Properties of DV

[bsuska]

My question is: Is all DV interlaced? Premiere gives an option to deinterlace rendered DV. How can this work? My Windows Media Encoder still recognizes it as interlaced when I have it detect. I know the DV that is recorded on my tapes is lower field dominant and subsequently transferred to my machine as lower field dominant file.

Thanks in advance...

-Brock

[edDV]

My question is: Is all DV interlaced?

DV is normally interlaced (bottom field first) and always structured as fields but progressive material can be transferred as DV.
(see adam wilt's discussion here http://www.adamwilt.com/24p/ ).

Premiere gives an option to deinterlace rendered DV. How can this work?

Premiere project settings can be interlaced or progressive. If progressive, DV can be deinterlaced in various ways from IVTC (good), drop field (good but half resolution) to various field blends (mostly bad).

Or, you can edit in interlaced DV format and convert to progressive if you need to in the output encoder. If you don't need to, DV to DVD works best with interlace encoding.

My Windows Media Encoder still recognizes it as interlaced when I have it detect. I know the DV that is recorded on my tapes is lower field dominant and subsequently transferred to my machine as lower field dominant file.

That is normal.

[bsuska]

Thanks for the quick answers. I guess what I'm trying to get at - we are looking at storing video files that will eventually be encoded to wmv. They are 3D renderings. Since we can output to anything from the 3D program, I'm looking for the most effective way to store these files without compromising the quality of these files. Uncompressed would be the obvious answer but the size of the files is just too much. We then started looking at DV (5 to 1). What I'm worried about is this - if I render to DV is the file automatically going to have a lower field dominance? This is unnecessary since I will then need to deinterlace it for wmv. Is there a better codec that I should look into using? Snooping around Premiere, it would let me export to a DV file with progressive box checked. Does this mean that I can output to a DV file that has no interlacing? I'm not concerned about project settings here...just export settings. From things I've read, I'm gathering that the actual DV spec says nothing about interlacing (DV tapes will accept whatever) it just happens that I'm using a NTSC DV camera which does put lower field dominance on the tape.

[Cornucopia]

Really, what you want to know is a good codec that you can use as:
1. Export option from your 3D app.
2. Import option to your editing app (Premiere it sounds like)
3. Export option from same editing app
4. Import option to WMV encoder.

You're concerned about size & quality.

So, try a losslessly compressed format: HuffYUV or Lagarith. You don't necessarily have to worry about interlacing, and you'll be able to retain better chroma resolution (4:4:4 or 4:2:2 as opposed to DV's 4:1:1).

DV explicitly supports interlacing, and implicitly expects it, but you've already seen that there are workarounds. Most of them, however, are forms of Telecine Pulldown.

A lot depends on if you expect to go direct to video (tape, DVD, broadcast, etc) or not. If you don't DV wouldn't be such a natural choice.

Scott

[edDV]

I agree. 3D should be preserved as progressive especially if progressive is the expected destination. You can always interlace later from the master.

DV format would be a force fit and is not the obvious choice. HuffYUV or Lagarith should be tested and researched. They were primarily developed for film or video sources. I'd want more info on the lossless claims when applied to sharp 3D renderings. 3D forums could advise the best mastering codec that also works for Premiere.

[dipstick]

DV can be encoded as Progressive or Interlaced, it supports both. If you plan on exporting to Tape, it needs to be encoded as Interlaced Lower Field First. For all else, it doesn't matter.

In your case, you'd be better of using Huffyuv or Lagarith codecs. Huffyuv is Lossless, Lagarith is very slightly lossy but much smaller file size. I use Lagarith for my 3D rendering and am quite happy with the quality and file size. It's much better than DV.

[bsuska]

Cornucopia, that is exactly what I'm trying to do. Get something out of 3d Max (that will need to be saved to a server), then imported to Premiere (where a title will be put on) and then exported to a format that the Windows Media Encoder will accept.

I'm a little bit confused by some of the answers posted. The mini DV tapes that I buy can house both a progressive DV signal or a interlaced DV signal? The link that edDV sent me would suggest that even though the camera is capturing in various formats, the signal is actually recorded to tape in an interlace fashion. So I guess I'm back to the question is all dv interlaced. dipstick says no.

[Cornucopia]

It might help if you divorce the idea of DV-as-a-codec from DV Tape-as-a-medium. They started out completely united, but they don't have to be. Now, the codec can be used lots without ever seeing the light of day as bits on tape, and DV tape is now used to store MPEG2 HiDef streams (HDV) and even plain data.

Nonetheless, DV the codec orders their bits by fields. Those field pairs could be time-syncronous (i.e. Progressive) or time-sequential (i.e. Interlaced). The decoder and application will do the work of associating the field pairs appropriately.
DV tape is, by nature, sequential and so lends itself to interlacing, but there are cameras that have a 30psf mode (I'm not even gonna go into the pulldown stuff).

The upshot is this: If it's DV. Assume that it's interlaced (or wants to be).

Scott

[guns1inger]

The best output format from your 3D program (e.g. Max) is a still sequence using TGA or Tiff format. This allows you to

1. Maintain a progressive frame
2. Include alpha channel information
3. Easily render multiple passes (e.g. seperate shadow, reflection etc. passes)
4. Restart interrupted renders without having to begin again at the beginning
5. Make changes to individual frames or sequences within a master sequence without the need to re-render the entire sequence.

Image sequences can be imported into all major editing apps and treated as if they were video.

[JohnnyMalaria]

The DV format doesn't care about interlaced vs. progressive. The data are encoded in an identical fashion.

It becomes important at the editing stage, especially if you add effects, change the size etc etc. Here's why:

A true progressive recording contains 25 or 29.97 whole frames every second (PAL or NTSC, respectively). Let's say you want to rotate the image 360 degrees gradually over a period of 360 frames. Each frame is simply rotated by one degree relative to the previous one as a single image and the new one created from it.

For the interlaced case, the recording contains 50 or 59.94 fields every second. To achieve the same rotation effect, each field has to be rotated. For a frame, the first field is rotated by 0.5 deg and the second frame by 1.0 deg. So the editing program has to do more work - it has to separate the fields from the frames, do its stuff and recreate the new interlaced frame. (If the software is smart enough, it won't actually separate them, it will just call the rotation effect twice, pointing to different parts of the same frame).

Complex transitions and effects (inc. moving titles), take more time on interlaced footage than progressive since the calculations are, in effect, performed twice. If you aren't performing complex editing that requires proper field processing, you can cheat and tell the software that you are using progressive even though it is interlaced. Simply color correction, fading in and out etc are good examples where this can be done. Of course, it depends on the software. Certainly, Premiere calls effects routines twice for interlaced material as did (R.I.P) EditDV.

Anyway, I agree with the other posters that, for your application, DV may not be the best choice....

John.