Which bitrate when re-encoding mpeg2 material and minimizing quality loss?

[newuserxyz]

Hello!

What's a good way to re-encode mpeg2 material?

For example after resizing some mpeg2 videos I've to re-encode the footage again to mpeg2.
My source material has a bitrate around 5000 kbit/s.

What bitrate should i use for the re-encoding when I want to minimize the quality loss?

1) Nearly the same bitrate(something like max 6000 kbit/s and min 5000 kbit/s)
-> because encoding at higher bitrate than the original material won't improve the quality

2) A very high bitrate (max 8500 kbit/s and min 7000 kbit/s)
-> because each re-encoding produces quality loss and I preferably want to keep the quality of the original material


hhmm?? Thanx for your advises :thanks:

[Guest]

It depends on the resizing you are doing. For example, if you are reducing the size, obviously a smaller bitrate can be used.

Also, your criterion of "minimizing quality loss" is vague.

[newuserxyz]

ok, you are right resizing is a bad example - let's say I cut the original.

what are more concret criterions of quality?

[Ghitulescu]

You can losslessly cut MPEG-2 if cut points are at GOP boundaries. There are also "smart" cutters, that reencode only the end GOPs, should the cut points be within them.

[Lester Burnham]

You can losslessly cut MPEG-2 if cut points are at GOP boundaries. There are also "smart" cutters, that reencode only the end GOPs, should the cut points be within them.

Indeed - and don't have to be payware - Cuttermaran does (think it uses quenc, not sure you can use HC).

[benwaggoner]

And if your source has any artifacts, using a deblocking/deringing decoder will help. Particularly if you're doing resizing as that will change the block alignment of the source so blocking artifacts in the source will become nice sharp lines in the middle of your blocks in the output. And those sharp lines won't move with motion, making motion estimation much harder.

DGDecode with cpu=6 FTW.

[GrofLuigi]

DGDecode with cpu=6 FTW.
Hi Ben,

Is this your personal preference (http://forum.doom9.org/showthread.php?p=559134#post559134) or you have some more information? :thanks:

GL

[newuserxyz]

Hi!

thank you for all replies! :thanks:

I actually don't have the posssibility of smart rendering, because I added some effects and could not make sure to cut only at the GOP boundaries.
In this case which strategy to use - nearly the same bitrate or a very high bitrate?

[Rumbah]

Indeed - and don't have to be payware - Cuttermaran does (think it uses quenc, not sure you can use HC).

You can use HC in Cuttermaran, just look at my signature ;) .

[benwaggoner]

Hi Ben,

Is this your personal preference (http://forum.doom9.org/showthread.php?p=559134#post559134) or you have some more information? :thanks:

Well, I'd say it's my default. cpu=6 applied deblocking and deringing wherever it can, and is reasonably neutrally weighted. So it doesn't tend to oversoften low QP or undersoften high QP.

But I've certainly had awful HD MPEG-2 sources with nasty QP spikes where I further tuned things.

But for the kinds of projects I use it for, I probably stick with cpu=6 at least 90% of the time.

[shmendrapolk]

I have a similar question:

I edit mpeg-2 1080i video using Pinnacle Studio. Because I add a lot of effects and use picture in picture, I often need to encode and export my semi-finished video, then start a new project using the semi-finished video as my source file, and then finalize it and export it again. So the end result is that I often need to reencode my mpeg-2 video twice.
When I'm done I export my final movies in 3 different formats: DIVX high-def (720p), Mpeg-2 for DVD, and Mpeg-2 1080i (for playing on PC & archiving).

So am I losing significant quality by having to re-encode the Mpeg-2 files twice?

[Ghitulescu]

I have a similar question:

I edit mpeg-2 1080i video using Pinnacle Studio. Because I add a lot of effects and use picture in picture, I often need to encode and export my semi-finished video, then start a new project using the semi-finished video as my source file, and then finalize it and export it again. So the end result is that I often need to reencode my mpeg-2 video twice.
When I'm done I export my final movies in 3 different formats: DIVX high-def (720p), Mpeg-2 for DVD, and Mpeg-2 1080i (for playing on PC & archiving).

So am I losing significant quality by having to re-encode the Mpeg-2 files twice?

You can save the intermediate files using a less agressive compression. You can't avoid de/re-compression during editing, but you can avoid multiple reencodings.

[shmendrapolk]

You can save the intermediate files using a less agressive compression. You can't avoid de/re-compression during editing, but you can avoid multiple reencodings.

It seems though that mpeg-2 is the only export format on Pinnacle that support 1080i. The mpeg-4 high def is only 720p and the avi only seems to offer standard definition.

[Ghitulescu]

It seems though that mpeg-2 is the only export format on Pinnacle that support 1080i. The mpeg-4 high def is only 720p and the avi only seems to offer standard definition.

Then use the maximal bitrate you can choose (15Mbps).

[benwaggoner]

Then use the maximal bitrate you can choose (15Mbps).
15 Mbps is the MP@ML max, but I'd think any encoder capable of 1080i would support a lot higher. And would need to!

[Ghitulescu]

15 Mbps is the MP@ML max, but I'd think any encoder capable of 1080i would support a lot higher. And would need to!

Oh, no! I've mixed up the two questions, I thought it was about a MPEG-2 stream of about 5000kbps bitrate.

Of course for native 1080i you need a higher profile to keep the details and minimize the artefacts as much as possible.