Digital Video

Pinnacle StudioDV
DV codecs
Win98SE capturing solution
Field order problem

I will only cover some of the most important aspects of DV here. For a comprehensive guide, I recommend to visit puremotion.com.

Camcorders and tape systems with analog recording have several disadvantages with quality.

Mainly these are time jitter, noise and dropouts.

Many ingenious solutions went into the first home VCR systems. These achieved a quality that was good enough for viewing. However even the 2nd generation copy of such a tape yet looks bad, further copies look awful. If one wants to edit such recordings, he therefore faces huge quality losses. Nothing has changed with this, since the 1st VHS machines came out 20 years ago. The signal processing of all analog tapes has principally remained the same except of some very little improvements. This applies to S-VHS, Beta, Hi8.

DV records the signal in a digital format. This way, tape noise is no issue any more. Time base errors are completely corrected by a digital buffer.
The best thing is that the digital signal can directly be transferred to a computer by a simple serial interface, can be cut and pasted without re-encoding (except when effects are applied), and then be rewritten to tape without the slightest quality loss.

Computer based editing solutions for digital are many times cheaper than for analog, because the hardware, an IEEE1394 board, is very simple.
Complete solutions with board and basic editing software are now below 100$ and cheap large harddrives make it possible to store and edit several hours of video. All this in broadcast quality (given a tape shot by a good cameraman).

Let me remark that digital signal is compressed by 5:1, each frame separately. This means a certain quality loss with the first compression, but it's so small that I don't think it matters. In fact it's the only loss once and forever, as long as tapes are only cut and copied. If really one recompression is necessary due to effects etc., the losses are still hardly visible. Compared to the generation loos of analog tape, they are not worth mentioning.

Formats

There is only one signal format but 2 tape formats: MiniDV and Digital8. Only the mechanical dimensions of the cassettes are different. Digital8 uses Hi8 sized cassettes, MiniDV uses DAT sized cassettes that are smaller and allow to build smaller camcorders.
The advantage of Digital8 is that the camcorders can play 8mm and Hi8 videos as well. They even do that better than the old machines, because D8 camcorders have a 'big' head drum like high end Hi8 devices, and the output is already digital and time base stabilized. This way you can already make a copy of an Hi8 tape without quality loss.
If you buy Digital8 or MiniDV depends on your wallet and on the question if you have many old Hi8 tapes. The only advantage of MiniDV is that the devices are smaller. They are not that much lighter however and at the moment they cost twice as much for no really obvious reason.

System Requirements

Pinnacle Studio DV

This reflects only my personal opinion. I'm not able to test each system on the market so I've gathered as much information as I could get and decided to buy the Pinnacle StudioDV because of its unique software and its fairly low price (<100$ street price was what I had to pay). A lot of info about it can be found at the Pinnacle Webboard (section: consumer webboard), and also on Mike Shaw's pages.

StudioDV can recognize scenes on the fly because of the time stamp information in DV data. It assembles a storybook where each scene is represented by one thumbnail image.
Then you simply drag these to the timeline as you want, trim them separately, insert transitions and play the result back to tape. The advantage is that this 'timeline' is not continuous like in other programs where you get lost trying to find seconds long scenes in hours of footage. Here each scene is one thumbnail image, be it 1/10 or 1000 seconds long. You get a linear display of each scene by double clicking on the thumbnail image and then you can trim just this scene.

Scenes can also be repeatedly used, single frames can be used and displayed for any desired length, audio can be dubbed and mixed in etc.

It's not as powerful as Premiere or Media Studio, but much more easy for beginners and in my opinion, also much more easy and useful than the Video Studio that comes bundled with most cheap cards.
Another advantage is that capturing is done in DV type 2 format, which can be read by VirtualDub and Tsunami and also Adobe Premiere without further conversion. There is no file size limit if you use Windows2000.

A problem is that StudioDV apparently recompresses the whole footage if output is directed to disc. Going over tape is a solution for this but not quite elegant. Also MPEG1 and RealVideo compressors are integrated, but most VFW codecs (like MPEG4) cannot be used for the output. Given the limited abilities of other editing softwares in this area however, this isn't too bad. The RealVideo encoder works great BTW.
With Windows98SE, you can only capture 18 minutes in one piece. That applies for all cards except the Canopus EZDV, which supports multiple files and even works with Windows95, but costs so much more that you could easily buy Windows2000 for the difference.

