Use Virtual Dub to edit / cut by frame numerically

[loninappleton]

That most recent problem of opening the compression options was solved by bypassing the
the shortcut on the desktop.

I checked and the full processing mode is active during that program load. They must work together or the compression menu is cancelled out.

What is the immediate need for this? We haven't gotten to manipulating bitrate or things
necessary for file size changes.

If there is a need to work in full processing mode I was wondering if those huge file sizes and HD space necessary are temporary? I'll check for the Huffy entry next time.

[loninappleton]

Your addition of screen shots is beneficial for all this. I'm going over more of it.

But the Huffy was not found under compression options. I'll try to run the INF inside the Virtual Dub 1.10.4
folder. Beyond that I don't know where else to run it.

[loninappleton]

Ran it in the folder and still not installed. The screen shot shows it installed, so what's the action to take with it?

In other words, what do _you_ do to set up?

[hello_hello]

I have loaded my .avi and the Video> compression selection is greyed out. I also tried the control key combination which was not fooled-- still no result.

If you're not using any VirtualDub filters, use the "fast recompress" option for encoding. That should output the video in the same format as the source. The VirtualDub filters are disabled.
The other options... (normal and full processing mode) convert the video to RGB by default. I'm not really sure why. Possibly because a lot of VirtualDub's filters only work with RGB. You can change the output format under the Video/Color Depth menu, but when using "fast recompress" there's no conversion and you don't need to think about it. If the source video is YV12 (for example) the output will be YV12.

[jagabo]

That most recent problem of opening the compression options was solved by bypassing the
the shortcut on the desktop.

I checked and the full processing mode is active during that program load. They must work together or the compression menu is cancelled out.

You set up that short to run your script that turns on Video -> Direct Stream Copy. If you run from other shortcuts or by double clicking directly on virtualdub.exe you will not run that script and the program will come up in Full Proccessing Mode.

What is the immediate need for this? We haven't gotten to manipulating bitrate or things
necessary for file size changes.

In Direct Stream Copy mode you cannot select a compression codec because the video will always be saved with the same codec as the source. The compressed video in the source file is copied straight to the output file so you can only perform simple cut/paste editing. In Full Processing mode you can perform filtering and transcode to a different codec. If you don't select a codec in Full Processing Mode you will get uncompressed RGB video in your output file.

But the Huffy was not found under compression options. I'll try to run the INF inside the Virtual Dub 1.10.4
folder. Beyond that I don't know where else to run it.

The problem isn't the location of huffyuv. It's that nobody every fixed the INF file to work under 64 bit windows. Earlier I gave you a link to instructions that showed how to install 32 bit huffyuv under 64 bit Windows. But if you don't need huffyuv in particular I suggest you use a different lossless codec -- one that installs easily under 64 bit Windows. UT Video Codec, for example. Or, as hello_hello suggested, you can install 32 bit ffdshow and uses its version of huffyuv.

[loninappleton]

I'm only half caught up here. As you can see this is a one-off tutorial where I know next to nothing.
I do recall Huffy and all in early attempts years ago to use a capture card with a VHS tape player and Hauphauge card. It was a complete botch. I never got it out to a file.

Trying to go over this repeatedly means when I don't drop one thing, I drop something else.
But I did simply run the UT Video Codec .exe. Where is that found? In the same compression menu?

If my original is 1280x720 as shown and I bring that into Virtual Dub or MeGUI how is the bitrate
calculated? That's the only place I can think of to start. I've seen bitrate calculators but you'd have to point it out.

More critical and related to another job I want to start, I've seen very sharp but very small
file sized images. 320x200 as I recall. These I believe were originally screen caps as well-- tv content. The question here is: How can I get a sharp image for playback on a CRT set in low resolution? Doing it manually should provide an advantage over using AGK which has given skewed audio etc. I've looked at those tv show files up close in playback-- no jaggies or anything. I don't know how it's done.

[jagabo]

But I did simply run the UT Video Codec .exe. Where is that found? In the same compression menu?

All video codecs VirtualDub can use are found in the Video -> Compression... dialog.

If my original is 1280x720 as shown and I bring that into Virtual Dub or MeGUI how is the bitrate
calculated? That's the only place I can think of to start. I've seen bitrate calculators but you'd have to point it out.

With losslesss codecs you don't get to select a bitrate. At most they'll have an option that lets you optimize for size or speed. The difference in file size (bitrate) between the two is usually pretty small.

More critical and related to another job I want to start, I've seen very sharp but very small
file sized images. 320x200 as I recall. These I believe were originally screen caps as well-- tv content. The question here is: How can I get a sharp image for playback on a CRT set in low resolution? Doing it manually should provide an advantage over using AGK which has given skewed audio etc. I've looked at those tv show files up close in playback-- no jaggies or anything. I don't know how it's done.

You just weren't looking closely enough. Depending on the quality of the scaler in the playback device you may get more or less jaggies. But generally the rule is the sharper the source and the sharper the upscale, and the more you upscale, the more jaggies you get. So if you enlarge a 320x200 video to 1920x1080 you're going to get a sharp image with lots of jaggies, or a big blurry image.

[loninappleton]

Today after looking at the link-- only one I saw not defined was to installing LameACM for audio-- I saw that the UT series of codecs seem to be installed on the Vdub compression window. That will have to do for now. Command line code is pretty confusing.

On that 320x200 No, I would not be converting up but down from 1280x720 on the original.

Let me ask this another way: If bit rate only makes small changes, how can I get a good and sharp -- ok half the size at say 640x480 from this 1280x720? Where's the starting place or does my original need some further preparation?

Also related, is there a generic method used to get a screen cap saved to a mass storage device smaller than when recorded? I think I'm trying to pull teeth here. Through all this I'm just looking for a method of work-- a series of steps. I will come up with those at the conclusion as I have done in other threads.

I realize this is slow- going.

[jagabo]

On that 320x200 No, I would not be converting up but down from 1280x720 on the original.

But if you are going to watch that 320x200 video full screen on a 1080p TV it will be upscaled by the player or the TV to 1920x1080. That's where you will get jagged edges.

Let me ask this another way: If bit rate only makes small changes, how can I get a good and sharp -- ok half the size at say 640x480 from this 1280x720? Where's the starting place or does my original need some further preparation?

If your source is of decent quality (that is, it doesn't require other filtering) just apply a resizing filter to reduce the frame size to the size you want.

Also related, is there a generic method used to get a screen cap saved to a mass storage device smaller than when recorded?

If you're talking about Windows just enable Icons View or Extra Large Icons View. That will give you a thumbnail for each video Windows recognizes (though that thumbnail may not be useful, for example, if a video starts with a long fade from black the thumbnail may just be a black box). If you're talking about some other media player it will depend on what that media player supports.

[loninappleton]

Resizing and filter don't sound like the same operation. We're looking to make a permanent 'save' to the smaller size 640x480 as example. I have no HDMI tv for playback, any DVD's from the friendly lending library get played on my old Sony Trinatron CRT.

Can you walk me through (as in a procedure with steps) for the resize?

Up there it looked like FFdshow has some input boxes for it. And a calculation for the reduction?

If you're concerned about the 'if you give a man a fish' canard, I'm alright with just the fish.

I'm only faintly familiar with MeGui but some processing/recalc is normally done in that, yes?

And scripts? After all this time there should be some canned ones by now.

[jagabo]

Can you walk me through (as in a procedure with steps) for the resize?

In VirtualDub:

1) File -> Open Video File.
2) Video -> Full Processing Mode
3) Video -> Filters... -> Add. Select Resize filter. Set the size you want.
4) Video -> Compression... Select codec and configure it.
5) File -> Save as AVI.

[loninappleton]

Very clear and straightforward. At times I didn't even know if we were still in the Vdub program.

Regarding resize, will a simple 640 x 480 do it or is a precise calculation needed?

To get enough room, I may set up a separate small HD or buy new which I've considered anyway.

Again, is that large workspace for full processing deleted at end of work?

Which codec of the UT group should I select-- what difference is there? UT has about 5 selections IIRC.

[jagabo]

You need to state exactly what you are trying to do. What are you starting with? What is your desired final product(s)? What editing and fitlering do you want to perform between those two?

[loninappleton]

I'll make a list. Back shortly.

I'll start with what I know from MediaInfo on the original, but snipped file which I viewed last night and appears 'all good.' Both audio and video.

It's good I'm not rushing ahead at this point.

[loninappleton]

Starting from the beginning, here is the info requested in the best screen shot I could make.
Note: No matter how often I repeated the simple act of asking MediaInfo to Export to text it didn't seem to copy anywhere for all my trying. For this reason some of the aduo part is dropped off.

[loninappleton]

As to your questions.

Let's shoot for a 1.3 GB size and maintain as much resolution as possible. That's a little bit over one half the original which should be 'safe' for this routine -- that is no audio skewing or other mess-ups.


I have made a fresh setup of Virtual dub 1.10.4 (x86) on my scratch disk. I have run UT video utils and FFdshow.

MediaInfo is a fresh download of 1.76

There are many other options showing for Vdub at the VideoHelp software page. Any further additions from those to add?

Also what is the right method to add these plugins etc: Within the Plugins folder or within the Virtual Dub 1.10.4 folder? I noted FFdshow makes it's own folder location in the (Program Files x86) directory.

I will go back and review your steps in the meantime.

[jagabo]

You don't want to use UT Video Codec (or any other lossless codec) for files you plan to watch. It's usually used for screen capture or intermediate files where you plan on performing filtering or editing in different programs. That intermediate file will be much larger than the AVC source you have now.

If you're just looking to make a smaller file for storage reasons you want to reencode with lossy codecs -- and you will lose some quality. Since the video uses AVC (aka h.264) you want to use that or something that compresses even better, say HEVC (h.265). But encoding with HEVC is very slow and many devices don't play it yet. So you're probably best off sticking with AVC. If you plan on using VirtualDub you will need a VFW AVC encoder like x264vfw, the 32 bit version for use with 32 bit VirtualDub. I'm pretty sure that codec installs correctly under 64 bit Windows so you don't have to use convoluted methods like those required for huffyuv.

The audio in your source is uncompressed and probably runs about 700 MB. Compressing that to 256 kbps mp3 (or MP2, AC3, or some other lossy audio codec) will reduce it's size by about 500 MB with almost no audible loss of quality. That puts you about 300 MB short of your goal of a 1.3 GB file. I wouldn't bother recompressing the video just for another 300 MB but if you really want to you should specify a video bitrate of about 2300 kbps. To get exact numbers use a bitrate calculator.

https://www.videohelp.com/calc

Since you're only reducing the video bitrate by about 20 percent I'd try keeping the same frame size as the source. The basic procedure in VirtualDub:

1) File -> Open Video File
2) Audio -> Full Processing Mode
3) Audio -> Compression... select codec and set bitrate
4) Video -> Fast Recompress
5) Video -> Set Compression... select codec and configure
6) File -> Save as AVI

If you find the resulting video is too block or lacking in detail you can try reducing the frame size. The process becomes:

1) File -> Open Video File
2) Audio -> Full Processing Mode
3) Audio -> Compression... select codec and set bitrate
4) Video -> Full Processing Mode
5) Video -> Filtering... Add... Resize to 640x360 -- or whatever size you want to try
6) Video -> Set Compression... select codec and configure
7) File -> Save as AVI

Configuring the video codec is the tricky part. For best results you want to use a 2-pass variable bitrate encode. VirtualDub doesn't support automated 2-pass encoding so you have to do it yourself. Set the codec up the way you want and with the first pass setting. Then Save As AVI. That codec will analyze the video on the first pass and create a small file with a log of what it found. A video file is produced during this pass but it's usually useless; you can delete it. Then go back and change the codec to second pass and Save As AVI again. This time the codec will use the information in the log file to determine how best to allocate bitrate to deliver the average bitrate you requested.

[loninappleton]

I appreciate your careful analysis and the step procedures. It's a lot to chew on. I will begin shortly. That business with the bit rate-- reduced by 20%-- is that a good number to use?

I will go over the options in Vdub before asking additional things. My audio needs no
further adjustment and so that element is settled.

And no, 1.3Gb is hardly a necessity to hit on the nose. I will avoid UT and follow the
other directions (it will be pretty much by rote) and see what happens.

[loninappleton]

It sounds like your trying to edit a video with long GOPs.

I've been trying to reorganize my procedure to get going on this. First the Group of Pictures issue.

In the dialog here it says:

" Use VirtualDub's Smart Rendering feature to re-encode only cut GOPs."

(Which is Group Of Pictures)

Is this a separate first step before any resize? Does it put humpty back together again rather than the GOP group of pictures? Let's see if we can get past that to begin with.


My only edit was on the original length AVI. I know of no other 'uncompressed' place to start

The place to start still eludes me and I have your steps close by.

Will close this post and revue the program again.

[jagabo]

Smart rendering is only useful during simple cut/paste editing, or when processing only a small part of the video (for example adding a fade-to-black at the end). The untouched GOPs will not be re-encoded when you save the result.

If you're resizing the frame all frames have to be re-encoded so there's no need for smart rendering. Whether you enable it or not all frames will be re-encoded.

You can perform both operations (cut/paste and resize) at once. But since you are resizing the frame the entire video needs to be reencoded.

[shans]

Smart rendering is only useful during simple cut/paste editing, or when processing only a small part of the video (for example adding a fade-to-black at the end). The untouched GOPs will not be re-encoded when you save the result.

If you're resizing the frame all frames have to be re-encoded so there's no need for smart rendering. Whether you enable it or not all frames will be re-encoded.

You can perform both operations (cut/paste and resize) at once. But since you are resizing the frame the entire video needs to be reencoded.

sorry to bump. I did so as my topic is relevant to this. I use virtualdub for editing as it is easy and also possible framewise. I edited a video with "Baseline@L3" profile, 1 ref, No Cabac AVC and re-encoded with smart rendering feature using the auto profile setting. The resultant video details as viewed in 'MediaInfo" is shown here:


It is seen that the encoded video parameters are different from that of the original. Can I conclude that this is because I used the Smart Rendering feature?

[jagabo]

I edited a video with "Baseline@L3" profile, 1 ref, No Cabac AVC and re-encoded with smart rendering feature using the auto profile setting. The resultant video details as viewed in 'MediaInfo" is shown here...

It is seen that the encoded video parameters are different from that of the original. Can I conclude that this is because I used the Smart Rendering feature?

I think so. Did you set the encoder to no cabac or a preset with no cabac? It doesn't automatically detect the properties of the source and match them. It doesn't have any idea what the properties of the source are. It just renders with the settings you specify.

Though, to tell you the truth, I don't really know if the smart rendering feature of VirtualDub works with AVC. I've only used it with Divx/Xvid.

[shans]

I edited a video with "Baseline@L3" profile, 1 ref, No Cabac AVC and re-encoded with smart rendering feature using the auto profile setting. The resultant video details as viewed in 'MediaInfo" is shown here...

It is seen that the encoded video parameters are different from that of the original. Can I conclude that this is because I used the Smart Rendering feature?

I think so. Did you set the encoder to no cabac or a preset with no cabac? It doesn't automatically detect the properties of the source and match them. It doesn't have any idea what the properties of the source are. It just renders with the settings you specify.

The "MediaInfo" of the original video and the x264vfw setting I used in virtualdub are attached. Actually, after editing I chose "Fast recompress"; smart rendering and then compression as seen here. I think the edited portion alone was re-encoded with revised setting as appplicable for "auto" setting leaving the original video to its own setting.

[jagabo]

The medium preset enables cabac. Setting Profile and Level to Auto doesn't mean automatically match the source. It means automatically set the profile and level to the settings corresponding to the medium preset and the other video properties (frame size, frame rate, etc.). If you want Profile=Baseline, Level=3.0 should specify that with the pulldowns.

Other things you can do to disable cabac are to use the Ultrafast preset, or enter --no-cabac in the extra options box.

[shans]

Thanks jagabo for the clarifications.

[loninappleton]

OP again,


I'll see what I can make of the tips given.

Putting aside Smart Rendering and still pondering a starting point which would clean many hings up such as GOP my question is:

In the Virtual Dub (now talking about Virtual Dub 1.10.4 available at Videohelp) there is a way to go to key frames and, I assume be able to select the cut process.

If I were using the original Part Two with the size of 2.1 Gb and cut on a key frame would that
eliminate all this talk of Group of Pictures?

Going to the next step would mean in sequence

Load Part Two 2.1Gb
Select Home key to begin the cut
Go to key frame at what is about 8 minutes in
Select End key to select that edit

And cut that beginning piece.

Then what is the best Save sequence (forgetting about the confusion of Smart Rendering)?

Just that much.

A resize procedure with codecs etc on the compression menu can be attacked separately and hopefully more simply.

[loninappleton]

An Update

A full stream edit is in progress.

Starting from the original AVI screen capture I did the following:

I first viewed the intro portion I wanted snipped in Virtual Dub. Using the correct keys
I had audio and video.

I marked the desired cut location on paper.

I engaged the Edit routine and found an edit point using the Goto KeyFrame keys option.

Selected the Cut.

The job is running. What I'll try to do is use all the tips provided as I go.

Full stream takes well longer than direct stream. This is of no concern. And I should have enough disk overhead on an older scratch disk I'm using for this job.

[loninappleton]

The job quit at 113Gb-- that's Gigabytes-- with a disk full notice.

I was able to find the huge work file and get rid of that.

To complete I have to get rid of more items on the scratch disk. Some of these are in a shared Network folder listed by Win7 as "Public.". If I'm just using a backup disk and do not need all the networked items, would that delete them from the other machine on the network or just the local one? I have been struggling with this.

[jagabo]

In full processing mode if you don't select a compression codec (Video -> Compression...) you get uncompressed video in your output file.

[loninappleton]

Ok, that's an error I made.

Two things:

Can I make the edit then go to compression and select that as part of one operation?

Which is a good default compression selection to make? I know we mentioned 640x480 (down from the 1280 x 720 original.)

I simply don't know how to do the setup yet, though I was able to go to a key frame, make the snip and do that much.

Also I know Hello Hello mentioned making two passes manually. Is that a relevant operation here?

Elsewhere I'm getting some forum advice about simply taking my work disk off the network and removing network folders thereby getting what should be plenty of room.