More capture/conversion (1 Viewer)

[Patrick Sun]

Before I found that Panasonic codec, I was pounding my head against the monitor last night...not a pretty sight.

I am going to do one of these conversion projects on my 3 year old PC to get a handle on the amount of time it took, and then do it on my new PC setup (still waiting on my video card to show up before I build it) and see if the extra horsepower made a difference. Plus, I'm working my way around the learning curve too.

 

[Gary_E]

Here are some captures from a current DVD project I just completed. These images are from 8mm home movies that were restored by Movies-2-DVD in Naples Florida. I used the mpeg-2 files to create the DVD. The results are quite amazing considering the age and condition of the original films.

I'm limited to 150kb per image, so the captures are not a true representation of the finished DVD but you'll get the idea.

http://home.bellsouth.net/p/s/commun...iew=thumbs&ck=

Regards,
-Gary

 

[Wayne Bundrick]

I believe "dvsd" is the four-letter AVI code for DV (firewire) video.

 

[Ken Chan]

There was a pretty good list at fourcc.org, but it appears to be down. Here's another.

//Ken

 

[Patrick Sun]

Okay, I have a problem with the audio in the final VOB files for the DVD that I'm authoring (backing up a 120 minute LD).

I've captured to AVI using a Datavideo DAC-100 via firewire/WinDV.

I have an AVI files around 27GB in size, the audio is embedded in the AVI file (plays normally).

I stripped out the audio with VirtualDub, resulting in a Wave file. I verified that the Wave file has the audio from the LD. I also used VirtualDub to combine the AVI files into the big 27GB AVI file.

I used TMPGEnc to crunch the video portion into a M2V files (it'll fit on a 4.7GB DVD). I'll have to play around with the bitrate to better utilize all of the DVD (I made one that was too big, and the current one has 500MB unused).

I used BeSweet/BeSweetGUI to convert the Wave file to an AC3 file. I used WinAmp5 and the AC3Filter to verify that the audio from the AC3 is present when it got created from the Wave file.

Then, the Muxing portion is where I guess I am having problems.

I know DVDLab is recommended for authoring, but being a little cheap and adverturous, I'm using IfoEdit 0.96 to do the muxing of the M2V file and the AC3 file. Once it creates the VOBs and the IFO files (I have them created in a directory "Video_TS" for burning later), I then have it do the "Get VTS Vectors" to make the IFO compliance with standalone DVD players. (I didn't do this initially, and the DVD I burned was unplayable in the DVD player). Now I did the VTS vector step, and the DVD will play, but all I get is the video. I get no audio at all.

The VOB files I play on the PC also have no audio embedded in them either.

So, if there something else I'm not doing to get the audio muxed into the final VOB with IfoEdit?

 

[Patrick Sun]

Okay, perhaps the AC3 audio file I've created just isn't up to snuff.

I gone ahead and muxed the M2V and the AC3 files using DVD Lab, and I still don't get any audio when I play it (but the video plays as expected).

Here are the audio attributes (using IfoEdit's Play DVD feature):

Audio Application Mode=DVD_AudioMode_None
Audio Format = DVD_AudioFormat_AC3
Multichannel Extension = false
Number of Audio Channels = 2
Sampling Frequency = 48000
Sampling Quantization = 16
Language Code LCID = 1033
Language Code Extension = DVD_AUD_EXT_NotSpecified

Is the first line telling me that my AC3 file isn't a valid audio file to be muxed into a VOB?

 

[Patrick Sun]

Okay, crisis over. For some reason, BeSweet wasn't making compliant AC3 files. I found out that another AC3 encoder frontend, FFMpeg creates AC3 files that can be muxed together to make VOBs with sound to them.

 

[Patrick Sun]

Okay, an update:

I got things working, though I need to re-do the chapter points (the time shown on my standalone dVD player is a little off from the DVD player in IfoEdit).

My reaction to the video quality is probably more to do with my expectations. I liked the video quality from the AVI file I captured, but the encoded M2V file is sort of lacking, the video looks a bit ragged and obscure in detail (pixelly) in some parts depending on what got encoded in the frame.

I used these encoding parameters (TMPGEnc): 2-pass VBR 1800-4500-9100 for the 121 minute run time of the LD. It took over 14 hours at the highest quality setting, I also did one at the 2nd highest quality setting, took around 7-8 hours, and the video quality difference between to the settings perceived by my eyes is minimal at best. Just for laughs, I am in the process of using the lowest video quality setting, and it's still taking 6 hours to do it.

My old PC took over 42 hours to encode at the highest quality setting for the same length AVI. (And I was too optimistic in my bitrate settings, and wound up with a M2V that's too big anyhow to use for a single DVD).

Is there anything I can do to get the video quality looking more like the AVI file I captured (besides splitting the movie onto two DVDs)?

