Morgan M-JPEG 3 Settings

[Teslacuted]

Could someone give me some advice on recommended settings for the morgan M-jpeg 3 codec? I'm capturing at 720X576 on a 2.66P4 so I'm looking for the best possible quality with roughly a 4.5:1 compression ratio. Thanks.

[Teslacuted]

My main problem is that I don't understand the "Fast Integer\Integer\Float" settings so I can't decide which one to use.

I'm not sure about interleaving either but since the option is only to support it or not I'm leaving it on.

Invert Field order to my knowledge will reverse the A&B fields in an interlaced stream (my source is interlaced) but I don't see any point in reversing them so I've left this option off.

Bitrate control sounds like something that would only interest people planning on streaming the finished video so I've left that off also.

I have no idea what the difference is between the 4:2:2 option and the 4:1:1 is, I'm fine with the processor extensions (there's no benefit in turning any of them off is there?) and the force quality option just seems to overide the other quality selection box.

Could someone plese advise me on which options to use for my situation and if possible explain briefly the few things I don't understand?

Thanks,
Teslacuted

[trevlac]

There are bugs in the download version. Search for posts by Sh0dan to find a fixed version. Personally, I stopped using it due to a strange field problem I had. I now only use PIC.

http://www.morgan-multimedia.com/review.htm

Originally posted by Teslacuted
My main problem is that I don't understand the "Fast Integer\Integer\Float" settings so I can't decide which one to use.

If I understand correctly, JPEG compression converts pixel 'spacial' info into 'frequency' info. This is really just running it thru a math formula that works in both directions. Fast Integer\Integer\Float are options that make the forumla more precise (less loss due to rounding). Use the best that is fast enough for you.


I'm not sure about interleaving either but since the option is only to support it or not I'm leaving it on.

Good. Because you will throw out 1/2 the data if you leave it off.


Invert Field order to my knowledge will reverse the A&B fields in an interlaced stream (my source is interlaced) but I don't see any point in reversing them so I've left this option off.

Sounds right, but these types of options don't always do what you'd expect. There are 2 aspects to interlaced. Spatial: Which field is on top. Temporal: Which shows 1st.


I have no idea what the difference is between the 4:2:2 option and the 4:1:1 is

Use 4:2:2 to keep more info. Use 4:1:1 for better compression. This is color subsampling. I'd use 4:2:2.


Hope this helps. ;)

[Teslacuted]

Thanks a bunch, I encoded a video clip in all of the possible options and found that firstly the Float option produces much less block artifacts but was only slightly higher in file size, I conpared the 4:1:1 to the 4:2:2 aswell and found the 4:2:2 was considerably higher in file size but I saw no noticeable difference between identical frames, what effect does the colour subsampling have on the finished video? (I captured in YUY2).

And I know this is officially a different topic but I'm using AviSynth to frameserve the video into CCE and I'm only getting an encoding rate of about 0.67 when with the Big3 method I've seen it break the 2.0 mark. Is there something wrong with my script?

AviSource("Episode 1.avi")
Trim(0, 18423) ++ Trim(22000, 32845)
Crop(8, 0, -0, -3)
LanczosResize (720, 576)

[trevlac]

Originally posted by Teslacuted
what effect does the colour subsampling have on the finished video? (I captured in YUY2).

NTSC DV is subsample at 4:1:1. For that some people complain about bleed of something like red against white. With proper interpolation, I'm not sure this should be an issue. Morgan may do a good job. But that's what you should look for. Red text on a white background (or the other way around) would be a great test.


And I know this is officially a different topic but I'm using AviSynth to frameserve the video into CCE and I'm only getting an encoding rate of about 0.67 when with the Big3 method I've seen it break the 2.0 mark. Is there something wrong with my script?

You script does not do much. You could test by taking away each line to isolate what is slow. I can't say much here. :)