Producing Pal DVD from mpeg2 hd content

[sjj1805]

Do not be overawed by High Definition.
In a nutshell you work with it the same as Standard Definition.
You cut out your unwanted bits, add your titles, insert your picture in picture effects and so on.

The only real difference is that you set your bit rates higher plus of course the screen sizes are larger.

[toni1]

If you just want to have a result then OK
But if you want to get the best out of it, then its good to acces to options.

[toni1]

I think I've got a possible explanation of voice de synchronizing probleme while dow sampling to dvd format

I think that this must be due to VBR methode of arranging data allocation;
the two times I had this type of voice de-synchronisation probleme and blocking it was always in the last part of the video;
Its as if videostudio is pushing more data in this last phase (more compression equal more errors
these errors i got producing the compatible dvd files prior to burning (it has nothing to do with dvd quality)

[toni1]

Would like to have trevor or Ken or other comment please

[Trevor Andrew]

Hi Toni1

You have two sections to your work flow.

1 / capturing and editing HD
2 / downsizing the HD video to fit single layer 4.3 Gb disc.

The first capture/edit step should be no different to capturing standard video.
You are capturing to Mpeg2, your settings should remain the same through-out the edit phase and final render.
This should produce a Hd Mpeg2 file. possibly about 10 Gb per hour.

This is of course too large to fit a standard disc, so we downsize.

We render the Hd file to a standard DVD format.
Only the frame rate and frame size need changing, all other settings should remain the same.

Sample properties:- HD

MPEG files
24 bits, 1920 x 1080, 25 fps
Upper Field First
(MPEG-2), 16:9
Video data rate: 25000 kbps
Audio data rate: 384 kbps
MPEG audio layer 2, 48 KHz, Stereo

Sample properties:- DVD

MPEG files
24 bits, 720 x 576, 25 fps
Upper Field First
(DVD-PAL), 16:9
Video data rate: 8000 kbps
Audio data rate: 384 kbps
MPEG audio layer 2, 48 KHz, Stereo

For one hour of video, This should produce a file under 4.3 Gb .

Toni I assume you have read J Hunters HD tutorial here:-
http://www.hd-video-editing.com/HDTutorial_Page1.html

[toni1]

Thanks Trevor
OK upper field has to be kept upper field through the whole process.
towo more questions :
-first , when producing the pal-dvd compatible file do I have to manually assign the maximum 4.3 giga size or is an automatic function of VS10
- Second what is the answer for my main question concerning data rate type (VBR CBR) . I know its a about pushing more data on a dvd; Does VBR have negative effect ( as to output quality and "readability" on home dvd readers)

Ive been discussing a lot with JCHunter and recognize having learned a lot while exchanging through his high def tutorial with all of participants. its sad that the kind of exchanged information has completely disapeared

[etech6355]

The first capture/edit step should be no different to capturing standard video.
You are capturing to Mpeg2, your settings should remain the same through-out the edit phase and final render.
This should produce a Hd Mpeg2 file. possibly about 10 Gb per hour.

toni1, I hope this is starting to sink in

Isn't most of your HighDefinition Video already de-interlaced? Going by many of your previous posts I would think they are. I suspect you have gigs of deinterlaced hd-mpeg2 on your harddisk or backed up to external usb / dvd's. If you haven't upgraded or changed your workflow then I suspect your still having playback problems and de-interlacing appears to fix some of your playback problems on the computer (I don't know about your projector, in another thread I asked you to connect your cam up to your projector via the component cables but you didn't have a component cable, you left that thread with no response so I guess you never tried this method). If your projector is 1280x720 max resolution but can still accept a 1080i signal just connect it up directly. The projector will change your source video to progressive format (1280x720).

If your having problems in VS going from hd-mpeg2 to sd-mpeg2 then use DV.AVI as an intermediate codec. Nothing wrong with this and you will have frame accurate editing. Going from hd-mpeg2 to SD DV.avi is no loss.

Here is a link to the other previous thread
http://phpbb.ulead.com.tw/EN/viewtopic. ... 786#104786

Your cam shoots 60/50 pictures per second, if you de-interlace the video you only have 30/25 pictures per second. To do this conversion to a lesser amount of pictures requires special software and not easy to perform to retain the original quality and motion.

Playing back the HD material is the most confusing, that's why whatever setup your using you need a starting point to know what the original footage actually look like, therefore the only way to do this is connecting your HD cam up to the TV.

