The DVD saga continues

TomG wrote on 7/16/2003, 9:38 PM
Last month I posted questions regarding conveting 16mm footage by a lab to a DVD . vob which I then took into dvd2avi and then into V4 for editing and back out to a DVD via DVDA. It took some time but I finally found a process that worked. Or so I thought. I then had another 2.5 hours of footage converted. This time the DVD burned from DVDA is not so good. There appears to be a lot of "vibration" when viewing people action shots.

Everything is the same regarding all settings. Even the media I am using is the same. The rendered output from V4 uses the DVD NTSC format. I have spent 2 days trying to isolate the problem. I finally ran the process against the two original DVDs from the lab. I followed the same procedure:

1. Copy .vob file to HD
2. Run .vob file through dvd2avi
3. Run the .dv2 file through vfapiconv to get .avi markers
4. Run .vfapiconv file into V4
5. Render 1 minute of footage to create .mpg
6. Run .mpg into DVDA to burn DVD

The first project came out just fine (as it did originally) even on a 54" screen. But the second project had the jitters, all 2.5 hours!!!! When I view the 2nd project on my monitor, it looks just fine (even from the DVD). But when I get it into the Toshiba DVD player and view it on the 54" screen, it's so bad it make you dizzy.

If seems like the problem is with the source DVD created by the lab. I know the guy and he swears that he uses a standard process in converting film to DVD (and I believe him). Both of the DVDs he provided me with look fine when played through my Toshiba DVD player.

Does anyone have any other ideas on what's happening here? I've racked my brain trying to find this problem. I thought maybe I had a sampling or interleaf problem but now I don't think so.

Thanks in advance for any ideas you may be willing to share.

TomG

Comments

BillyBoy wrote on 7/16/2003, 10:02 PM
I would rule out Vegas as the source of the problem by repeating the steps using TMPGENC and Flask.
TheHappyFriar wrote on 7/16/2003, 11:26 PM
Is the footage from the lab progressive or interlaced (most likely interlaced)? Also, is it 24fps or 29.97 (30 maybe?)? They didn't render it for Pal instead of NTSC or vice versa? It sound like an interlace problem. Did you save some frames that are next to eachother and compare them? Wierd problem.
TomG wrote on 7/17/2003, 5:46 AM
The footage from the lab is interlaed at 29.97fps and was rendered in NTSC (or so I was told). If I had an interlace problem, would the only point this would be visible be after the final DVD burn?

I have not attempted the flask tmpgenc route yet although I do have flask. Billyboy, how does this work? Is it a similar sequence to the dvd2avi/vfabiconv process? Is there an article on how to do this somewhere on doom9?

Thanks,
TomG
JJKizak wrote on 7/17/2003, 7:42 AM
When you find olut what the problem is let me know how to fix it. I had 11,000
ft og 16mm film transferred to beta sp tape and if I capture it
and render it from the Vegas timeline it gets the "judder" you talk about. If I
put it into DVDA after capture (not from timeline) it is perfect.

JJK
TomG wrote on 7/17/2003, 8:01 AM
JJK, sounds like you experienced the same problem I am having. I guess the right "technical" term is judder. When you said it was perfect after putting it into DVDA, where exactly was that in your process? What "capture" point were you referring to?

I still can't explain why I had a problem with one DVD and not the other, even though they were converted by the same lab using the same process. I hate to think I blew $500 converting the film to DVD and not being able to edit it!!!!

I'll keep you posted if I discover anything new.

TomG
mikkie wrote on 7/17/2003, 8:21 AM
FWIW, possible tests I think I'd try:

1) Take the VOB file, as is, use DVD shrink if nec, or more accurately perhaps, just cut a portion for testing using tools from doom9, digital-digest etc. Burn to dvd-rw or +rw, see if the prob is there from the start.

2) If film is originally involved, and source is to be NTSC DVD, footage is digitized at 23.976p, then flags are usually inserted in mpg2 that instruct player to maintain constant 29.97, as well as pull down instructions regarding which frames to duplicate. Might check out original vob to see what flags were included in the orig. mpg2. Assume that when you used DVD2AVI you checked for frame versus NTSC 29.97i - setting force film should remove any of the above flags.

3) Assuming you've got separate streams for DVDA, send the mpg2 through restream to check & possibly alter any flags - try a portion of that on a rewritable DVD.

Have seen this problem where the player didn't like the pulldown flags inserted for a 23.976 mpg2. Could see how it might come about if the original is actually 23.976 with NTSC flags as is common on most any commercial DVD with film source.

Have also seen it as the result of converting to 29.97 - this might give some insight into the process: http://www.cs.tut.fi/~leopold/Ld/FilmToVideo/ - Doom9.org also has a guide on this stuff, but to me it was less straightforward, a bit more work to get one's head around.

Last, perhaps most obvious so I only mention it in passing - probably checked it already so please don't be offended - if interlace order was reversed on encoding, even if only pulldown flags...
TomG wrote on 7/17/2003, 8:55 AM
Thanks, Mikkie, for the valuable information and suggestions.

Don't know why or how, but just for grins I ran one more render. This time I used the PROGRESSIVE SCAN OPTION instead of the interlace. The results were stunning. The judder was gone!! Don't know why this was the case. Is it something done in the lab when the film was converted? The technical "nuts and bolts" of digitizing often overwhelms me. I still have some bounce in the shots but this is due to a shaky camera. Isn't there a filter or plug-in that you can use to reduce this?

But the good news is that I think my problem is solved. I am going to call the lab and explain what I found and see what they think.

Thanks again for reminding me about the scanning options. Trouble is, you just don't know when to stray from the "boilerplate" templates that V4 gives you.

TomG
JohnnyRoy wrote on 7/17/2003, 12:09 PM
> If I had an interlace problem, would the only point this would be visible be after the final DVD burn?

Yes, actually the only place you’d see this is on an interlaced TV. That’s why the DVD looked fine on your computer. Your computer monitor is not interlaced. Based on the fact that rendering it as progressive scan fixed the problem I would say that interlacing was the issue. Perhaps this happened with you captured it? This just means the video was interlaced either top field first and rendered as bottom field first or visa versa. The simple solution would be to render it as the other type (i.e., if the template says top select bottom) and you should see the jittering stop. As I said, bypassing interlacing all together by rendering as progressive had the same effect. I don’t know why one DVD would have worked while the other didn’t.

~jr
TomG wrote on 7/17/2003, 12:17 PM
Thanks, JohnnyRoy

I did render the file interlaced from bottom and then from top and saw no difference. It wasn't until I went Progressive (no interlace?) that things looked OK. I guess when the film was actually "captured" was at the lab? As I mentioned before, I followed the same procedure of DVD2AVI and VFAPIConv for both .vobs but the results were different. The first one sailed through Interlace-bottom-first with no problem.

TomG
nolonemo wrote on 7/17/2003, 3:03 PM
There is a software package that you can use to remove camera shake (or tilted horizons) from video (AVIs and I'm not sure if other formats), you will lose a little around the edges of the frame. I tried the demo and it seemed to work fine. Unfortuneately, I can't remember the name for the life of me.
JohnnyRoy wrote on 7/17/2003, 4:12 PM
Tom,

Wow that's weird about how both interlace options looked the same. If they interlaced the video on the DVD one should have worked while the other should not.

For the "camera shake" software, I would try SteadyHand by DynaPel. It works really great and they have a demo version. While you’re there download their free VideoScope and check out the insides of your AVI file. ;-)

~jr