My thoughts on Avisynth & VV3

BradK wrote on 10/9/2002, 1:02 PM
A note to all who will listen. :-)

I use AVISynth scripts daily to help me clean up my source footage (primarily VHS and TV Caps) before I do any editing/encoding. This is typically done with TMPGEnc for MPEG-1 and VirtualDub for DivX whereas my DV projects are all done in VV3. There have been occasions where I'd like to use my analogue footage in VV3 but the following issues would have to be dealt with before I make any serious attempts at it.

1) VV3 lacks good pre-processing filters which forces the usage of other utilities such as AVISynth and/or VirtualDub filters. These filters are most often essential when dealing with analogue captures.

2) The use of external tools is complicated further since VV3 it seems, only supports "real" AVI's and won't even attempt to open .avs files.

3) To work around this issue, one can re-render the AVISynth generated AVI in VirtualDub, re-encoding it with Huffyuv to a new file before importing it in VV3. This requires twice the amount of disk space which can amount to tens of GB when dealing with analogue footage. This also eats up a great deal of time.

4) The above Huffyuv source must also be converted to RGB during the encode or else VV3 has issues displaying the file. (Most likely Huffyuv's issue)

If VV3 had some good preprocessing filters, or could open AVISynth files directly I would use VV3 to edit and encode analogue footage as well as DV. I've also heard the MainConcept MPEG-2 codec is very good, only wish I had better options beyond DV.

Comments

SonyEPM wrote on 10/10/2002, 9:08 PM
Hi Brad-

Please explain what your actual situation or problem is, in terms of "how the picture looks", or "the sound is abc", or "my desired destination format is xyz"- that kind of thing. There are many people here interested in helping with these kinds of issues. The more details you can provide the better.
BradK wrote on 10/16/2002, 6:16 PM
I'll do my best.

A typical scenario is as follows.

I capture with VirtualDub using the Huffyuv codec (http://math.berkeley.edu/~benrg/huffyuv.html) at 720x480 resolution with my ATI Radeon 8500 AIW 128Mb. Audio is captured at 48 kHz 16-bit stereo uncompressed PCM.

In order for my capture to be viewable in VV3, the Huffyuv codec must be set to RGB for output (http://members.shaw.ca/brad_k/HuffyuvRGB.jpg) or I end up with garbage in VV3 (http://members.shaw.ca/brad_k/noise.jpg) Setting up to output in RGB isn't such a big deal, but before I bring it into VV3, I want to clean up the source, esp if my source is vhs or a tv cap (my current project is Laserdisc).

To clean up my Huffyuv source AVI file, I rely heavily on VirtualDub filters and/or AVISynth filters for noise reduction. Most of the time, I'll use VirtualDub filters within my AVISynth script which streamlines the process greatly, esp if some filters require conversion from RGB to YUY2 and visa versa. I can also pick the order in which the filters are run (makes a big difference). This is also where I'll apply Inverse Telecining for film sources, pick my favourite resize filter etc.

Once this is complete, I use AVISynth to frameserve the file to VirtualDub or TMPGEnc. I open the .avs file as I would any .avi file. This feature is what I'd like most for VV3. I could preview the entire source with all filters applied in my timeline. However, VV3 won't allow any source file with the .avs extension to be imported.

So, my option is to then run the .avs file through Virtualdub, and re-encode with the same compression settings for audio & video. This will generate a clean .avi file which I can then (finally) import into VV3. The problem with this is, it takes a long time to re-encode. It also eats up twice the disk space for my source files.

The silly thing is, I don't even want to edit my source once it's in VV3. All I want to do is to use the MainConcept codec rather than TMPGEnc to encode to MPEG-2. I've seen some comparisons of the two recently and while I believe TMPGEnc is superior for MPEG-1 (VCD's), the MainConcept is clearly the winner for MPEG-2 content especially when viewing text captions. It's also much faster, that is if you can get your source imported into VV3. :-)



I use this capture setup as I find it achieves the best possible results.
owlsroost wrote on 10/18/2002, 1:37 PM
Hi Brad,

Have you tried the 'Link2' software (see http://www.videotools.net/index.php?htm=intro2link2.htm and http://www.videotools.net/index.php?htm=virtualdub_frameserver_premiere.htm for more info).

From the website :

"Introduction:

This program is designed to add additional input formats to programs who can read AVI files. This program adds a new input engine to the existing programs and feeds the video/audio through the already implemented AVI interface. Currently my program is designed in such a way that it adds support for Avisynth, Virtualdub frame
server files and AVI files with and without compressed audio. So this tools enables you to e.g. read Avisynth files in any program that reads AVI files (Like Premiere, CCE SP, Ulead Media Studio Pro )

Common applications:
Import / Encode your Avisynth scripts in know editing programs and/or encoders like Premiere, Ulead Media Studio Pro, Cinemacraft Encoder (CCE), Ligos, Windows Media encoder and Cleaner. Besides this, it enables you to import MPEG/DVD material in your encoder/editing programs in a combination with other programs. (See guides section). Above all, this tool enables you to frame serve from Virtualdub directly to any encoder or editing program. Detailed information about this can be found in the guides section."

Tony
DDogg wrote on 1/22/2003, 9:28 PM
BradK,

I hope you do not mind me jumping in. While not as flexible as native AVS support, VFAPI 1.04 will allow you to frameserve, via avisynth to VV 3.0c. I have used this to serve 6 gig mpeg files to VV and got excellent results. Please let me know if I can be of help. VFAPI can be slightly aggravating to install. Definitely read the small doc that comes with it. I am sure you can find VFAPI on any of the better video hobby sites. This is where much of the active development of avisynth is taking place.

One does wonder why native support for the incredible power of avisynth has not been brought to a fine product like VV. Perhaps it is my ignorance speaking, but I would assume all needed source code would be available for study in the Virtual Dub source. As you mentioned, VDub actively reads file.avs scripts.
wk wrote on 2/12/2003, 2:54 PM
Well, having Vegas frameserver would be SUPER!

I myself was looking for it, becuase i wanted to frameserve from vegas to virtualdub so that i could use the easy but powerful filters of virtualdub to add subtitles to the movie....i dont know how to do it, as i dont have soo much disk space to make a raw avi.