Highest Quality Video Editing

[lomaidala]

I've successfully combined clips from two different DVDs into a single video using DVDFab HD Decrypter, DgIndex, and VirtualDub. The problem is, I had to compress the video twice to get the final cut, and the resulting video quality suffered.

Is there a way to do what I'm doing and only compress the video once?
If not, is there a better method for doing what I'm doing?

The final result does not have to be an AVI. Any format that works with external subtitles is fine by me. I'm currently using Xvid to compress, but I'm open to other suggestions.

[qz3fwd]

just keep the mpeg2 video and remux to mkv or ts your combined clips presuming there are no resolution changes between clips.

[Baldrick]

Why compress twice? But you can export to a lossless codec like huffyuv in virtualdub.

Or join the dvd mpeg2 with for example Mpg2Cut2, videoredo, tmpgenc mpeg editor, etc. No quality loss.

[lomaidala]

just keep the mpeg2 video and remux to mkv or ts your combined clips presuming there are no resolution changes between clips.

Remux to mkv with MKVToolnix? What is ts? There are no resolution changes.

Why compress twice? But you can export to a lossless codec like huffyuv in virtualdub.

Or join the dvd mpeg2 with for example Mpg2Cut2, videoredo, tmpgenc mpeg editor, etc. No quality loss.

I compress twice because first I create two AVIs, one from each DVD, then I create a third video from those two AVIs. It's the only way I know how to append video for cutting in Vdub, otherwise the file size ends up being larger than my available disk space (over 45GB per video). Is it possible to cut the two DVDs before compressing the first time?

I actually have experience with Mpg2Cut2. It's a great program, but unless there's something I don't know, it isn't suitable for me in this case because it creates a brief fast-forward-then-freeze effect just before the seam where the cutting in the video is. Do videoredo and tmpgenc mpeg editor also do this?

Thanks for recommending huffyuv, I'll give that a try.

[hech54]

I compress twice because first I create two AVIs

XVID like you mentioned in your first post?
If so, that is absolutely, 1000% wrong.

[manono]

I compress twice because first I create two AVIs, one from each DVD, then I create a third video from those two AVIs.

As hech54 said, that's a very poor way to go about it.

It's easy to limit yourself to a single reencode by creating lossless AVIs for joining in VDub, as Baldrick mentioned. I think it's possible to do the joining without the problems at the join places, something I've also noticed with MPEG2Cut2.

I'd try it this way; do the cutting in MPEG2Cut2. Create a bunch of MPEGs. Add them one after another in DGIndex until they've all been added. Go File->Save Project and Demux Video. The result will give you one joined M2V video file and one joined AC3 (or whatever the audio originally was) audio file. Open Muxman, add in the video and audio and create the DVD.

This depends on not only the video qualities being the same, but the audio qualities as well, especially the number of channels and bitrate. Once the pieces are joined in DGIndex you could also create an MPG or MKV or whatever. No reencoding. No guarantees, but it's worth a shot, I think. Maybe do a test first with two or three of the pieces.

[lomaidala]

I've decided to go the huffyuv route because I want to use filters to color-correct some of the clips. This is what I intend to do:

• Load the video from the first DVD into Vdub.
• Apply a deinterlace filter (both DVDs are interlaced).
• Save as AVI.
• Load the video from the second DVD into Vdub.
• Apply a deinterlace filter.
• Apply a gradation curves filter for color correction.
• Save as AVI.
• Append the two AVIs.
• Apply a resize filter to correct anamorphic distortion.
• Cut the video.
• Save as third and final AVI.

Is there anything that looks funky with this method, or anything I could be doing better?

Thanks for the suggestion, manono. I tried it out just for the hell of it but got a bunch of error messages. I'm sure it's my fault, but since I plan on using huffyuv, I'm okay leaving it a mystery.

I compress twice because first I create two AVIs

XVID like you mentioned in your first post?
If so, that is absolutely, 1000% wrong.

Good to know, thanks. I knew something was up.

[jagabo]

The learning curve is steep but you could do all your cutting, pasting and filtering in AviSynth. Then encode to your final output format directly. You won't need an intermediate file.

Or you can frameserve from VirtualDub to your encoder to avoid an intermediate file.

[lomaidala]

Is there a guide you could direct me to for frameserve? I looked here and it said frameserving does not support filters.

Thanks for mentioning the AviSynth method, but I think I'll hold off on it for now since it sounds complicated. I'm curious though, would there be a quality difference compared to using the intermediate file I'm using now?