I also had playback problems on my computer and had to upgrade the video card to a HD-Certified card. Some older HDTV's & Projectors have slow response times which you may be calling arti-facts. I think what your seeing is the incapability of you playback devices to properly reproduce the fine high detail in each field that's compressed in the hd-mpeg2 file.

Personally I don't go through all this baloney to get SD video. I put the cam in the down_convert mode, record dv.avi and I'm done with it, or go directly into a dvd recorder. If I have to use the computer then it's hd-mpeg2 to dv.avi. All Interlaced Video.

ps- If you make a hd-wmv video or a divx video from the hd-mpeg2 you still leave it as upper_field_first and let the winodws encoder or the divx encoder perform the de-interlacing. Those formats force you to do this because that's part of their codec. Divx does interlacing but playback is only good on a hd-divx certified player.

[toni1]

etech6355
All my originals films are kept in the original captured upper fiels interlaced format.
I déinterlaced only while producing a unique edited hd file (thinking upon discution on ex hd forum that its better for panning artefacts)

I used that deinterlaced file to produce the sd pal compatible file

this is a complicated issue and I know your opinion since we have already discussed that topic and will try to connect directly my cam to display (to see since I havent done it yet)

Interlacing , panning, fast movement artefacts are quet common in hd cams (maybe less with newer models) and will testing trevors, Kens yours.

I still have the suspended above question I asked Trevor about , concerning compression type (CBR VBR) which of both is the most likely to keep quality . which of both is the less likekly to produce compression bugs and which of both is better for old sd dvd readers

[etech6355]

That's great you archive the original recordings and don't have to re-capture again.
You may be surprised to hear this. My HC3 has better motion than my other SD DV Cams. Two of my dv cams are good ones. A nice feature on the HC3 is you can pause & step the video, take a snapshot and transfer it to the memory card. At least on my HC3 stepping frame by frame is very clear, almost looks frame_based.

We performed a motion test on the HC3 after first getting it. We mounted it on the dashboard of the car, then drove down some winding roads and fast turns, not to mention some bumping.
The HD mpeg2 is very good and great motion. The HD -> SD is excellent with the motion. It is nice.

You can browse through the tutorials and previous posts to answer your questions on encoding.
If you encode 8000vbr or 8000cbr with mpeg/dolby audio it's going to look good. I encode almost to the highest setting for playing on my players at home. If I'm making a family dvd and not sure of the players then 6500 to 7000 is good.
If you use dv there shouldn't be any sound sync issues. I don't think mpeg conversions are always reliable.

[Trevor Andrew]

Hi Toni1

Constant or Variable bit rate.

Well I use Constant for all my work, usually at 6000 Kbps.

Which one is the best? is very debatable.

If your video is about one hour then it will fit to disc at the highest bit rate (8000)
Variable bit rate will have little effect, Constant should be used.

Be aware that when using Variable we are using an Average bit rate
When using Constant we are referring to a Maximum bit rate.

Therefore Constant 6000 is a higher rate than Variable 6000.
Variable will produce a smaller file.

Just my thoughts.

As I said I use Constant.

Regards

Trevor

[daniel]

Variable rate was introduced to have a smaller file size by reducing the bitrate for "easy" GOPs.

In Videostudio the bitrate you select is maximum (thus never higher than the corresponding CBR) with a VERY close average and a much lower minimum. So the net result is about the same quality with a smaller file (even smaller with dual-pass).

The caveats are 1. it is slower to encode and 2. nothing is perfect below heaven including the algorithms and sometimes it will for a split second fail to identify a glitch due to lower bitrate.
If you see it or not is a very hot debate. You'll get about as much advices are there are users in this forum.

It all depends on the contents of your video, a motionless speaker on a static background or a cartoon will show dramatic reduction in size, allowing for a much higher max bitrate than CBR. A sports video or one with numerous "cut" changes as a rock concert will be degraded with little size benefits.

Use high CBR as long as you have the space. If you are tight and need to go below 6000, try VBR at 6500-7000 and see if you win enough space.

An exception is stills/slide shows where it would be stupid to use CBR to encode a clip that is essentially encoded as strings of high-bitrate-void B frames. This is filling disc space with useless and redundant data.
Zero is as good as 0 as it is with 00000000.

[GeorgeW]

Variable rate was introduced to have a smaller file size by reducing the bitrate for "easy" GOPs.

