VirtualDubMod 1.4.13 Cluster compression Helper, ver. 0.1.

ReadMe file.

(c) 2003, Andrey Gule [krolyk@hotmail.com]


  USAGE:
        VDMCH[.exe] <video_file_processed_by_cluster>[[[.##].avi]|[.parts]]


As you might already know, VirtualDubMod 1.4.* has very useful feature:
Net compression. It's limited to only 1-pass compression tasks, but yet
it's useful. Having 1-pass encoding done by 5-10 PCs is VERY good thing, 
believe me ;)

I have a VHS VCR and I do some video capture of camera records.
And here is where it kicks in: processing video with some filters to get it
look better makes my iCel 1,4 run at 1 FPS, which is miserable. I have
10 PCs in my office: 1 FPS on ten PCs turns out to be 10 FPS, that is much
more better ;)   This way I can have 30 hr task completed in 4-5 hrs.

But there is a major drawback in the method VirtualDubMod 1.4.13 use.
First it processes the source file via multiple PCs into multiple result 
files, and then runs a task to append all these parts altogether. 
Thus, it requires twice as much space, as the result file would take.

As far as I am quality-obsessed, I use HuffYUV (i.e. loseless compression)
as intermediate compression. And HuffYUV means up to 40 GB/hr. Usual tape is
1,5-2 hr. This means that I can process only one hour at a piece on my 80G
hard disk. And if I buy a new 120G HDD I will get only 60G more of working
space. This seems to be nonsense for me. I tried to work around this feature.

The idea is simple: why should I put these files all together into one
if I only need them to be processed and deleted? Also Dub can easily
append several files and work with them as with one big file.

First thing came into my mind was just open the 1st file in Dub and then
append all other files. Two problems: if you split your result file at 
1 file per minute, that means you have to open 80 files in correct order.
The second problem is the order. Result files are numbered as they 
are processed, not as they are played back in video.

DubMod stores this information in .parts file. OK, I just have looked 
through it and guessed what means what. Then I exported it to Excel,
sorted the table by the start frame number: this way I received correct
order of parts numbers. Then I go to Dub, Append AVI segment, Append,
Append... I've done this operation for two videos and then I started to
think: 'there must be another way'.

The other way I found in Dub script: .jobs file. There is a command Open
followed by multiple Append's. OK, this is it. I multiply the line with
Append, then I copy as a vertical block a sorted column from Excel and
I can see a correct numbers right in correct lines of .jobs script.
I only have to put the numbers into their place inside the line of script.

Doing this once more made me think again. And the thing is that
I'm a programer, and this is rather a diagnosis than an occupation ;)

Well, enough of talking. I've coded a utility that takes .parts file for
input and generates Open-Append part of .jobs script, all based on the info
from that .parts file. Interface should be very simple for anyone who is
able to use CLI (command-line).

It's very simple to use: after your task was processed by the cluster of PCs,
abort the creation of one big avi. Open Dub, open any avi with part of result
video, setup everything you need and save the job. You may setup several jobs
with your video if you need to: for instance the 1st and the second passes
of DivX compression. Close Dub, run my tool (VDMCH) passing the name 
of the video being processed as a parameter. You'll get a file named 
"your_video_name.jobs.part". Edit you Dub's .jobs script: replace line with 
Open(... with the contents of the .jobs.part file: there are one Open and 
multiple Append's. Now save the Dub's script, open Dub and run your jobs.

That's it, no more disk space waste.

If anyone wants, I may send him a source code (coded in MS VC++ .NET 2003,
in C++, plain WinAPI console application, STL, also some comments ;).

Thank you for your attention, this is all. The tool is available at
http://Hopka.org.ua/ in both zipped and uncompressed versions.


Avery Lee, thank you once again for keeping up with this essential tool.
I just wonder, what would we do if there's no VirtualDub ;)


PS sorry for so much talk. Feel free to transfer this info to forums/etc.,
    wherever you think this would be listened to. The soft is usual
    'freeware/no warranty/use at your own risk' - feel free to distribute.