StudioDV can also do scene recognition with analog tape, however this has to be done by image analysis. StudioDV does this automatically after capturing. It takes quite a while though (1.5 x playing time on a Celeron 600).
There is a 3rd party Freeware called
Scenalyzer which can also do the job and is about 8 times faster. Scenalyzer can also cut the capture file physically into scene files, and it can automatically read multiple capture files, merge their endings and beginnings if they overlap and cut them into scenes, all in one step. This can provide a solution for the file size problem under 98SE, but the split scenes are not very useful for further editing in StudioDV.
If you want to improve pictures with VirtualDub however, using this as an intermediate step makes sense. VirtualDub has better tools for image cosmetics than most of the sophisticated video editors, and it can also merge the files afterwards, using its 'add video segment' function.

Normally with StudioDV, it would be better to capture overlapping files, make scene lists for each and then open them in StudioDV in one project. The problem is that StudioDV cannot do this automatically, it only lets you set a time limit for capturing. So it captures 18 minutes, and with analog tape, it even calculates scenes for half an hour afterwards. Then you have to return to it, set the tape back a bit and start the next capture segment.

AVI-IO now also has DV capabilities and can continuously capture to several sequential files. This offers a convenient method to overcome the 4 GB barrier under Windows98SE because no frame is lost and therefore capturing overlapping files is not necessary.
VirtualDub can also be used and it it is free software. How to do this is described below.

However it takes half an hour to open each single file in StudioDV, because StudioDV will do an optical scene scan for each, even if the source was digital tape !

Not very convenient all this. If you use WIndows2000, the problem is not as big because you can capture at least each tape in one piece. However I think I have a solution for 98SE as well.

Scenalyzer is capable just to recognize scenes and store the start positions in a text file. It can also handle multiple files in one pass. Because of its speed, I thought it would be good anyway if StudioDV could use this scene list.

For the above reasons, I have written a little program I called
ScTrans
that can convert Scenalyzer's list format to that of StudioDV.

ScTrans can handle Scenalyzer output from multiple file scans. So you can capture to 4GB segments with AVI_IO, let Scenalyzer run over the files in one step, let ScTrans convert to .seg files in one step. The whole process is 8 times faster and, most important, it's one step for all segments at once. Afterwards, all files can quickly be opened with StudioDV and combined into one project. If the files all reside on one disk, switching between them can be done with StudioDV's dropdown list, which is almost as fast as flipping pages in the thumbnail booklet. There is no size limit for the output as long as this goes directly back to tape.

The algorithm of ScTrans is based on the output I got from Scenalyzer 3.0 and StudioDV 1.05. Because I had to re-engineer the file formats from the output I got, it may be that it doesn't work for other versions of these programs (1.10 tested OK meanwhile).
There is also only basic error handling built in.

The Scenalyzer file contains the name of the video file it belongs to, so everything else is automatic. At runtime, the name of the video and scene files are displayed with a list of the start and end frames of all scenes. Scenalyzer can also analyze multiple files in one action. ScTrans will handle this and generate separate .scn files for each video file.

This little utility is in beta state and provided as is, no warranty whatsoever. I have not been able to crash it so far, but I recommend not to run other applications when you try this, because if it would crash, you could loose data if files are open.

Because of virus and trojan hazards, I recommend to download the program only from this site. The zip file also contains the source so you can compile it yourself.

 

DV Codecs

Software codecs are used for display and rendering of DV files. StudioDV comes with its own codec, which is decode-only and not even necessary for the StudioDV because from version 1.04 on, SDV uses the Microsoft codec. The MS is not based on Video for Windows (the old 16 bit interface) but on DirectShow. Given that you use the latest StudioDV (from version 1.04) and Microsoft DV updates, you could remove Pinnacle's DV300 codec from your Multimedia settings and StudioDV will still work. More about this can be found at the Pinnacle Webboard.
Scenalyzer as well as VirtualDub however would not read DV files anymore with this method, under Windows98 as well as 2000. Apparently They can't use DirectShow codecs. So I leave Pinnacle's codec in place and it works well for me because the programs mentioned need only to read.

Under Windows98SE, an alternative is to install the Adaptec codec over StudioDV. Their license statement doesn't forbid this.
However this doesn't work under Windows2000. What also doesn't work to open files compressed with it with StudioDV and then exporting to the camera. SVD will open the files but you would have to save to an AVI file in DV format (complete recompression), open this again and then you can send to the camera but sometimes even this process gets problems. Reason unknown.
So if you want to do edit jobs with Premiere or VirtualDub, either save as MJPEG (SDV then recodes only these parts to DV) or get the MainConcept codec.

