
I bring all the values (min/avg/max) up by 1500...???

That doesn't make any sense.

Don't do that.

It's best in a "safe" kind of way to limit the MAX video bitrate to no higher than 8000kbps.

Now if you are using LPCM WAV audio you definately should not cannot push the MAX but if you are using MP2 or AC-3 audio then you could push the MAX video bitrate from 8000kbps to 9000kbps but no more than that for sure.

The AVG of course is based on the running time of the video you are encoding. So you have to adjust that accordingly.

The MIN is the one spot where most people tend to disagree as to what to set it to.

- John "FulciLives" Coleman

[tarrickb]

what am i being "safe" about capping at 8000 (or 9000)?

[tarrickb]

what am i being "safe" about capping at 8000 (or 9000)?

I'll answer my own question and say compatability....correct?

[FulciLives]

what am i being "safe" about capping at 8000 (or 9000)?

There is a DVD spec and the DVD spec has a strict limit on the bitrate used. You can only go so high before you go out-of-spec.

Even if you get really close to but still under the spec you might encounter playback issues on some DVD players ... a lot of this also has to do with the quality of the blank media and the quality of the burn etc.

LPCM WAV audio has a fixed bitrate of approximately 1536kbps so when you use that with a MAX bitrate of 8000kbps you got a total of 9536kbps but the MAX bitrate a DVD can handle (this includes video, audio and subtitle streams) is approximately 9800kbps so 9536kbps is close enough.

However I say you can use a video bitrate of 9000kbps with MP2 or AC-3 audio because the highest bitrate MP2 can use is 384kbps and AC-3 can use no more than 448kbps so again 9000kbps + 448kbps = 9448kbps which again is close enough to the MAX a DVD can handle.

- John "FulciLives" Coleman

[Rudyard]

Another quick question , didnt want to start a new thread as this one should fit...... when converting captured sport programs would I be correct in thinking a CBR is better as there is to much movement for VBR (im talking rugby league, so there is constant player movement, for those in the US picture NFL but without the stopage between each play).

I wanted to put some games onto DVD and dont think there will be enough still moments for VBR to be effective, unless the half time break adds up to be enough low motion?

Would CBR be the usual way to encode fast moving sports games?

[daamon]

If the length of the footage is short enough, then CBR can be used anyway - this can be calculated from the VideoHelp Bitrate Calculator, assuming you don't want to go below around 5,000 for the video.

You'll get more video at CBR if you use compressed audio (AC3 or MP2) at around 192kbps.

If it's longer than the running time you figure out above, then VBR is the better option. Or CBR and two discs, or more...

As it's rugby league, you're looking at 40 mins per half, plus half time and extra time? That'll add up to about 90 mins per game.

[vhelp]

Evening everyone.

Hand held cams has a tremendous amount of motion in them. They are usually
in the form of up/down/left/right jittery motion of the human arm and hand.
And, dispite the cams' on-board stabalizers, it is not enough for the final
MPEG process. Also, it is a safe assumption that home footage has various
scenarios of scenes with lots of user zooming around. This adds to the
MPEG's inability to compress well and still produce quality.

When I shoot video, and based off of lots of practice and experimentation,
etc., I use the following basic receipe:

1 - Videos: break your projects into 1hr videos
2 - MPEG: encode each 1hr segment using your encoders CBR mode and 9000 bitrate
3 - Audio: can use AC3 or MP2 (I use MP2 @ 192k) for starters

Some things to note regarding Encoding, Bitrate and Audio..

One hours worth of video will fit onto 1 -r/+r disk, with some possible
room left over for other things like menues, etc.

Also, you can experiment a little with the audio. But, since your audio is
coming from a low-budet mic to begin with, I see no point in maintaining it at
*suggested* higher bitrates, unless there is no sacrifice to the video project,
overall.

-vhelp 3859