PDA

View Full Version : AVI vs Mpeg2


nymph
10th July 2005, 15:57
I have a Panasonic PV-GS120 camcorder. The firewire port is fried on it so I am forced to transfer video via S-video cable and compsite audio cables. I have Pinnacle Studio 9 and I am trying to figure out what the best method of video transfer would be in my situation.

From camcorder to AVI using PICVideo MJPeg Compressor?
From camcorder to AVI using DV Encoder?
From Camcorder to Mpeg2?

The first 2 options give me about 13-14.2GB file with the DV Encoder being the largest. The weird thing is that when I goto the properties of the file and the summary tab it says the bit rate is 1536Kbps. Isn't that pretty bad? And why such a big file if the bitrate is so low?

With the Mpeg2 file it gives me a playable bitrate of around 8-9Mbps but the file is half the size.

My plan was to transfer the video from the camcorder with the DV encoder to an AVI file and then use Tmpgenc Express3 to encode to an Mpeg2 with a fixed bitrate of 9500. Am I going about this all wrong? Can anyone help me?

Thanks,

nYmph

Mug Funky
10th July 2005, 16:45
the 1536 is for the audio only...

as far as the method goes, can you not borrow a camera with a working firewire? going through analog doesn't strike me as the kindest way to treat your video... i suppose you may run the risk of frying the firewire on a camera you don't own though (if your computer was the problem). that could be bad...

nymph
10th July 2005, 17:20
OK 1536 bitrate for audio makes sense. How can I find out what the bitrate for the AVI video is?

Is there any software out there that will transfer video from my camcorder using S-Video and composite audio with NO compression at all?

nYmph

Delphin
15th July 2005, 14:28
OK 1536 bitrate for audio makes sense. How can I find out what the bitrate for the AVI video is?

Is there any software out there that will transfer video from my camcorder using S-Video and composite audio with NO compression at all?

nYmph

I agree that figuring out a way to grab the DV format bitstream would give the highest quality but if you are determined to use Analog capture then you could try the HUFFYUV lossless codec if it will work with your capture hardware.

The HUFFYUV codec is the only 'lossless' codec I have found that is fast enough to capture in realtime without a lot of 'dropped frames'.

You can get a very good version right here on doom9, but I had to use a two step process to get the latest doom9 version installed.

Get the original version from . . .

http://neuron2.net/www.math.berkeley.edu/benrg/huffyuv.html

Here's the direct link to the zip file you need . . .

http://neuron2.net/www.math.berkeley.edu/benrg/huffyuv-2.1.1.zip

This will get you the all important huffyuv.inf file you need to install the codec.

Now get the latest stable bug fixed version from doom9 at . . .

http://www.doom9.org/Soft21/Codecs/huffyuv_ccesp-patch_025.zip

Now here's how to install the updated version . . .

1. Create a temp folder and move the huffyuv_ccesp-patch_0.2.5.dll DLL file into it.

2. RENAME the above file to just plane "huffyuv.dll"

3. Now unzip the huffyuv-2.1.1.zip from the first link I listed above and get the "huffyuv.inf" (NOTE the .INF and NOT the .DLL) and move JUST THE INF file into the same TEMP folder you created above.

4. Now that you have the original "huffyuv.inf" and the patched file (renamed to plane "huffyuv.dll) in the same folder, just RIGHT-CLICK the huffyuv.inf file and pick 'install' from the dropdown menu.

First the good news . . .

HUFFYUV can shrink the video a bit without degrading it's quality at all.
(It's one of only a few truly LOSSLESS codecs)

NOW the bad news . . .

We are still talking about HUGE files.

Uncompressed 720x480 30fps RGB video is over 30 megabytes a SECOND (Ouch!).

Drop the resolution to 640x480 and the frame rate to 24 (the rate used for Hollywood films) and you can get this down to 'only' about 22 megabytes a second.

Huffy UV can only shrink this by a factor of about 2 up to about a factor of 4 or 5 depending on the options and content of the video, so we are still talking about a LOT of data.

To get numbers in the 4-5 range let HUFFYUV change the colorspace from RGB to YUV. This happens later in MPEG compression anyway, so you would still not be degrading the video (This is the bottom dropdown option in the RGB list in the HUFFYUV codec configuration options). Note that if your capture card already captures in the YUV mode the color space conversion is already done so you still get the larger compression factor vs. the horrible RGB numbers listed above).

This means that, even after compression by HUFFYUV the video will still eat about 6-15 MegaBYTES of hard drive space every second. (not the 4 to 8 megaBITS/second we usually talk about in MPEG2 encoding)

By comparison MJPEG starts to look reasonably 'lossless' (though still technically lossy) at compression ratios of 6 to 10, which would cut the required bit rate down by 2 or 3 verses HUFFYUV, but if you have a huge hard drive and only want to keep the video until it's MPEG2 encoded you can't beat HUFFYUV for speed and quality (true lossless encoding).