[poisondeathray]

Are you sure the DVD's are interlaced? If they are films / theatrical content (eg. a movie) - the actual content won't be interlaced ; you would need to inverse telecine, not deinterlace

If they are truly interlaced, you have access to higher quality deinterlacers in avisynth e.g. QTGMC (again, some learning curve) . And yes there will be a visible quality difference

[jagabo]

Is there a guide you could direct me to for frameserve? I looked here and it said frameserving does not support filters.

That page specifically says to use frameserving when you

want to use VirtualDub filters without creating an intermediate file.

Bascically, you set up your filtering and editing and instead of saving to a file you select File -> Start Frame Server. This creates a signpost file which is used to communicate between VirtualDub and the encoder (or another editor). You open that signpost file in the encoder as if it was a video file. Try frame serving from one instance of VirtualDub to another to see how it works. Not all editors/encoders support frame serving.

Thanks for mentioning the AviSynth method, but I think I'll hold off on it for now since it sounds complicated. I'm curious though, would there be a quality difference compared to using the intermediate file I'm using now?

There's no difference in quality between frame serving and using a losslessly compressed or uncompressed intermediate file. But there are better filters in AviSynth.

[manono]

I looked here and it said frameserving does not support filters.

If I'm understanding the question correctly, that's not right. I don't frameserve using VDub but, when necessary, make lossless intermediate AVIs and frameserve using AviSynth, and the page to which you linked says:

The following options will work:

Video filters.
Audio offset (displacement/skew correction).
Audio source switching (remultiplexing).
Range cropping (start/end) and segment deletion (i.e. crappy edits).

Thanks for mentioning the AviSynth method, but I think I'll hold off on it for now since it sounds complicated.

Yes, it takes some learning, but if you're serious about your video hobby you'll eventually get around to learning it. And then you'll kick yourself for not having learned it earlier. This is especially true if you like to do video restoration and improvement.

[lomaidala]

Sorry everyone, I read that page I linked to too fast, and missed the key line. Without that line, it looks like this:

Which options do not work during frameserving?

• Video filters.
• Audio offset (displacement/skew correction).
• Audio source switching (remultiplexing).
• Range cropping (start/end) and segment deletion (i.e. crappy edits).

Sorry about that.

Are you sure the DVD's are interlaced?

I guess now I'm not sure, but they look pretty interlaced. They are episodes from a TV series. How can I be sure the actual content is interlaced?

Try frame serving from one instance of VirtualDub to another to see how it works. Not all editors/encoders support frame serving.

I started frameserving, but how do I get it to another instance of VirtualDub?

Yes, it takes some learning, but if you're serious about your video hobby you'll eventually get around to learning it. And then you'll kick yourself for not having learned it earlier.

Well now you got me extra interested.

[manono]

I guess now I'm not sure, but they look pretty interlaced. They are episodes from a TV series. How can I be sure the actual content is interlaced?

Bob it and look for fields/frames repeating in a 3 2 3 2 3 2 pattern. If you find them then it's film sourced and can be IVTC'd (with AviSynth). If every field/frame is unique in motion scenes, it was shot on video and is really interlaced. Or post a 10 second sample that has steady motion.

[newpball]

I've decided to go the huffyuv route because I want to use filters to color-correct some of the clips. This is what I intend to do:

• Load the video from the first DVD into Vdub.
• Apply a deinterlace filter (both DVDs are interlaced).
• Save as AVI.
• Load the video from the second DVD into Vdub.
• Apply a deinterlace filter.
• Apply a gradation curves filter for color correction.
• Save as AVI.
• Append the two AVIs.
• Apply a resize filter to correct anamorphic distortion.
• Cut the video.
• Save as third and final AVI.

1. Deinterlace (I assume it is not film based) does not make any sense if you put this back on a DVD.
2. What anamorphic distortion are you talking about?

[jagabo]

Try frame serving from one instance of VirtualDub to another to see how it works. Not all editors/encoders support frame serving.

I started frameserving, but how do I get it to another instance of VirtualDub?

Just start an other instance of VirtualDub (double click on the VirtualDub icon, or in the first instance of VirtualDub select Tools -> Open New Program Instance) and select File -> Open Video File.

[newpball]

So basically the two videos have to be placed one after the other and be put back on a DVD?

Before you start getting the Avisynth battery of filters ready, deinterlace things and fix anamorphic things and do other things to 'improve' the video I think it would be better you post a small 3-5 second mpeg-2 sample (use DGIndex for this) from each source so we can see what we are dealing with.

