AviSynth to fix laser rot...

[Guest]

Archived from groups: alt.video.laserdisc (More info?)

I have been hearing about doing this. How exactly would I go about it?
Never used AviSynth, but I am very experienced with other video software
and have even programmed a few of my own video utilities when needed.

-Mike

 

[Guest]

Archived from groups: alt.video.laserdisc (More info?)

half_eaten wrote:
> I have been hearing about doing this. How exactly would I go about
it?
> Never used AviSynth, but I am very experienced with other video
software
> and have even programmed a few of my own video utilities when needed.

Karyudo is the advisor for this job. Do a google search, find his
e-mail address, and write to him. I essentially did this for capturing
AC-3 to DVD, and he was very courteous, helpful, and patient with my
questions. Eventually, my project was successful.

-Junior

 

[Guest]

Archived from groups: alt.video.laserdisc (More info?)

That's a pretty good theory. The only problem is that with an analog source
it's practically impossible to capture something with the exact same frame
count, let alone have them all match up right. Even if you edit them to
start start at the exact same frame. Somewhere along the line, whether
it's the 100th, 1000th, or 10000th frame... you will get a desync.

The only way you could possibly do this would be to have CAV discs, and
capture each frame manually. While this would work it would be absolute
hell.

Or you could go through each capture frame by frame comparing the images
by eye and fixing desyncs like that. But that would be equally hellish.

Let me know if I'm wrong though...

-Mike

 

[Guest]

Archived from groups: alt.video.laserdisc (More info?)

On Mon, 20 Dec 2004 01:32:00 -0500, "half_eaten"
<half_eaten@hotmail.com> wrote:

>That's a pretty good theory. The only problem is that with an analog source
>it's practically impossible to capture something with the exact same frame
>count, let alone have them all match up right. Even if you edit them to
>start start at the exact same frame. Somewhere along the line, whether
>it's the 100th, 1000th, or 10000th frame... you will get a desync.
>
>The only way you could possibly do this would be to have CAV discs, and
>capture each frame manually. While this would work it would be absolute
>hell.
>
>Or you could go through each capture frame by frame comparing the images
>by eye and fixing desyncs like that. But that would be equally hellish.
>
>Let me know if I'm wrong though...

OK: you're wrong.

What I posted last time is not theory; I've done it. For three
feature-length films, no less.

Sure, to use TooT to best advantage you should capture with no frame
drops, but even with my Athlon 1700+, that's easy. Just checked, and
one of the longest sides I capped was 51,700 frames -- no desyncs,
times three. Times four sides, times three movies. On the order of
600,000 frames, and no problems.

If for some odd reason you weren't able to get caps with no drops,
then you can still use VDub to find the frame drops (there's a button
for going to the next dropped frame) and just duplicate it in
AviSynth. That one frame might not get TooTed right (it'll still be
averaged between the two good copies, though), but that's no biggie.
The end result will still be near-perfect.

Lop off a few frames from the starts to get all three caps (for a
given LD side) lined up, and you're in business. With AviSynth you can
use StackVertical and/or StackHorizontal to confirm all the frames are
the same before you TooT them.

 

[Guest]

Archived from groups: alt.video.laserdisc (More info?)

half_eaten <half_eaten@hotmail.com> wrote:
: That's a pretty good theory. The only problem is that with an analog source
: it's practically impossible to capture something with the exact same frame
: count, let alone have them all match up right. Even if you edit them to

Mike,

you're incorrect. I haven't captured multiple times from LaserDiscs, but I did
capture 2 times from 2 VHS of the same footage using 2 different VCRs. I needed
to do that 'cause 1 VHS had a better image and another one had a better sound.
I used 2 different VCRs 'cause, as you might guess, 1 VCR also produced a better
picture and another one produced a better sound.

The frame count was *precisely* the same over 87 minutes of non-stop footage.
The only thing that was different - was the sound. The sound was off by
120ms over the long run. This is due to VCR's slightly different speeds.

