Mpeg encoding involving 3rd party products

[cyeary]

I am a big fan of a encoding program called Episode. I have been using this program in my workflow for years. I have noticed lately that the encoding setting i have been using in it are great for playing movies forward but once the movie has played I have to go back to the beginning of the movie to be able to scrub to a certain place. I was wondering if you think that using ALL I-Frames as the encoding setting will do anything harmful to my sync or anything else. The file size seems to be identical to the same file that is encoded with auto from the Coolux encoder.

Thank you in advance.

[malkuth23]

From what I understand, if you only use I-frames you will get a file that is the same size, but you will lose quality.

Because we are using constant bitrate, we only have a certain amount of information per frame. This is the problem with constant bitrate. On a complicated frame we lose quality and on a simple frame we waste space... Of course this is all a trade off to get the steady play rate we all know and love.

By using P and B frames, we can put more information in each frame about the important stuff (the changes from each frame). This makes the quality higher.
You want I-frames at a scene change because everything changes there. You want P frames during more subtle changes.

You will get the best quality from a GOP.

There are a lot of parts of the encoding process I do not fully understand (like why a B frame vs. a P frame or the downsides of Scene Detection in mpeg2), so I hope other people jump in here if I was wrong on any points above.

[Daniel Kaminski]

Hi all,

Matt is right. Since CBR (Constant Bitrate) defines how big your video will be you will loose quality.
I Frame only means that a full Image is written every frame.

There are a lot of books written about the MPEG2 standart, that explain in detail how it works.
I found this short beginners guide. Some of the information is outdated, but the principal of how it is encoded and how information are stored is explained pretty well.

http://www.iem.thm.de/telekom-labor/zin ... ginnzi.htm

I hope this helps.

Daniel

[cyeary]

I guess the best answer would be what is the built in encoder based on then. When I use the Coolux encoder everything is perfect but the batch and audio capabilities of it are VERY LIMITED. Plus you have to have a dongle on each machine to use it. With the program I am using EPISODE the batch functions and audio splitting capabilities are endless and I do not have to have a dongle for each computer. Just a code. So is there a recommended GOP structure? My current settings allow playback fine but once the now pointer has gone past the container i cannot simply go back 30 to 60 frames to get to the end. I have to jump to before the container then play forward into it then go to my outpoint. When you have a show with 40 some odd VT rolls with over 8 channels of audio each roll this is a lot of time. SO What type of GOP format should I use, How many P-frames between I-Frames, how many B-frames between P-frames,Should I force a Frame header with each GOP grow, and finally what should my GOP type be Open or Closed?

[Dennis Kuypers]

Hi,

have a look at this post: viewtopic.php?f=2&t=3239&p=9526#p9524

There is a screenshot of the default encoder settings for the PB encoder. Maybe you can use that as a base.

As for the GOP: I guess that it should rather be closed for performance. Cross-GOP lookups sound kind of expensive...

Regards
Dennis