Is this an interlacing issue?

[jagabo]

When I run the AVI file there are NO artifacts, NO weird lines, nothing. It is clear and smooth.

Not it's not. Pause the video while it's playing and you'll see artifacts like:



Those are different than the artifacts in your first post (because of further processing by Nero) but they are the root of your problem. The fact that opening your AVI in VirtualDub and applying the bob doubler filter shows the same artifacts as your first post proves the above artifacts are in you AVI file.



Use a screen magnifier if you have to. Or use VirtualDub's zoom option on the source pane.

[bvdd]

Yes, I agree the artifacts are definitely there.

As I said , I could live with a DVD that looks like the AVI file, but not with the heavy jagged lines.

What is Nero doing to make it worse? Is there another software that might avoid this?

I tried converting the AVI to VOB files using AVStoDVD, then I tried burning with both Nero and ImgBurn.
Then I converted the AVI to MPEG2 using AVStoDVD, then burned again. Same result.

I know the standard on this forum is high and probably none of you would settle for this quality. But it's good enough for me if I can just get a
DVD that looks no worse than the way the AVI looks on my PC.

Thanks.

[jagabo]

The problem isn't Nero's. It's what you're doing in VirtualDub. If you don't screw up your video in VirtualDub Nero won't cause it to get worse.

[bvdd]

The problem isn't Nero's. It's what you're doing in VirtualDub. If you don't screw up your video in VirtualDub Nero won't cause it to get worse.

What am I screwing up in VirtualDub?

All I did was trim clips and use the AVISynth scripts I got from this forum.

[davexnet]

In the first few posts it seems you started with some footage from a camera. Perhaps if you went over your work flow in a little
more detail we might get somewhere. If your source (presumably good) went into Virtualdub as-is and from that you made a 2-hour
avi, which, as we've seen was bad. It looks as if something you diid in Virtualdub is to blame.

How/where did the footage get resized to 640*480? What else did you do ?

Perhaps provide 10 seconds of the actual camera footage ?

[jagabo]

The problem isn't Nero's. It's what you're doing in VirtualDub. If you don't screw up your video in VirtualDub Nero won't cause it to get worse.

What am I screwing up in VirtualDub?

All I did was trim clips and use the AVISynth scripts I got from this forum.

You've been told this several times now: you resized the video vertically either in VirtualDub or AviSynth. If you don't specify what scripts and filters you used nobody can tell you where you went wrong.

[bvdd]

In the first few posts it seems you started with some footage from a camera. Perhaps if you went over your work flow in a little
more detail we might get somewhere. If your source (presumably good) went into Virtualdub as-is and from that you made a 2-hour
avi, which, as we've seen was bad. It looks as if something you diid in Virtualdub is to blame.

How/where did the footage get resized to 640*480? What else did you do ?

Perhaps provide 10 seconds of the actual camera footage ?

Thanks, Dave.

Here is one script I used that was posted here:
https://forum.videohelp.com/threads/375313-Lowering-AVI-frame-rate-causes-jerky-movement

AVISource("SHORT TEST.avi")
QTGMC(Preset="faster")
Srestore(Frate=19.98)
AssumeFPS(17.982)
Crop(8,0,-8,0)
LanczosResize(640,480)

If I still have the camera footage I'll post it tomorrow.

[davexnet]

I think you must have taken that script out of context, that was designed to improve a digital file that was created from Super 8 cine film. Or did you ? You're the same poster as that other thread .

Post you original footage tomorrow and we'll go from there.

[jagabo]

Here is one script I used that was posted here:
https://forum.videohelp.com/threads/375313-Lowering-AVI-frame-rate-causes-jerky-movement

AVISource("SHORT TEST.avi")
QTGMC(Preset="faster")
Srestore(Frate=19.98)
AssumeFPS(17.982)
Crop(8,0,-8,0)
LanczosResize(640,480)

That is not the script you used on this video. It would have produced fully progressive output at ~17.982 frames per second. The video sample you provided had interlaced frames (encoded progressive) at 24 fps.

[bvdd]

I think you must have taken that script out of context, that was designed to improve a digital file that was created from Super 8 cine film. Or did you ? You're the same poster as that other thread .