: it's the 100th, 1000th, or 10000th frame... you will get a desync.

Nope.

--Leonid

PS. You may ask how I know that the frame count was the same. Loading 2 AVI
files into VirtualDubs and comparing a frame in the beginning between 2 and
then at the end gave exactly the same frame count result.

 

[Guest]

Archived from groups: alt.video.laserdisc (More info?)

On 21 Dec 2004 00:12:22 GMT, Leonid Makarovsky <venom@csa3.bu.edu>
wrote:

>half_eaten <half_eaten@hotmail.com> wrote:
>: That's a pretty good theory. The only problem is that with an analog source
>: it's practically impossible to capture something with the exact same frame
>: count, let alone have them all match up right. Even if you edit them to

Yep, that's been my experience as well, but maybe I need to scrap my
procedure and start from scratch.
>
>Mike,
>
>you're incorrect. I haven't captured multiple times from LaserDiscs, but I did
>capture 2 times from 2 VHS of the same footage using 2 different VCRs. I needed
>to do that 'cause 1 VHS had a better image and another one had a better sound.
>I used 2 different VCRs 'cause, as you might guess, 1 VCR also produced a better
>picture and another one produced a better sound.

I've never seen two tapes of the same title that would play or capture
exactly the same on the same VCR, much less two different tapes
on two different VCRs and I've had to do it quite a few times with
two, sometimes three different tapes to achieve the desired footage.
I'm curious about the equipment used.

And the same applies to Laserdiscs, just to keep it on topic. <g>
If a frame is dropped in one of the three of multiple captures, how
does it affect the sequence in the encoding process. Does it just
ignore the dropped frame and choose the best of the other two?

>
>The frame count was *precisely* the same over 87 minutes of non-stop footage.
>The only thing that was different - was the sound. The sound was off by
>120ms over the long run. This is due to VCR's slightly different speeds.

How can the frame count be that exact and the sound be off 120ms
(which is a little over 4 frames) if the VCRs are running at different
speeds? Not being sarcastic or cynical, but there are a lot of
variables and you seem to lead somewhat of a charmed life when
it comes to video.
>
>: it's the 100th, 1000th, or 10000th frame... you will get a desync.
>
>Nope.

Wasn't there a scientific study done on how much difference in
A/V sync the human eye could detect? Anyone else seen that?
>
>--Leonid
>
>PS. You may ask how I know that the frame count was the same. Loading 2 AVI
>files into VirtualDubs and comparing a frame in the beginning between 2 and
>then at the end gave exactly the same frame count result.

Mike

 

[Guest]

Archived from groups: alt.video.laserdisc (More info?)

On Mon, 20 Dec 2004 21:40:22 -0600, Michael D
<michaeld_54@NOSPAMyahoo.com> wrote:

>If a frame is dropped in one of the three of multiple captures, how
>does it affect the sequence in the encoding process. Does it just
>ignore the dropped frame and choose the best of the other two?

Yes. You can even run TooT on three caps that have frames missing
(dropped frames), but that's sort of dumb. Remember, TooT is just
comparing every pixel of every frame, and picking the two that are
closest together and averaging them. The results wouldn't necessarily
remove the rot and other disc artifacts, although in general you'd
still get some noise reduction due to averaging the two caps that were
on the same frame -- or closest to it.

The better way to do it (if you insist on having any frame drops at
all...) is to duplicate a frame where the drops are, so all the frames
you did cap correctly line up right.

In other words:

12345578
12345678
12335678

....is better than:

12345789
12345678
12356789

Don't worry: finding dropped frames is super-easy with VirtualDub
(among other tools).

 

[Guest]

Archived from groups: alt.video.laserdisc (More info?)

