DV quality for broadcast

[Dan Uneken]

I use the XL-1 for documentaries and as second camera on bigger TV shoots, with Betacam SP as main camera. But often producers refuse DV material even without looking at it, saying that this is not "Broadcast Quality" and their policy is not using anything but "Broadcast Quality". I'm sure there are lots of you out there who have the same experience. What do I tell these people, what exactly is "Broadcast Quality", by the way. At the time of broadcast the DV looks great to me, better in many cases. Thanks.

[Martin Munthe]

In it's first generation without recompression DV is up to broadcast specs. Tell the producers I've onlined at least ten music videos (shot on 16mm) in DVCAM (4:2:0, 5:1 compression) for broadcast on MTV Europe (prime time).

But please note. If you recompress DV it get's really bad. Like if you add a title and render it in DV. If you are going to do any recompression be sure to do it in a less lossy environment. Uncompressed is ideal.

[Josh Bass]

I'm sorry, explain the bit about the titles again? I've never noticed quality loss after adding effects and such.

[Dan Uneken]

Does this mean that if I edit a program in NLE and output it to DV again, it's lost a lot of quality?

[Henrik Bengtsson]

Basically, each time you need to recompress DV material, ie. calculate it again, you get the compression artifacts doubled again. So if you are to use DV materials, make sure to always be maximum of 1 generation from the original.

For example: You capture a shot on dv. Lets say you want to do colour correction and a compositing onto this. Now, if you do it all in the same program (for example FCP or AFX) you are home free. Since you will be 1 generation from the source. But lets now say that you need to run it through a second program. (Retimer or something.) If you now use the DV material, do colour correction & compositing, render it out to a DV stream and then put it into retimer. You are working with 2nd generation material. With added artifacts and all.

So how do you bypass this. Well, by using uncompressed material for all inbetween generations. This takes a lot more cpu, mem and hard disk to handle. but its the only way to keep it from degrading the shots. Also, if you add effects and want to go broadcast, try to put the end result onto uncompressed also. Why? well, if you add effects to a DV sequence, you WILL need 2 generations degradation of it. 1 when you capture the material and the 2nd when you recalculate it after applyhing the effects.

It is very dependant on how the material looks if this to obvious or not. I recently made a documentary about whalesharks, all blue tints, all gradients due to the ocean. After 2 generations (and especially when blown up on a film screen) this looks like war of the square blocks :) Now, the end result will be on uncompressed video, and this dv preview was just that, a preview, so im not to concerned with it.

Do not expect capturing on dv, applying 2315 effects, run it through 8 programs and then send it off to dv again and not loose quality. Because you will.

/Henrik Bengtsson
DocuWild

[Rob Lohman]

It is true that your loosing quality the more times you recompress
(ofcourse). There are however 2 other ways to avoid loosing
quality and to not have to do UNCOMPRESSED files (which take
up huge huge space).

- Use a lossless compression codec. For stills this is easy. I know
there are some video lossless codecs available as well, but I
don't think they are included with windows for example. Probably
need to purchase this.

- Use frameserving. It is possible to use applications like Avisynth
or Virtualdub (with vfapi (?)) to directly go from one application
to another without writing any files to your harddisk. The
information gets transferred directly to the other application
in uncompressed format whenever that application requests
a frame. No extra (almost) harddisk space needed.

[Josh Bass]

You all say each time you render, you're losing a little quality, correct? In Vegas Video, the only time you have to render (except when compositing) is when outputting to a tape. Is this good?

[Henrik Bengtsson]

Well, yes and no. That means that you have two generation degradation (1.st in the camera, 2nd. when you output to tape again). This is the minimum quality degradation you will have when outputing to a new DV tape. And in some cases, this will destroy the material enough to be useless in a broadcast environment. The only way around this is to make sure that the final output is uncompressed or using a lossless compression codec and using say Digibeta as final output tape.

/Henrik Bengtsson
DocuWild

[Bill Ravens]

What you all are saying is true ONLY for the re-rendered portions of the video. The vast majority of the video is simply a 1 for 1 copy unless you do color corrections, fades, etc. on every frame. I guess it might pay to get it right when you shoot it, eh?

[Henrik Bengtsson]

True true =) If you get it all perfect in the camera then you wont have to bother with re-rendering. But.. seriously. How often do you use footage that you don't even tweak a little black & white levels on? I know i always play with the vectors&waveforms on every image i make.

But of course, if you get the image 100% perfect in camera, you wont have to re-render (except if using dissolves :)

[Bill Ravens]

Your point is made, docuwild. In light of this discussion, when CANON foists something on me like black bars across the bottom of my footage that forces me to re-render every frame...I utter promises to myself about my future commitment to Canon products.