Post you original footage tomorrow and we'll go from there.

Yes, I'm the OP in that thread.

My original issue was that the frame rate was too fast. I posted a clip and someone here slowed it down using that script.
He also said at the time, that 17fps was probably still too fast and 12fps might be closer.

So I changed line 4 to (AssumeFPS(12) and used that script to process several clips. That seemed to be correct.

I only had to do that on about 5-6 clips (there were 40+ in all). I worked on them individually and then sequenced them
together for the 2 hour final movie - I did all that in VDub.

I'm still looking for the original captured clips.

[davexnet]

Well there's a sample clip in the other thread called SHORT TEST.avi. Seems to be DV. Is that an original capture?
As jagabo has said, the script you posted to do not explain the 24 fps of the 132MB sample you posted in this thread.

[SameSelf]

My original issue was that the frame rate was too fast. I posted a clip and someone here slowed it down using that script.
He also said at the time, that 17fps was probably still too fast and 12fps might be closer.

So I changed line 4 to (AssumeFPS(12) and used that script to process several clips. That seemed to be correct.

I only had to do that on about 5-6 clips (there were 40+ in all). I worked on them individually and then sequenced them
together for the 2 hour final movie - I did all that in VDub.

I'm still looking for the original captured clips.

Was this videocam footage of film projected at the wrong speed? Just curious.

Judging from the Avisynth script, your original captured clips were 720x480 with 8-pixel-wide black bars on the left and right edges. That is why the Crop(8,0,-8,0) line is there, to reduce the full raster image to 704x480, then resize the 10:11 pixel aspect ratio (4:3 display aspect ratio) to square pixels. This is fine if the video only exists as a digital file since all modern displays are square pixel. However, if you want to burn a DVD, it has to be 720x480 (or 704x480), which means you should not have resized the clips.

Secondly, DVD only supports 29.97 frames per second. Hollywood video is shot at 24 fps, and a Telecine process called 3:2 pulldown is required to convert it to 29.97 fps (not to mention drop-frame considerations). I guess 12 fps would require 6:4 pulldown. I don't know. But are you sure you applied an appropriate pulldown to your 12 fps footage?

My advice is to find the original captured clips. And if you can't, then recapture the video. Then many people here can guide you on the proper workflow starting from the original clips. Seeing that there are some playback speed issues that need to be resolved before burning to DVD, your workflow needs careful scrutiny far beyond vanilla DV AVI.

[bvdd]

Ok, here's the original clip I captured with WinDV. The only thing done to this is that I trimmed it in VDub/Direct Stream Copy.
Maybe this will reveal something?

https://www.youtube.com/watch?v=NRABUxdUpe0&feature=youtu.be

[manono]

No, that's not the original clip. That's a YouTube reencoded clip. Haven't you already been told not to provide YouTube videos as samples? However, you'll notice they deinterlaced it before resizing it, something you could have done but didn't.

[bvdd]

My original issue was that the frame rate was too fast. I posted a clip and someone here slowed it down using that script.
He also said at the time, that 17fps was probably still too fast and 12fps might be closer.

So I changed line 4 to (AssumeFPS(12) and used that script to process several clips. That seemed to be correct.

I only had to do that on about 5-6 clips (there were 40+ in all). I worked on them individually and then sequenced them
together for the 2 hour final movie - I did all that in VDub.

I'm still looking for the original captured clips.

Was this videocam footage of film projected at the wrong speed? Just curious.

Judging from the Avisynth script, your original captured clips were 720x480 with 8-pixel-wide black bars on the left and right edges. That is why the Crop(8,0,-8,0) line is there, to reduce the full raster image to 704x480, then resize the 10:11 pixel aspect ratio (4:3 display aspect ratio) to square pixels. This is fine if the video only exists as a digital file since all modern displays are square pixel. However, if you want to burn a DVD, it has to be 720x480 (or 704x480), which means you should not have resized the clips.

Secondly, DVD only supports 29.97 frames per second. Hollywood video is shot at 24 fps, and a Telecine process called 3:2 pulldown is required to convert it to 29.97 fps (not to mention drop-frame considerations). I guess 12 fps would require 6:4 pulldown. I don't know. But are you sure you applied an appropriate pulldown to your 12 fps footage?

My advice is to find the original captured clips. And if you can't, then recapture the video. Then many people here can guide you on the proper workflow starting from the original clips. Seeing that there are some playback speed issues that need to be resolved before burning to DVD, your workflow needs careful scrutiny far beyond vanilla DV AVI.

This footage was projected at the best speed for reducing flickering. I'm using a Sankyo Dualux 200H, and even with the speed all the way down, some clips were still obviously too fast. Thus the effort to slow them down.

If I resized the clips, I did it unintentionally. I was simply using the scripts that were suggested here. I don't know at what point resizing would have happened.

I don't understand what 'Pulldown' is so, no, I'm not sure about that.

I agree my workflow needs to be carefully planned; unfortunately, that awareness would have really helped me a year ago.
I'm up against a deadline now and I'm trying to exhaust every possibility for getting one watchable DVD copy of this file.
With the soundtrack, I have at least 100 into this project.

The thing that messed me up was that I never saw a problem with the AVI file until I burned it on DVD. It still looks good enough for
this project when played on the PC.

Thanks again for your help and any advice.

[bvdd]

No, that's not the original clip. That's a YouTube reencoded clip. Haven't you already been told not to provide YouTube videos as samples? However, you'll notice they deinterlaced it before resizing it, something you could have done but didn't.

Here's the clip.

I would have done what Youtube did if I had known about the process - but I don't.

[bvdd]

No, that's not the original clip. That's a YouTube reencoded clip. Haven't you already been told not to provide YouTube videos as samples? However, you'll notice they deinterlaced it before resizing it, something you could have done but didn't.

Why would I deinterlace and resize the clip? Are you saying the root of my current issue is that I did NOT do that right after
the footage was captured?

[SameSelf]

This footage was projected at the best speed for reducing flickering. I'm using a Sankyo Dualux 200H, and even with the speed all the way down, some clips were still obviously too fast. Thus the effort to slow them down.

If I resized the clips, I did it unintentionally. I was simply using the scripts that were suggested here. I don't know at what point resizing would have happened.

I don't understand what 'Pulldown' is so, no, I'm not sure about that.

I agree my workflow needs to be carefully planned; unfortunately, that awareness would have really helped me a year ago.
I'm up against a deadline now and I'm trying to exhaust every possibility for getting one watchable DVD copy of this file.
With the soundtrack, I have at least 100 into this project.

The thing that messed me up was that I never saw a problem with the AVI file until I burned it on DVD. It still looks good enough for
this project when played on the PC.

Thanks again for your help and any advice.

Just because something plays fine on your PC does not mean it is suitable for DVD. For example, many people shoot video on their cellphones these days which is VFR = Variable Frame Rate (or almost but not quite 29.97 fps video). That means the speed of the video fluctuates based on the bitrate that the phone was able to handle and very often drops below the target 29.97 fps. Playback of VFR video is a cinch on a PC. However, DVDs require CFR = Constant Frame Rate. So good luck trying to encode a VFR video for DVD. Don't worry about wrapping your head around 3:2 pulldown, that is just an example of how 24 fps film is squeezed into the 29.97 fps DVD spec.

Someone else might chime in with a better advice, but if you have footage that is at some random frame rate that was chosen based on flickering (as opposed to a target delivery format like DVD), I would say, forget about DVD unless you are willing to live with the wrong speed. Or delete the messed up footage from the DVD. But don't blame yourself, DVD is a very restrictive format. We all have to fight hard to get our footage onto DVD without mangling it in the process. Sometimes it is just not worth the effort.

[bvdd]

Just because something plays fine on your PC does not mean it is suitable for DVD. For example, many people shoot video on their cellphones these days which is VFR = Variable Frame Rate (or almost but not quite 29.97 fps video). That means the speed of the video fluctuates based on the bitrate that the phone was able to handle and very often drops below the target 29.97 fps. Playback of VFR video is a cinch on a PC. However, DVDs require CFR = Constant Frame Rate. So good luck trying to encode a VFR video for DVD. Don't worry about wrapping your head around 3:2 pulldown, that is just an example of how 24 fps film is squeezed into the 29.97 fps DVD spec.