In Videostudio the bitrate you select is maximum (thus never higher than the corresponding CBR) with a VERY close average and a much lower minimum. So the net result is about the same quality with a smaller file (even smaller with dual-pass).

The caveats are 1. it is slower to encode and 2. nothing is perfect below heaven including the algorithms and sometimes it will for a split second fail to identify a glitch due to lower bitrate.
If you see it or not is a very hot debate. You'll get about as much advices are there are users in this forum.

It all depends on the contents of your video, a motionless speaker on a static background or a cartoon will show dramatic reduction in size, allowing for a much higher max bitrate than CBR. A sports video or one with numerous "cut" changes as a rock concert will be degraded with little size benefits.

Use high CBR as long as you have the space. If you are tight and need to go below 6000, try VBR at 6500-7000 and see if you win enough space.

An exception is stills/slide shows where it would be stupid to use CBR to encode a clip that is essentially encoded as strings of high-bitrate-void B frames. This is filling disc space with useless and redundant data.
Zero is as good as 0 as it is with 00000000.

I recently wrote the following in the MovieFactory forum concerning VBR. If the encoder adheres to the Average bitrate, then regardless of number of passes, it should produce a file roughly the same size as a single-pass. Are you seeing noticeably smaller files with 2-pass VBR encoding within VS10

I noticed that there's a common belief that VBR encodes automatically means "smaller" file. While this is TRUE in ULEAD software, it is NOT true with all software (outside of Ulead).

MPEG encoders usually allow for more variables for VBR encoding (i.e. MINIMUM, AVERAGE, and MAXIMUM bitrates). Ulead, however, does not allow you to enter all 3 (most of their out-of-the-box software). The bitrate you enter is actually the MAX bitrate. So, that is why you get a smaller file using VBR 6000kbps vs. CBR 6000kbps in ULEAD -- because the bitrate you are specifying is the MAX, and the AVG is actually slightly lower than 6000kbps (in this case).

But in other software, you are able to specify all three, or if only one, then that software might treat that as the AVERAGE bitrate -- so the file sizes would be about the same as a CBR encode at the same VBR encode. For instance, AVG VBR of 6000kbps should be about the same size as CBR 6000kbps (regardless of number of passes -- as long as the encoder maintains the specified AVERAGE bitrate for the encoding).

I only mention this in case users here decide to try other software, and automatically assume VBR = smaller file sizes (vs. CBR at the same bitrate). You have to know how the other software treats VBR encodes, and what parameters you are able to enter for the VBR encode.

Regards,
George

[daniel]

I recently wrote the following in the MovieFactory forum concerning VBR. If the encoder adheres to the Average bitrate, then regardless of number of passes, it should produce a file roughly the same size as a single-pass. Are you seeing noticeably smaller files with 2-pass VBR encoding within VS10 :?:

Smaller Yes.
Noticeably, well Windows tells you the file size so you notice it...

I am one of those that by default use VBR because 75% of what I do is shows from photographs, where it is "mandatory" if you don't get DVDs for free. Since this is my default setting I use it for video also.
In both cases for high bitrates (I use 6000 7000 or 8000 depending on final size) the MPG is smaller in dual-pass. Is it worth double the time?
At some point I decided no for video, yes for slide shows.

I must add that I render at night so time is of no concern (the PC must stay on until the morning anyway). If I had to wait I would not use dual.

Please note that since the second pass makes a comparison with the first, probably by GOP, quality may be affected too. I mean the bitrate for a given GOP could be the same but the result more similar to input. I don't know but they definitely check lots of things between the two.

The only conclusion I see is that the posted Average is a target but not truly respected if lower values reach the target quality.

Edit: I also had rare cases where the 2-pass was heavier than the 1-pass... But usually it was about 3-5% smaller.

[GeorgeW]

Edit: I also had rare cases where the 2-pass was heavier than the 1-pass... But usually it was about 3-5% smaller.

When you say "heavier" -- does that mean larger

Regards,
George

[daniel]

Edit: I also had rare cases where the 2-pass was heavier than the 1-pass... But usually it was about 3-5% smaller.

When you say "heavier" -- does that mean larger :?:

Regards,
George

Yes, that's what I meant, as in bigger :)
I was surprised but it shows how all this size and compression thing is immensely dependent on the contents of the video.

I should add this happened rarely and always on short videos (5 min or so...).
That's why I thought they can be larger thus maybe they are better?
But comparing two 5 minutes videos looking for differences frame by frame was not my cup of tea... I leave that to retired people..