I may back down the AC3 bitrate from 384 to 224 (which gives me just a little more room to bump up the average video bitrate setting to 4650, which probably also won't make that much difference.

I'm not sure if the menu creation aspect of DVDLab is enough to warrant buying it, though I did play with it, and got a feel for how things needed to be set up to incorporate the menu into the DVD authoring process. The "off-the-path" stuff, creating the menus from scratch will be my next endeavor soon.

 

[Ken Chan]

Some people say they get better results with TMPGEnc by using Constant Quantization (CQ). The problem with that is that you can't predict the final file size. I also recall someone mentioning that setting the "B picture spoilage" to zero helps -- although I'm not sure exactly what that means

//Ken

 

[Patrick Sun]

I'll try the B Picture Spoiler setting. I've also created a 224kb AC3 file, and bumped up my VBR setting to 1000-4900-8000 on my latest encoding effort. We'll see how it looks in just over 8 hours...

 

[Patrick Sun]

Wayne suggested:

There is one more step that I'm doing before encoding to MPEG. I convert the file from Microsoft DV AVI to Canopus reference AVI, using a utility I downloaded from Canopus. It doesn't actually convert the video data, it just changes the AVI signature to Canopus and copies the data into its own layout. Then the AVI is no longer decoded by the lame Microsoft codec, instead the much better Canopus DV codec (a free decode-only version is also available for download) becomes responsible.

I tried this and while I was surprised that the conversion creates a bunch of split up files from the original AVI file, it appears that this gets rid of the weird vertical striations that I was getting by using the Microsoft AVI codec. Now my end product looks closer to what I captured. Thanks for the tip.

BTW, here's the link to the Canopus site for the utilities.

 

[Wayne Bundrick]

Like I said, the Microsoft DV codec is lame.

Yeah I didn't mention that a Canopus reference AVI consists of a smallish .AVI file which only contains reference pointers to multiple .nnn files containing the actual video, with each one up to 4GB in size. It's Canopus' way of getting around the 4GB file size limit that AVI used to have. There are some other benefits to using reference AVI files but they generally require you to use Canopus editing products to take advantage of them.

 

[Patrick Sun]

I also noticed that the Canopus AVI file was a little brighter (I usually bump up the Contrast, and I subtract some of the redness out of the final TMPGEnc encoded version).

Now I'm playing around with splitting up the 2-hour films into 2 discs and bumping up the encoding bitrate to around 8000 (the single disc version was set to around 4750 via 2-pass VBR) to improve the video quality.

 

[Holadem]

Patrick, did you ever find an uptimum setting for MPEG2 encoding with TMPGEnc?

I am quite unhappy with the obviously interlaced and pixelated MPEG2 files I am getting.

--
H

 

[Patrick Sun]

Well, here's the final post in my thread/adventure in finding a good set of settings for turning interlaced video (from a film source 24 fps) to a progressive MPEG-2 video file.

If you needed more clarification, you'd have to wait for a reply by me until after work today because my setup is on my home PC, but perhaps others will chime in as well.

 

[Holadem]

Thanks. I wish I could show you guys some "before" and "after" clips, but the files would be too large, even for the 2 minute clip I am expriementing with. That would allow me to see if it's my expectations getting in the way.

In any case, playing the tape from my camcorder, directly plugged into the TV, looks MUCH better than the picture I get from the encoded file playing in my el-cheapo Magnavox DVD player (I wonder if that is also a factor).

--
H

 

[Jeff Jacobson]

I'm capturing video with a Canopus ADVC-100 (using WinDV). Would this step benefit me as well?

 

[Ken Chan]

obviously interlaced

If this is your Digital8 stuff, it is interlaced, and probably should stay that way. It may not look good on a computer monitor, but when you play it on TV, it should be correct.

If you see visible stuttering during motion -- kinda like two steps forward, one step back -- you've got the field order wrong. DV is "bottom field first".

 

[Holadem]

If this is your Digital8 stuff, it is interlaced, and probably should stay that way. It may not look good on a computer monitor, but when you play it on TV, it should be correct.

It isn't.

I have left everything interlaced, I don't even have a progressive display.

I got the "bottom field first" right. I have no problems with motion. The picture itself sucks, I can see what I can only describe as "scan lines". And there are these EE things around fine objects, except it's obviously digital/bloky.

Dunno what do do at this point. Keep experimenting I guess.

--
H

 

[Holadem]

OK, Question:

Does anyone know where I could download MPEG2 samples from?

They MUST be compressed from a camcorder's DV-AVI file, for comparison's sake. I know the quality of the camcorder is going to be a factor, but I don't expect dramatic differences - such a file should give me an idea of what a MPEG 2 file properly encoded froma camcorder should look like.

--
H