Someone else might chime in with a better advice, but if you have footage that is at some random frame rate that was chosen based on flickering (as opposed to a target delivery format like DVD), I would say, forget about DVD unless you are willing to live with the wrong speed. Or delete the messed up footage from the DVD. But don't blame yourself, DVD is a very restrictive format. We all have to fight hard to get our footage onto DVD without mangling it in the process. Sometimes it is just not worth the effort.

Thanks.
So are you saying that because my captures are VFR they cannot be encoded to DVD - and that's why I'm seeing heavy comb artifacts on the DVD?

I can't delete the messed up footage from the DVD because the whole 2 hr movie has the same problem artifacts. Every scene, including the clips that I simply trimmed and did no other processing to (that's most of the clips).

So whatever is happening in the burn process to cause the artifacts, is happening to every clip throughout the film.

Is there a way to convert the clips or the full AVI to CFR?

I should add, the audio track is not an issue. It was added after the video was done.

I can also live with the speed being too fast or too slow if I have to.

[SameSelf]

No, that is not what I am saying. You have CFR video if I understand correctly which means you have something along these lines:

1. Super 8 video that was shot at 12 frames per second
2. Super 8 video that was projected at a faster frame rate, perhaps 18 frames per second
3. Video of the projected footage at the faster frame rate recorded at 59.94 fields per second
4. DV AVI captures of #3 which are 59.94 fields per second

The complexity makes the plot of Inception look trivial!

So when you play the captured DV AVI, it is the same speed as what was projected. IOW it looks sped up. Now you can tell an decoder to slow down the playback using Avisynth which you have done using Srestore() and AssumeFPS(). But you no longer have 59.94 field per second video do you? Think about it.

How you would turn that into NTSC DVD spec at the proper playback speed is beyond my pay grade.

[manono]

Here's the clip.

Finally. For nine days we've been waiting for that. And there's nothing wrong with it. Treated properly you won't get the artifacts as seen in the earlier pictures. Since there are lots of blended fields and since the source was progressive home movies originally (right?), I'd do it this way:

QTGMC()
SRestore(Frate=19.98)

If you don't believe 19.98fps to be the correct framerate, you can slow it to whatever speed it's supposed to be with an AssumeFPS(xx.yyy) line in your script after the SRestore line.

But, with 3:3 pulldown 19.98fps can be used in an NTSC DVD with no problems. Something lower and you'll have to make other adjustments.

[jagabo]

Here's the clip.

Finally! If you open that AVI file directly in VirtualDub and apply the bob doubler filter you see it doesn't have the artifacts of your image in the first post of this thread.

That film was projected at about 20 frames per second then recorded by the camcorder as 29.97 interlaced frames per second.

You could simply encode that video as interlaced MPEG 2 and burn it directly to DVD. My guess is if you import the AVI into Nero it will handle it correctly when it makes a DVD. Almost all editors know how to handle DV AVI.

If you want to convert it to all progressive video in AviSynth you need something like:

AviSource("OrigClip.avi") 
ConvertToYV12(interlaced=true)
QTGMC(preset="fast") # or the faster, not as accurate Yadif(mode=1, order=0)
SRestore(frate=19.98) # eliminate blended and duplicate frames

That will leave you with progressive frames which can be encoded with 3:3 pulldown flags for DVD. But I think the video needs to be slowed to 16 or 18 fps to get natural motion. Rates that low can't be burnt to DVD with soft pulldown, they'll need hard pulldown, or duplicate frames.

There are also some aliasing problems and other issues that you might consider fixing. Once the video has been restored to progressive frames (ie, after SRestore()) you can address those if you want. For example

AviSource("OrigClip.avi") 
ConvertToYV12(interlaced=true)
QTGMC(preset="fast") # or the faster but less accurate Yadif(mode=1, order=0)
SRestore(frate=19.98) # eliminate blended and duplicate frames
Spline36Resize(width/2, height)
Spline36Resize(width*2, height)

Gets rid of most of the aliasing.