What would you do to improve this video?

[Caiosouza]

If you join it during authoring, you will hear it. You do need a process that will accept the AC3 file, and I am not familiar with the free authoring apps. It's been years since I used one of those. We're not familiar with all of the software you have available.

Amost anything will import .wav files, if it comes to that. I had no problems with AC3 or .wav with TMPGEnc. Later, you might want to check their website and try some of their software. Not free, but you can get trial versions. I guess I'm lazy, because for years their popular products have served well, to the point where I hardly use anything else except a few free things like HCenc or X264 encoder..

So I caught the filtered raw AVI video and AC3 audio, join TMPGEnc and I'm ready to author?

Or I get the m2v File passed the HCgui_025and AUDIO AC3, and join on TMPGEnc, then authoring (VIDEO_TS, ifo, vob) ! ?

[jagabo]

Finally I can listen to that AC3 after installing VLC, but ready after the DVD will be able to hear?

Of course you will. AC3 is the most common audio format found on commercial DVDs.

[Caiosouza]

Finally I can listen to that AC3 after installing VLC, but ready after the DVD will be able to hear?

Of course you will. AC3 is the most common audio format found on commercial DVDs.

But in any other DVD authoring, I've heard the audio on any player,windows media player, media Player Classic
And in this case, the final DVD, i listen the audio only with VLC installed.

And in the source file I hear the audio without having VLC, what changed?

The audio file should not be the same?

After all was not only separated first?

I can not risk that, time to move on to the final project, are 150 minutes of video Source to work!

[sanlyn]

You should make your test MPEG first. There is no point to working 150 minutes of video before your basic procedure hasn't been tested.

[Caiosouza]

You should make your test MPEG first. There is no point to working 150 minutes of video before your basic procedure hasn't been tested.

But I've done several tests by authoring sample,including your (sample_DVD 6500k_interlaced.mpg), all PC players no audio, only I listen when run with VLC

Do you not think that's weird?

I think I'm doing something wrong in DGIndex

PS : The DVDAuthorGUI is pretty good,thanks!

[jagabo]

But I've done several tests by authoring sample,including your (sample_DVD 6500k_interlaced.mpg), all PC players no audio, only I listen when run with VLC

Do you not think that's weird?

No. VLC has a built in AC3 decoder. Windows and some other players don't. You need to install a DirectShow AC3 decoder or enable the AC3 decoders in the players. Or maybe your AC3 decoder is mis-configured.

[Caiosouza]

But I've done several tests by authoring sample,including your (sample_DVD 6500k_interlaced.mpg), all PC players no audio, only I listen when run with VLC

Do you not think that's weird?

No. VLC has a built in AC3 decoder. Windows and some other players don't. You need to install a DirectShow AC3 decoder or enable the AC3 decoders in the players. Or maybe your AC3 decoder is mis-configured.

Solved!! And something wrong with my PC, in my other PC the audio works in any player! Without VLC!


How long do you think I work all 150 min, in a Core l5 2.8ghz 8GB DDR 3?

[jagabo]

How long do you think I work all 150 min, in a Core l5 2.8ghz 8GB DDR 3?

10 to 20 hours.

[Caiosouza]

I am amazed with the results obtained in the work of video,

I think it was miraculous what you managed to do and share with me, incredible !

It would be too risky to retouch this red leaking?

On TV are too too red

Look:

[jagabo]

Try adding

MergeChroma(aWarpSharp2(depth=30))

right after QTGMC(). Sanlyn probably has some better filters for that. You can decrease the overall saturation a bit with Tweak(sat=0.9) or so.

[sanlyn]

jagabo's aWarpSharp2 routine is an old standby among chroma fixers. I started there, too, but ran into complications with the poorly encoded original, especially red. It's shifted left as well as smeared. I finally settled on the routine below. Not 100% perfect -- slight overkill in some scenes, slight underkill in others, but nothing will completely fix low-bitrate damage or VHS chroma smear.

Place this statement at the top of the script, before the script opens the video (change the "path" data to match your system setup):

Import("Drive:\path\to\Avisynth plugins\FixChromaBleeding2.avs")

Place this code following the "MCTemporalDenoise" statement, but before the "Splin63Resize" statement:

ColorYUV(cont_y=-10,off_y=1.0)
FixChromaBleeding2() 
ChromaShift(v=4)
MergeChroma(aWarpSharp2(depth=30))
MergeChroma(aWarpSharp2(depth=20))

A copy of the filters is attached. Place the filters in your Avisynth plugins folder. Chroma fix routines such as these slow down the script by 5 to 10%.


