on the fly mpeg1??

[disastermaster]

Could anyone tell me which software prog will allow me to capture mpeg1 (vcd pal) on the the fly. I`m useing a matox marvel g200. What version of video tools would I need for this prog??

cheers.

[Dr Mordrid]

IF you can find a patch to open the YUY2 channel you can do all manner of alternate codec captures. I think there's one on http://www.murc.ws called the "Flying Dutchmans YUY2 patch". If this patch works on your G200 then MediaStudio Pro or Video Studio can capture either MPEG-1 or MPEG-2 by setting it up to capture YUY2 and using the Ulead MPEG capture plugin.

The limitation is how large a frame you can capture, which is determined by the speed of your CPU. A 1 ghz +/- most often allows full frame MPEG-2 so....

Dr. Mordrid

[disastermaster]

Thanks for the info, I`ve just downloaded a trial of Ulead Videostudio. I`ve also got the yuv patch.
It works fine in mpeg1 mode (pal vcd) but when I try Svcd pal the template in videostudio does not match any of the video formats with the marvel, and therefore it refuses to capture

Any Ideas?? Cheers

[Dr Mordrid]

This can be a problem with the G-cards because they are limited to the 704x480 using the ATSC frame size and its half and quarter frame counterparts. As such they don't do custom frame sizes in the drivers.

The workaround for many of us has been to install the HuffYUV codec (2:1 compression, lossless) and capture serial (properly: segmented) SVCD sized *.avi's using AVI_IO or VirtualDUB, both of which can set up custom frame sizes.

To so this you first set up the HuffYUV codec, which you can get here;

http://www.math.berkeley.edu/~benrg/huffyuv.html

and set the capture program to use YUY2 as the source and HuffYUV as the compression. Then you set up a custom frame size for SVCD (480x480 NTSC or 480x576 PAL) and capture the *.avi's.

In AVI_IO (my fave) you first set up the 704x576/480 YUY2 capture mode then set up the SVCD custom frame (480x480 NTSC, 480x576 PAL) in the Capture Options right before starting the capture.

Some G-card drivers won't do this with 704x576/480, giving a scrambled result. If this happens try using a 352 width. Experiment.

These can then be imported into VideoStudio or MediaStudio for editing and export using the SVCD output selection.

Dr. Mordrid

[disastermaster]

thanks

Thanks for the help . I downloaded the trial of avi_io and the huffy codec and tried them out. The results where great in avi format.I then encoded them to svcd using the svcd(Pal) template in TMPGnc. The results didnt look too good on my 19inch monitor, but when I outputed the signal to Tv (via PowerDVD Marvel composite Scart) The results were noticeably better. I presume that if I burn the svcd and play it on my standolone dvdplayer the results should be the best way of viewing this file?

Why is the format for svcd 480 x 576 ie taller than it is wide?

The trial of avi_io seems complete (no timelimit,or restrictions) is that right or am I missing something?

The huffy codec did produce a good quality avi file but also used a lot of disk space, Could I use the Divx4 codec at 6000 low motion to produce a good quality file ready for encoding to svcd or is the quality going to suffer?

Anyway thanks for your time again!

[Dr Mordrid]

You asked for it

SVCD is what's called an "anamorphic" format, which means it's horizontally squeezed. The same thing gets done when widescreen is encoded to a 4:3 screen, as is done with 16:9 DV cameras. It gets stretched back to full frame coverage when played on aware devices like the G-cards.

Technically this 480 wide format is called 2/3 D1 since 480 is 2/3 of a full ITU 720 frame width.

There is also a 1/2 D1 which is 352 wide or 1/2 of the ATSC 704 D1 frame width, but a full 480/576 high. VCD is also 1/2 D1 wide but only 240/288 high.

Why there are two D1's is explained by the fact that MPEG macroblocks are 16 pixels wide. 704 and it's 1/2 D1 width of 352 are both evenly divisible by 16 while 720 is not. This makes for the ability to encode VCD's and 1/2 D1 without horizontal scaling.

1/2 D1 and 2/3 D1 are commonly used to reduce the bitrate in satellite & cable transmissions. One example of a 2/3 D1 user is DirecTV.

Using DivX as an intermediary compression is NOT recommended as it's temporally compressed.

Normal codecs like MJPeg etc. are called Inter-frame compressed, which means each frame is compressed separately as a whole bitmap. Each of these real image bitmaps is called an I-Frame in all compression types.

MPEG's, including MPEG-4 (which is what DivX is), are called Intra-frame (subtle difference. intra=temporally = time) compressed, which means there is only about 1 real bitmap (I-frame) per every 15 +/- frames. The rest are synthesized mathematically, meaning MPEG fakes them.

These "fake" frames are called P (predicted) and B (backward) frames. P provides playback compression and B allows for reverse playback and searching, Re-compressing Intra-frame/temporally compressed video by editing or transcoding to another lossy compression, especially another MPEG variant, introduces quality losses you don't really want to deal with.

HuffYUV is not strictly speaking compressed at all, which is why it's lossless. It's what they call "Huffman encoded", which means a statistical table is set up to represent all the sample values in a smaller space. It therefore gives the highest possible quality in the least possible space. The compression ratio ends up being about 2:1.

A far better intermediate format would be PICVideo MJPeg. While still lossy it's not an intra-frame codec and its highest quality setting (Q=20) is low enough in loss to be very useful.

PICVideo will give you about 3-4:1 compression vs. HuffYUV's 2:1.

Dr. Mordrid