Filters- need more speed....

[rob1]

I'm converting VHS to DVD (Canopus->DVIO->VDMod (filter and edit, save a temp file)->CCE Basic (2-pass VBR ave=3,000).

I need to use VDub to edit out commecials as I highly prefer it to other options. Now, I'm getting very nice results, looks better than the original. BUT, only if I filter (I use SmartSmooth- looks great, little loss of detail). Filtering is taking FOREVER! About 9 hours for 1:15 minutes of video.

Having read fairly extensively on this, can someone confirm a few things that might help speed up this process? Particularly with regard to Avisynth, which I find very confusing and don't want to learn unless I'm sure it will make a big difference.

1. Obvious question- will using avisynth to filter really speed things up that much? I believe part of the speed improvement is due to using YUY rather than RGB- correct? But if I'm still going to VDub for editing, will I have to convert color and will this reduce the speed bonus (sorry- avisynth just confuses me so I ask dumb questions). So- in brief, will filtering via avisynth be considerably faster than Vdub with similar results?

2. If avisynth is worth learning due to speed increase, what's the best way to use it? Send the Dvio captured file to avsynth filters, then frameserve to VDub, save a temp file after editing and then encode via CCE? Or- edit with VDub, save temp file and send to avisynth, filter and either save temp file or frameserve to CCE?

3. Partially answered above, but for clarity- my impression is it is better/faster to save a temporary file rather than frameserve to CCE. Is this correct? (I've got the hard drive space to do that.)

4. Would overclocking help with the speed? How much? I'm using 1800 P4 and have downloaded SOFTSFB which allows overclocking on the fly. Haven't tried it out yet and am hesitant as I swear I read somewhere that overclocking can screw something up with the video processing thing- but I could be wrong. Anyway- any advice on overclocking to get greater speed?

Thanks in advance. I have read up on this, but I wanted to clarify before I really start to learn avisynth. I've screwed around with it and find it difficult.

[DJRumpy]

You don't need VDub at all if you use AVISynth. You don't even need to modify your original AVI. AVIsynth can remove all of the commercials, filter with smartsmoother, etc.

The only thing I use VirtualDub for is to get the frame number where a commercial begins and ends. With those numbers, plugged into the TRIM command, you simply remove them from your output.
Simply point your AVISynth script to your avi. The Trim command would look something like this:

clip1=Trim(5004,12029) #retains frame 5004 through 12029
clip2=Trim(13282,26289) #retains frame 13282 through 26289
clip3=Trim(29878,37282) #retains frame 29878 through 37282
clip1++clip2++clip3 #splices all of our retained frames together

Using VirtualDub, the slider, the keyframe button, and the forward/rewind buttons, you can quickly locate the start/end frames from commercial segments. The above command would start your video input at frame 5004, and stop it at frame 12029. At frame 12029, that would be where your commercials start (example only). The next line picks up at frame 13282, where your commercial ends, and your movie starts again. The last line, takes each segment you selected to retain, and splices them together as one movie. You can parse through two hours of footage in about 10 minutes. It requires no editing of your original avi. No resaving of avi segments, etc. Just take the script and drop it directly into CCE for encoding.

AVISource("c:\mymovie\movie.avi")
clip1=Trim(5004,12029)
clip2=Trim(13282,26289)
clip3=Trim(29878,37282)
clip1++clip2++clip3

1. Obvious question- will using avisynth to filter really speed things up that much? I believe part of the speed improvement is due to using YUY rather than RGB- correct? But if I'm still going to VDub for editing, will I have to convert color and will this reduce the speed bonus (sorry- avisynth just confuses me so I ask dumb questions). So- in brief, will filtering via avisynth be considerably faster than Vdub with similar results?

Yes it will. You won't need VDub.

2. If avisynth is worth learning due to speed increase, what's the best way to use it? Send the Dvio captured file to avsynth filters, then frameserve to VDub, save a temp file after editing and then encode via CCE? Or- edit with VDub, save temp file and send to avisynth, filter and either save temp file or frameserve to CCE?

Input AVI to AVISynth, and encode. No other steps. No temporary files.

3. Partially answered above, but for clarity- my impression is it is better/faster to save a temporary file rather than frameserve to CCE. Is this correct? (I've got the hard drive space to do that.)

No. This is a waste of time. There is no quality improvment in creating a temporary file. Although the actual encode time will improve minimally, with a raw avi input file over a frameserved file, the time wasted actually creating the temporary file and then encoding takes longer than frameserving alone.

4. Would overclocking help with the speed? How much? I'm using 1800 P4 and have downloaded SOFTSFB which allows overclocking on the fly. Haven't tried it out yet and am hesitant as I swear I read somewhere that overclocking can screw something up with the video processing thing- but I could be wrong. Anyway- any advice on overclocking to get greater speed?

Yes. How much? Depends on how much you overclock. Would it be worth it? Probably not. The gains you get would be a matter of minutes. Why make your system unstable for 5 minutes of gain? You take the risk of crashing in the middle of an encode, and having to start over. It's a personal choice. My opinion. Don't.

I don't know the specifics for the smartsmoother plugin, but I'm sure it's been ported to AVISynth. Someone here should be able to help, or just search google.

[menace]

Follow up to trim question--can i take 3 different avi sources, trim them then join them,all in one step? what would the script look like for something like that? Thanks!

[bunyip]

To menace,
Yes you can but they need to be of the same resolution and Avisynth needs to treat the video as a continuous source so you have to join them in the script first then edit as a single file.

You can do this one of two ways.
If the files are a segmented source (e.g. captured as 'video_1.00.avi', 'video_1.01.avi' & 'video_1.02.avi') you can add the following as your first command,

SegmentedAviSource("video_1.avi") #this concatenates all files named 'video_1.nn.avi

If they are just seperate files then you can use,
AviSource("video1.avi","video2.avi","video3.avi")

The trim commands will need to refer to the total number of frames in all 3 files.

[menace]

Thx!--I guess my math will have to be up to snuff to do that!

[bunyip]

Menace,

Not necessarily - DJRumpy already mentions the 'trick' - load all your videos together in the correct sequence into VirtualDub and use the slider control to pick the frame numbers that you want to cut and just plug these numbers into the TRIM commands. (VDUB strings all the videos into one).