[Attachment 21446 - Click to enlarge]

[Caiosouza]

jagabo's aWarpSharp2 routine is an old standby among chroma fixers. I started there, too, but ran into complications with the poorly encoded original, especially red. It's shifted left as well as smeared. I finally settled on the routine below. Not 100% perfect -- slight overkill in some scenes, slight underkill in others, but nothing will completely fix low-bitrate damage or VHS chroma smear.

Place this statement at the top of the script, before the script opens the video (change the "path" data to match your system setup):

Import("Drive:\path\to\Avisynth plugins\FixChromaBleeding2.avs")

Place this code following the "MCTemporalDenoise" statement, but before the "Splin63Resize" statement:

ColorYUV(cont_y=-10,off_y=1.0)
FixChromaBleeding2() 
ChromaShift(v=4)
MergeChroma(aWarpSharp2(depth=30))
MergeChroma(aWarpSharp2(depth=20))

A copy of the filters is attached. Place the filters in your Avisynth plugins folder. Chroma fix routines such as these slow down the script by 5 to 10%. In the motorbike shots, it helps to keep the cycles from looking like plastic toys.


[Attachment 21446 - Click to enlarge]

Very good, I'll try it
Unintentionally you enlighten me further any questions, would be as the order of placement of comand lines, it makes a significant difference from what I understand.

[sanlyn]

Unintentionally you enlighten me further any questions, would be as the order of placement of comand lines, it makes a significant difference from what I understand.

The order of the lines isn't absolutely mandatory, but its significance is that each line is a refinement of the line above it. More importantly, you would want to make these corrections before resizing the frame.

First, the ColorYUV statement reduces overall contrast a little (which is more important in other scenes than it is in the sample shot). Then as jagabo suggested and as I often do myself, the statement MergeChroma(aWarpSharp2(depth=30)) is a starting point that "shrinks" the overflow of the color. MergeChroma affects only color, not luma. Often this is enough to make the correction. But it has little effect on the intensity (saturation) of the color or its displacement a few pixels to the left.

So I allowed that statement to remain, but preceded it with something to first correct intensity and give aWarpSharp2 a little help. FixChromaBleeding2 affects the overflow and the intensity to a visible extent. It doesn't affect overflow completely. ChromaShift(V=4) shifts the placement of the red channel toward the right. Because you are working with YUV color, you must shift horizontally by an even number of pixels.

MergeChroma(aWarpSharp2(depth=30)) reduces the overflow even more, on each side of the affected edges. For me, this correction was not quite "narrow" enough, so the statement MergeChroma(aWarpSharp2(depth=20)) works with the more narrow extent of overflow, and narrows it even more. Usually the first aWarpSharp2 would be enough. But I felt that this sample needed a little more refinement.

The original script will therefore look like this:

Import("Drive:\path\to\Avisynth plugins\FixChromaBleeding2.avs")
MPEG2Source("Drive:\path\to\project\1992 MX 250 Sample.d2v", CPU=6, Info=3)
QTGMC(preset="fast")
McTemporalDenoise(settings="low")
ColorYUV(cont_y=-10,off_y=1.0)
FixChromaBleeding2() 
ChromaShift(v=4)
MergeChroma(aWarpSharp2(depth=30))
MergeChroma(aWarpSharp2(depth=20))
Spline64Resize(width*2, height)
Crop(2,10,-6,-16).AddBorders(4,12,4,14)
# --- re-interlace
AssumeTFF().SeparateFields().SelectEvery(4,0,3).Weave()

Hm. That's a lot of verbiage. Too much coffee this morning, I guess.

[Caiosouza]

I Trying it

Code:
Import("C:\Arquivos de programas\AviSynth 2.5\plugins\FixChromaBleeding2.avs")
MPEG2Source("C:\Documents and Settings\USER\Desktop\WORK FOLDER\1992 MX 250 Sample.d2v", CPU=6, Info=3)
QTGMC(preset="fast")
McTemporalDenoise(settings="low")
ColorYUV(cont_y=-10,off_y=1.0)
FixChromaBleeding2()
ChromaShift(v=4)
MergeChroma(aWarpSharp2(depth=30))
MergeChroma(aWarpSharp2(depth=20))
Sharpen(0.3, 0.0)
nnedi3_rpow2(2, cshift="Spline64Resize", fwidth=width*2, fheight=height)
Sharpen(0.3, 0.0)
Crop(2,10,-6,-16).AddBorders(4,12,4,14)
AssumeTFF().SeparateFields().SelectEvery(4,0,3).We ave()


In red is suggestion of jagabo

Decode it:

Estou adorando suas explicações!

Bom café da mahã,

Grande abraço !

[sanlyn]

A big hug back. I should add: for someone who has never gone into Avisynth, you're doing well. Thanks for your patience.

[Caiosouza]

WORKS FINE !!!

I'm thinking of leaving the yellow lines like this in his code.
The yellow lines code, i think that brought a (little) bit more visible mosquitoes,

The red leak resolved, perfect

Now tell me with your eyes clinicians, the red type is right

Originally a red bike that appears early on has this color, not a red and well, I do not want perfection, it does not exist, the photo and so you have somewhere to cling

Result:

[Caiosouza]

A big hug back. I should add: for someone who has never gone into Avisynth, you're doing well. Thanks for your patience.

Wow, I'm the who appreciates your kind patience,

What killed me was the problem with the AC3 audio to my PC to play

Now I follow: Demux Video+Track TO AVS code IN VirtualDub TO GET Raw Filtered Video TO PUT ON HCenc (6500-8000K DVd compliant) TO MUX Audio AC3+ m2v Video Result HCenc = Ready to Authoring DVD !

I need your eyes, to decide on the Sharpen yellow lines in the code, and type of red

You think you are ready for final project?

[sanlyn]

jagabo's code in yellow sharpens the video slightly, then resizes to full-frame DVD video, then does a little more sharpening of the resized results. A half-frame 352x576 video might play suitably on a CRT, but it will not look very good on an upsampling HDTV. It would be like enlarging a very small negative to a very large photo, and imperfections would be amplified.

Yellow on white is very difficult to read.

[Caiosouza]

jagabo's code in yellow sharpens the video slightly, then resizes to full-frame DVD video, then does a little more sharpening of the resized results. A half-frame 352x576 video might play suitably on a CRT, but it will not look very good on an upsampling HDTV. It would be like enlarging a very small negative to a very large photo, and imperfections would be amplified.

Yellow on white is very difficult to read.

I will not use it, but I thank jagabo, for giving me so many possibilities

I play the videos on standard resolution in a "modern" CRT TV(That was good! ahah), I like interlaced solution on this, leave the LCD to improved HD something

And what do you say about the red type?

PS:

Eu gosto muito do desenho do seu avatar,me traz boas lembranças e ja assisti muitas vezes, fez parte da minha infância !

[jagabo]

jagabo's code in yellow sharpens the video slightly, then resizes to full-frame DVD video, then does a little more sharpening of the resized results.

Doing it like this also gives less aliasing artifacts compared to a simpler resize like Spline64Resize(). But this video is in such poor shape and has so much motion it's hard to tell. It's much more noticeable in cartoons from cleaner sources.

[Caiosouza]

jagabo's code in yellow sharpens the video slightly, then resizes to full-frame DVD video, then does a little more sharpening of the resized results.

Doing it like this also gives less aliasing artifacts compared to a simpler resize like Spline64Resize(). But this video is in such poor shape and has so much motion it's hard to tell. It's much more noticeable in cartoons from cleaner sources.

Yes is effective, but in this case better and get some impurity in disguise?
What do you think?

I really like the softness.

[jagabo]

It's up to you whether you want the sharper picture or not. I was just gaving you an option. Even if you choose not to use it here it may come in useful for some cleaner video in the future.

[Caiosouza]

It's up to you whether you want the sharper picture or not. I was just gaving you an option. Even if you choose not to use it here it may come in useful for some cleaner video in the future.

"Originally Posted by Caiosouza..."...but I thank jagabo, for giving me so many possibilities "

Thanks again!

Tell me something about the kind of red.

[manono]

Hm. I thought DVDAuthorGUI should be able to do that. G

It does. As does Muxman and most DVD authoring programs. That's the first I've heard of an authoring program requiring already muxed sources. However, what's more important is what this TV station for whom this work is being done wants. If it'll take an MPG then Imago is very good for that. If it needs a DVD (and doesn't need a menu), then I'd always go with Muxman.

[Caiosouza]

After several testing, I'm going to filter all file 150 min, was already spoken, but what bitrate to use in HCenc to get decent quality to get everything in 1 DVD9 (Min ,MAx)

And why the line (in Red), if the original file already and interlaced?

I withdraw # --- re-interlace ?

These are my final questions.
And I will use this code:

Import("Drive:\path\to\Avisynth plugins\FixChromaBleeding2.avs")
MPEG2Source("Drive:\path\to\project\1992 MX 250 Sample.d2v", CPU=6, Info=3)
QTGMC(preset="fast")
McTemporalDenoise(settings="low")
ColorYUV(cont_y=-10,off_y=1.0)
FixChromaBleeding2()
ChromaShift(v=4)
MergeChroma(aWarpSharp2(depth=30))
MergeChroma(aWarpSharp2(depth=20))
Spline64Resize(width*2, height)
Crop(2,10,-6,-16).AddBorders(4,12,4,14)
# --- re-interlace
AssumeTFF().SeparateFields().SelectEvery(4,0,3).We ave()

[manono]

And why the line (in Red), if the original file already and interlaced?

Because QTGMC deinterlaced it and made it 50fps.

but what bitrate to use in HCenc to get decent quality to get everything in 1 DVD9 (Min ,MAx)

That's what bitrate calculators are for. You fill in the video length, the final file size, and the audio bitrate, and it spits out the average bitrate. You decide what min and max bitrates you want. Depending on the audio bitrate (mine are usually dual mono and 256), I might use 250/9500 for min/max. Others use different numbers.


And if you're making or already have a menu for it, you have to allow for that, too.

[Caiosouza]

And why the line (in Red), if the original file already and interlaced?

Because QTGMC deinterlaced it and made it 50fps.

but what bitrate to use in HCenc to get decent quality to get everything in 1 DVD9 (Min ,MAx)

That's what bitrate calculators are for. You fill in the video length, the final file size, and the audio bitrate, and it spits out the average bitrate. You decide what min and max bitrates you want. Depending on the audio bitrate (mine are usually dual mono and 256), I might use 250/9500 for min/max. Others use different numbers.


And if you're making or already have a menu for it, you have to allow for that, too.

But I can choose interlaced, time to encode the HCenc, and he give me 25fps
Not?

I intend to pass the HCenc Overall 6500 to 8000MAX , without audio.
Then mux Video M2V and AC3 audio on muxman and have ready to dvd, this correct?

I will not use the menu,
I will use your recommendation and go in Muxman, as you quoted
Thanks for everything!

[manono]

But I can choose interlaced, time to encode the HCenc, and he give me 25fps

Not if you haven't added the reinterlacing portion of the script. Progressive 50fps isn't DVD compliant and won't author for DVD, no matter how HCEnc handles it. You may wind up with a 25fps DVD that has twice the original length.

Where it says:

# --- re-interlace
AssumeTFF().SeparateFields().SelectEvery(4,0,3).We ave()

Weave is supposed to be a single word with no space in it. That's a drawback of the code used here.

I intend to pass the HCenc Overall 6500 to 8000MAX , without audio.

Is 6500 your average bitrate? In my opinion, 8000 is way too low for a max bitrate. Others might disagree. Where did you come up with 6500? Did you just make it up out of thin air? Did you consult a bitrate calculator?

Then mux Video M2V and AC3 audio on muxman and have ready to dvd, this correct?

If you have DVD compliant M2Vs and audio, then they're ready for Muxman. Aren't you creating any chapters for your DVD?

If so, scroll through the script with it open in VDub and choose frame numbers where you might want them. Create a celltimes.txt for muxman like this:

5000
10000
15000
20000
.
.
.

and put the frame numbers for the chapters inside the celltimes.txt. In Muxman add the chapters in File->Import Chapters (something like that).

[sanlyn]

I intend to pass the HCenc Overall 6500 to 8000MAX , without audio.

Is 6500 your average bitrate? In my opinion, 8000 is way too low for a max bitrate. Others might disagree. Where did you come up with 6500? Did you just make it up out of thin air? Did you consult a bitrate calculator?

That is the bitrate I used in the mpeg demos posted here. Earlier the O.P. posted a sample result from the script being used, encoded at ~4000 kbps. Most action shots had the same mosquito noise and other motion problems that appeared on the original video. I used HCenc to encode the results of the script and kept raising the bitrate until motion was smooth and clean. If the video is going to be viewed on a CRT, one could get away with 6000 instead of 6500. The maximum could go to 9000 IMO.

A bitrate calculator for 150 minutes of MPEG2 on a single layer DVD disc will give a bitrate of approximately 3800 VBR/9000 max. That will result in the same problems the O.P. complained about with the original video, made a bit worse by the fact that it will be a re-encode of a poorly encoded, low-bitrate, fast-action half-frame DVD encoded at a 6-hour MPEG1 bitrate. However, at bitrates higher than 3800 VBR, the O.P. will not be able to fit the entire video on a single-layer disc. He will have to split the video into two discs, or split the video onto a double-layer disc. Besides fast action, there is considerable object displacement, wrinkling, and distortion on the original encode. There would be little sense in spending all this effort at recovery, just to end up with the same low bitrate results that the owner started with. The original encoded video was 352x576 @ about 1300 kbps.

[manono]

A bitrate calculator for 150 minutes of MPEG2 on a single layer DVD disc will give a bitrate of approximately 3800 VBR/9000 max.

Okay, but he's already said he wants it on a dual-layer DVD. So that 6500 figure doesn't have any relation to what's needed to do that. Maybe it really is the bitrate needed to fill the DVD with whatever audio he has on there, but that would just be a coincidence as you came up with that figure as the bitrate you thought was needed for some sort of acceptable quality. And he had better make a bunch of chapters so there's a good place to split it for the second layer. Without chapters it'll never be able to be burned to a dual-layer disc.