On Tue, 21 Dec 2004 06:57:46 GMT, Karyudo
<karyudo_usenet@yahoo.com.remove.me> wrote:

>On Mon, 20 Dec 2004 21:40:22 -0600, Michael D
><michaeld_54@NOSPAMyahoo.com> wrote:
>
>>If a frame is dropped in one of the three of multiple captures, how
>>does it affect the sequence in the encoding process. Does it just
>>ignore the dropped frame and choose the best of the other two?
>
>Yes. You can even run TooT on three caps that have frames missing
>(dropped frames), but that's sort of dumb. Remember, TooT is just
>comparing every pixel of every frame, and picking the two that are
>closest together and averaging them. The results wouldn't necessarily
>remove the rot and other disc artifacts, although in general you'd
>still get some noise reduction due to averaging the two caps that were
>on the same frame -- or closest to it.

I agree that it would be a nightmare to work with three caps, each
with it's own set of dropped frames.
>
>The better way to do it (if you insist on having any frame drops at
>all...) is to duplicate a frame where the drops are, so all the frames
>you did cap correctly line up right.

Thanks I was just curious about the progression if
frames were dropped in one of the three captures. Most of
the films that I convert are from the '20s and '30s and it's
would be difficult to find two copies that were consistent enough
to consider using this.
>
>In other words:
>
>12345578
>12345678
>12335678
>
>...is better than:
>
>12345789
>12345678
>12356789
>
>Don't worry: finding dropped frames is super-easy with VirtualDub
>(among other tools).

Yes, I'm pretty familiar with VDub and scripting with AviSynth, but
I appreciate you weighing in with your thoughts. : ) I'm off to
Donald Graft's site to read more.

Mike

 

[Guest]

Archived from groups: alt.video.laserdisc (More info?)

Michael D <michaeld_54@nospamyahoo.com> wrote:
: I've never seen two tapes of the same title that would play or capture
: exactly the same on the same VCR, much less two different tapes

Yes 2 tapes of the same title. One was sealed in shrink wrap, and another one
was rather used. But the one that was used had crisper sound though so I
decided to use it for audio.

: on two different VCRs and I've had to do it quite a few times with
: two, sometimes three different tapes to achieve the desired footage.
: I'm curious about the equipment used.

Capture card FlyVideo 3000FM which is Philips based. Capture in Huffyuv avi
format. Terratec Aureon 7.1 Universe soundcard.

: And the same applies to Laserdiscs, just to keep it on topic. <g>
: If a frame is dropped in one of the three of multiple captures, how
: does it affect the sequence in the encoding process. Does it just
: ignore the dropped frame and choose the best of the other two?

I didn't have any dropped frames. But in case the frame is dropped, it is
replaced with the previous frame.

: How can the frame count be that exact and the sound be off 120ms
: (which is a little over 4 frames) if the VCRs are running at different

In this project I was working with 2 PAL tapes, so it was precisely 3 frames.

: speeds? Not being sarcastic or cynical, but there are a lot of
: variables and you seem to lead somewhat of a charmed life when
: it comes to video.

That's the miracle of it. The capture card grabs frames from the VCR. The
sound card samples the analog sound. So if one VCR is running at precisely
29.976 frames per seconds and another one runs at 28 frames per second,
what's going to happen? Your capture card will be grabbing frame by frame
and store them on your computer at 29.976fps in both cases. It will be grabbing
frames at slower or faster speed, but it will be storing them at the
correct speed. But what's gonna happen to the sound? It will be sampled
differently depending on speed of the VCR. Remember soundcard has now idea
about the capture card.

However, when you capture from LaserDisc and using digital sound in, a sound
card doesn't sample anything. It just copies digital data bit by bit. Frames
and sound samples here are descrete, so you will not have to worry about
synchronizing.

: Wasn't there a scientific study done on how much difference in
: A/V sync the human eye could detect? Anyone else seen that?

I can notice 60ms.

--Leonid

 

[Guest]

Archived from groups: alt.video.laserdisc (More info?)

I guess you are right..

Well, I could give it a shot. I have two copies of Mortal Kombat on LD. I
have a Hauppauge WinTV card I'll try it with. I could put together some
software dedicated to fixing laser rot. That would be a nice tool to have!