[lomaidala]

Attached are samples of each of my two videos. This is with Honor Pulldown Flags since the original video is 29.97fps. But it's strange. Now that I'm viewing these m2v files in my player, they appear different than both the original DVDs and what I get in Vdub. These clips appear to have no interlacing.

Bob it and look for fields/frames repeating in a 3 2 3 2 3 2 pattern.

By "bob it," I take it to mean apply a Bob doubler filter and then go frame by frame? The result seems to be 2 frames for every 1.

1. Deinterlace (I assume it is not film based) does not make any sense if you put this back on a DVD.
2. What anamorphic distortion are you talking about?

1. As of right now, my plan is not to create a DVD, but rather to create AVI files from my DVDs.
2. The original DVD is anamorphic 720x480, but I'm concerned that if I make an anamorphic AVI and take it to a friend's house that it won't display properly, and appear stretched or squashed. So I resize it to 640x480 because I figure a ratio of 1:1 is safer.

Just start an other instance of VirtualDub

I get this error when I open the frameserve file:
VirtualDub Error
AVI Import Filter error: (Unknown) (80040154)

So basically the two videos have to be placed one after the other and be put back on a DVD?

I place them one after the other because that's the only way I know how to cut video using Vdub, but the end result is that video 1 has certain clips replaced using video 2.

[newpball]

Attached are samples of each of my two videos. This is with Honor Pulldown Flags since the original video is 29.97fps. But it's strange. Now that I'm viewing these m2v files in my player, they appear different than both the original DVDs and what I get in Vdub. These clips appear to have no interlacing.

Good, because they are movie based then the interlace problem is solved!

However your sources show blending and duplicate frame problems! It might need more frames and further examination to see what is exactly happening.

1. As of right now, my plan is not to create a DVD, but rather to create AVI files from my DVDs.

Excellent choice!

Personally I hate AVI as a destination container, it is an old container format that lacks certain flags and features.
I would consider MKV.

2. The original DVD is anamorphic 720x480, but I'm concerned that if I make an anamorphic AVI and take it to a friend's house that it won't display properly, and appear stretched or squashed. So I resize it to 640x480 because I figure a ratio of 1:1 is safer.

That is a good thing if it is anamorphic. That means it is higher quality!

Resizing to 640x480 is totally wrong as you would destroy detail in the horizontal space. Leave the 480 as is and 'upsize' the horizontal instead.

Ignore finger pointing "standards" people who choke over the fact you would scale to a non-standard size, they prefer standards over video quality.

[jagabo]

The two m2v files in post #18 are encoded interlaced and have interlaced frames. There's field blending but I think at least some of it is intentional. It's very common for intro sequences to do odd things like this because they speed up or slow down video to match the music timing. You will not be able to cleanly restore these to m2v files to their original progressive film frames. You should upload samples from the main body of the show. Include a long medium speed panning shot if possible. Because character animation in cartoons is usually done at low frame rates (like 12 fps) but panning shots are typically done at the normal film frame rate, 24 fps.

[manono]

There's field blending but I think at least some of it is intentional.

I have a slightly different take on it. I don't think any of it is intentional, but just the result of creating a really crappy master for DVD. And even if it was intentional, that doesn't mean you shouldn't do your best to improve it by unblending it as best you can. Yes, I understand that's not what you were saying, but you were only asking for a more representative sample.

You will not be able to cleanly restore these to m2v files to their original progressive film frames.

Not entirely, but you can do a pretty good job of it and, as you speculated, the main body of the episodes should be even 'cleaner'. But even the last 2/3 of the samples are pretty 'clean' after having been unblended.

Yadif(Mode=1)#or the better QTGMC
Srestore(FRate=23.976)

And this is nothing lomaidala can do in VDub.

[lomaidala]

Attached are panning samples from the main body of the show. There are other panning shots I can upload if these aren't the right speed.

You will not be able to cleanly restore these to m2v files to their original progressive film frames.

Does this mean I will not be able to cleanly get rid of the interlaced look?

And even if it was intentional, that doesn't mean you shouldn't do your best to improve it by unblending it as best you can.

So undo the field blending? Is this different from deinterlacing?

Personally I hate AVI as a destination container, it is an old container format that lacks certain flags and features.
I would consider MKV.

Can you name something specifically that I would gain by using MKV instead?

[manono]

And even if it was intentional, that doesn't mean you shouldn't do your best to improve it by unblending it as best you can.

So undo the field blending? Is this different from deinterlacing?

The end result for both is it will no longer be interlaced, but unblending is different from deinterlacing. Just deinterlacing won't do anything about blending.

Anyway, my take on it is the fields don't line up properly in the two new samples and I'd handle it as in my earlier script - a bobber followed by SRestore. Others may have a different or even a better solution.

[lomaidala]

I'd handle it as in my earlier script - a bobber followed by SRestore.

Oh, so that was a script for AviSynth? I tried pasting it in to the end of the basic script I use, and got an error when I opened it in Vdub:

VirtualDub Error
Avisynth open failure:
Script error: there is no function named "Yadif"
(C:\Movie.avs, line 3)

I guess this is what you meant by it's nothing I can do in Vdub? If so, what filters would you recommend I use in Vdub in this case?

[manono]

I guess this is what you meant by it's nothing I can do in Vdub?

I meant that VDub filters can't unblend anything. Only AviSynth filters can.

Script error: there is no function named "Yadif"
(C:\Movie.avs, line 3)

Yadif has to be loaded in a different way from most other AviSynth filters, either as a C Plugin or as a Stdcall_plugin. It's explained in the included doc. I do it like this:

LoadCPlugin("C:\Path\To\Yadif.dll")

[lomaidala]

I've started doing all of my cutting using AviSynth and much prefer doing it this way rather than doing it in VirtualDub. Thanks to those who recommended it! It's especially useful since I will likely be doing this multiple times if I change my mind on a color correction setting. There's also now the possibility of uploading cutting settings for people who want to create the edit I'm making for themselves.

How much more is there to AviSynth in terms of what I'm doing? If I'm able to do cutting just fine, is color correction difficult?

I do it like this:

LoadCPlugin("C:\Path\To\Yadif.dll")

Thanks, but now I get this error:

Avisynth open failure:
Script error: there is no function named "Srestore"
(C:\Movie.avs, line 6)

I'm curious though, would this line change the video to 23fps? If so, why change it if the original is 29fps?

[jagabo]

You need the SRestore() package and all the filters it relies on. http://avisynth.nl/index.php/Srestore

You've barely touched on what AviSynth can do. There are hundreds of internal and external filters for all kinds of things.

internal filters: http://avisynth.nl/index.php/Internal_filters
some external filters: http://avisynth.nl/index.php/External_filters
more: http://avisynth.nl/index.php/Main_Page#Filters.2C_external_plugins.2C_script_functions_and_utilities

[manono]

I'm curious though, would this line change the video to 23fps? If so, why change it if the original is 29fps?

It'll be changed to 23.976fps, yes. You change it because that was the 'original' or 'base' framerate, except maybe for that fast moving intro part. If the intent is to create another DVD it'll still be better unblended and then encoded as progressive 23.976fps with 3:2 pulldown added, something the idiot DVD makers should have done to begin with. For anything else you want just the unblended and progressive 23.976fps video. I suppose a regular IVTC could have been done for most of it, except those fields didn't line up quite right. I've seen scripts before that can fix the problem but I'd rather just use QTGMC on that sort of thing.

[lomaidala]

You need the SRestore() package and all the filters it relies on.

I'm still getting an error. Should I be using Srestore 2.7e or 2.7g? And should I be loading the AVSI file in my AVS script using "Import(Path)"? My script currently looks like this:

LoadPlugin("C:\dgmpgdec158\DGDecode.dll")
LoadCPlugin("C:\AviSynth Plugins\yadif17\yadif.dll")
LoadCPlugin("C:\AviSynth Plugins\masktools2-25.dll")
Import("C:\AviSynth Plugins\Srestore.avsi")
MPEG2Source("C:\Movie\VTS_01_1.d2v")
Yadif(Mode=1)#or the better QTGMC
Srestore(FRate=23.976)

It'll be changed to 23.976fps, yes. You change it because that was the 'original' or 'base' framerate, except maybe for that fast moving intro part.

It's my understanding that using Forced Film in DGIndex will result in 23.976fps, and using Honor Pulldown Flags will result in 29.97fps. I've been using Honor Pulldown Flags for these videos. Should I be using Forced Film instead now that I know the final video should be 23.976fps?

[manono]

It's my understanding that using Forced Film in DGIndex will result in 23.976fps,

Yes, but in the absence of the TFF and RFF flags to guide it, it blindly removes every 5th frame, not what you want. Unless you have a very high percentage of Film content (as defined by DGIndex), you'll always want to use Honor Pulldown Flags.

I'm still getting an error.

And the error is...?