The MainConcept codec also works with Windows2000. This one costs 50$. The trial version will read files without restrictions but watermarks them when it writes to files. One of the fastest and best codecs, recommended except for playback with Media Player. There it is slow and has a green shift.

Under Windows98, I could achieve perfect playback with Media Player even with the Adaptec codec installed. Apparently Media Player in this case still uses the MS codec dll because it plays fluently in full format (PAL, on a Celeron 600).

Under Windows2000 w. the MainConcept codec, I had Media Player stuttering after installation. Sometimes Media Player also uses the MS codec when MainConcept is installed, but If not, it is helpful to switch between codecs. Under Windows2000, codecs can be changed on the fly by using .reg files:
All codecs have different dll files that do not conflict on your disk. The only problem is that there is only one registry key that determines which dll is to be used. Changing this key can switch between codecs. No reboot necessary.

This is for only people with some computer knowledge:
The key where the active codec dll is assigned, is located in the system registry under:
[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Drivers32]

The key for StudioDV it is:
"VIDC.DVSD"="miroDV2avi.DLL"

The key for MainConcept it is:
"VIDC.DVSD"="mcdvd_32.DLL"

The key for the Microsoft codec is:
"VIDC.DVSD"="qdv.DLL"

You can use Regedit.exe, located in your system folder, to look at the keys and to export keys to .reg files.
Reg files can be edited (right click and choose edit), and they can be re-imported to the registry by double clicking on them.
So you can easily make several .reg files containing only this path and key (remove other keys that may have been exported as well), with different key values, and change just this single registry entry by double clicking. One separate .reg file for each codec is necessary.

Complete solution for StudioDV under Win98SE

(This is a 'beta' suggestion, not yet entirely tested for reliability)

Example: capturing a 90 min tape to disc.

We need AVI_IO or VirtualDub, Adaptec or MainConcept codec (only if SDV's doesn't work), Scenalyzer, ScTrans (see above for details):
- capture a 90 min. tape: AVI_IO generates 5 gapless files in 90 min.
- you can also capture with VirtualDub (on some systems this will crash, you'll have to try if it works for you). The trick is to connect the analog audio outputs of your camera to the line in of your sound card. Set up the spilling system, select the WDM capture device, set audio 'compression' to 32k or 48k 16Bit Stereo depending on your source tape sound format. Use compatibility mode to capture.
- run Scenalyzer: ~ 20 min for all
  (hint: select option for high or highest scene detection sensitivity)
- run ScTrans: some seconds
- open the files in StudioDV: some seconds

The 'normal' process would be:
- capture segment one: 18 min
- with analog tape, scene detection another ~ 30 min
- manually rewind tape a bit
- repeat to capture segment 2 .. 5
This occupies you all the time, for up to about 4 hours.
(If you capture with AVI_IO or VirtualDub, capturing will be automatic but StudioDV will detect scenes when you open the files, so opening each file would take 30 min. and has to be done manually, even with digital tape source)

I'm glad about any feedback from Win98 users because I'd have to revert my entire video partition from NTFS to FAT32 to test this process complete tapes (AVI_IO does not work under Windows2000).

 

 
DV Field order problem

Own observations and problem reports by several users have revealed that there must be compatibility issues in the way interlaced video fields are ordered (or decoded, as DV stores fields into 1 frame) against all other video formats available.

Obviously, the captured DV has inverted field order compared to TV captures as well as industrial DVDs, and not only that, it has inverted even and odd TV lines AND inverted sequence of field playback, so the effects compensate for still images but become apparent if something moves.

The effect shows up when DV is converted to another video format interlaced, or when another interlaced source is converted to DV and then sent to tape. The inversion seems to occur in the Camcorder's hardware codec, bot compressing and decompressing, so it compensates for mere editing and recording back to tape.

Several software DVD players can deinterlace, and also my TV-out (Matrox) has a help file that misleads users to switch to one field with interlaced material. No wonder only a few people stumbled over the problem and even less could identify it.

As long as there is no other solution, here is a workaround:

Donald Graft's Smart deinterlacer filter for VirtualDub, latest version, has more field ordering options.
I've set threshold to 255 to disable deinterlacing but that's not really necessary (field swap disables deinterlacing anyway with this filter). Just check phase shift, field swap after phase shift. That does resolve it.

VirtualDub can encode to MPEG4 directly or frameserve to TMPGenc (MPEG2) or MS WindowsMedia8 encoder. All work well with interlaced material. A definite advantage for storing DV footage on a CD, if you have a good TV-output.

Here's a screenshot of the settings: