DV edit->HDV/HDCAM-> 4:2:2 online workflow
 

Tried to search for this but came up with naught.

So we have shot on a combo of HDV and HDCAM and are downconverting through the Z1 camera (HDCAM through the deck) so all the footage is DV anamorphic. We will edit the film in dv until the picture is locked.

Then we will do an online and need to recapture both formats into the timeline.

While doing this we want to maintain the highest quality picture with the least amount of generation loss.

What is the best way to do this with the two formats on the same timeline? Do we import the HDV in one timeline (replacing the DV) and then uncompress in FCP. then in another timeline do that for the HDCAM?

We want to do the online in uncompressed 10bit and then we will do color correction on that (not sure if we will do cc in AE, FCP/color finesse or divinci yet).

All original footage is 1080 50i but we are not sure what we want to end up for digital masters. I am assuming we will end up with master on HDCAM. We want the most flexible master to convert down to all the formats. I am thinking 1080p and everything will convert down from there. we will not be going to film but will be going to HDTV 1080i and 720p/i, SD TV (pal and ntsc) and we also plan to show digitally in festivals in HD and SD. Oh, and we will be compressing to DVD (ntsc, pal and one of the new hd dvd formats) as well.

Thanks in advance for your advice!

I wouldn't bother with HDV if you're going to an uncompressed timeline. Recapture the HDV footage straigth to uncompressed by using the analogue component outputs of the Z1 and a component-to-HD-SDI converter. This will fit right in with the onlining of the HDCAM footage.

Be aware that speed changes will most likely upset the Media Manager. Apply them on the final online footage only, or remove them before recapturing.

HDCAM masters will most likely be sufficient. I don't understand why you would want to go progressive since all source footage is interlaced and you're not doing a film out. You may be able to keep the final movie as a self contained 1080i uncompressed Quicktime file if it's a short movie.

I guess I forgot to mention. the piece is 1hr long documentary.

The reason I was thinking of keeping in progressive is twofold:
- the format changes in post. for example to get from 50i to NTSC have to go 25p, 24P then 3:2 pulldown up to 29.97 then interlace again. As well as any HD conversions that need to be done (if we decide to do them in the computer instead of tape, to make dvds for example.)
- I thought that if I had a 1080p master then I could just goto any transfer house, give them my master tape and convert down to any format tape to tape in realtime.

Questions about re-capturing analogue component to HD-SDI:
- will i lose quality vs. going digital via firewire to hdv and then uncompressing in FCP?
- I guess for the HDCAM the deck should be able to do HDCAM HD-SDI straight into a blackmagic or kona card right?

Thanks.

Did you shoot 1080i50 as cheap way to get to 24p, or do you have the international market (PAL) in sight too? If you do, I'd keep a 50i or 25p master around.

Regarding capturing from the analogue component of the Z1: a lot of people have been doing it this way before FCP supported the Z1. It will fit right into your workflow. I do not believe that you're going to lose a lot of quality compared to a HDV->1080 uncompressed decompression- and scaling job on the computer, but it will certainly save you a lot of time and potential timecode issues.

But you will need to test it, and seeing your other questions, you have other stuff you need to test. Did you account for adequate storage? What deinterlacer are you going to loose? How and when will audio post production be done (in 50i or 24p)? How long are you going to hire the HDCAM deck? Will the final cut be done on your machinery or in a post house?

Something that's useful, but not widely known, is that all footage in FCP is handled at 4:4:4 while you edit. HDV, Uncompressed, DV everything.

If you keep it all in FCP you will get the full 4:4:4 work flow, regardless of what you originate in. It keeps everything in 4:4:4 until you do your final render for output.

So bring in your HDV as HDV and intercut it, it will only lose one "generation" regardless of what you do with it: transcode, key, title, whatever!

Think of FCP as a big uncompressed playground for all your footage.

For more info see this post:

http://www.dvinfo.net/conf/showthread.php?t=53247


John

Thanks for your comments again.
We never shoot in 60i, it seems to be a pain in the butt. We would have loved to shoot in 24P but the Z1 doesn't do it as you probably know. Then for the HDCAM we thought we would also keep that the same, 1080i to keep things consistant in the timeline. Then we can have a base starting point for all the footage and treat it all the same.

We plan on doing everything in house, well except for the sound post where we will do the basic editing and then go into the studio to do recording, folly and finally the mix. I am leaniing towards doing the sound at 25P and then slowing it down for the 1080 24P master.

For storage we plan on setting up a 2.8 TB local raid so that should be more than plenty enough.

We have done extensive tests with about 10 de-interlacers (only on DV though) and came up with magic bullet, nattress and built in FCP deinterlacer (yes, the deinterlacer in 5.0 is much improved!). We will have to re-test with the HD footage and select one.

For the deck, we will rent the deck and import all the reels as DV 50i (anamorphic) to start. Then the director can do our offline easily on any mac. Then when it comes time to online, we will just get the deck back for another day or two to import the timeline at full resolution.

I think we are going to go with your advice of an SDI-HD import. Sounds like it will be best for timecode and to get all the footage uncompressed in one step.

John, thanks for the article. I will have to do some more research into this though. I can't bring myself to believe that everything is handled in 4:4:4 in FCP. He says in the post..

"If you take an HDV stream, whether you’re doing color-correction or a 16-layer composite, we decompress all that video into a 4:4:4 color space, do our composites, and then do one single re-encode back down to the HDV format. So you’re only, ever, incurring one generation of re-encoding."

To me this sounds like they are decompressing to an uncompressed codec before they do any of their work. that makes sense.

But if you take for example DV pal (or HDV 50i for that matter). The color space is 4:2:0. That is all that is captured and that codec does not handle 4:4:4 as far as I know. Also, if it was uncompressing it automatically, you would need a shedload of storage (or processing) for that.

I would love to get more info on this but in the meantime, i will stick with a low res offline and then a full res online. If you happen to have more links or reading on this I think it would be appreciated by all!

Sounds like you have everything well thought out. Just a couple of pointers to help you:

Here's an article on (re)capturing HDV with the Z1: http://www.creativecow.net/articles/...capturing_HDV/
The important bit is at the bottom: remember to set the device control to Firewire (-PAL in your case) when using the analogue to HD-SDI conversion.

This product might be handy instead of a analogue to HD-SDI converter: the HDV Connect converter from http://www.convergent-design.com/ I don't think it's going to make that much of a difference though. Miranda has a similar box, but rumours have it that has problems with timecode.

If all else fails you still can convert HDV to 1080i uncompressed with the Media Manager.

John is right and wrong concerning effects handling in Final Cut Pro. He's right concerning the uncompressed part, but I'm not sure about the 4:4:4 part. DV might be converted to uncompressed 4:2:2 when rendering. Final Cut uncompresses the footage, renders all effects and finally recompresses the end result back to the codec of the sequence. This means that you can drop HDV footage in an 1080i uncompressed sequence, but it will need decompression and rendering every time you change something. Better uncompress the HDV footage before bringing it into the timeline.

Good luck with the documentary.

Thanks for all your advice on this. I think it is cleared up in my mind. We will try SDI-HD first and if that fails then we will just import the hdv and uncompress it to 10 bit 4:2:2 and go online from there.

Cheers!!

Sorry if I seemed too glib in my previous post:

Some clarifications:

Obviously if you shoot in hdv/dv (4:2:0), DVCPRO50 (4:2:2) or HDCAM (3:1:1), That's the color space you get, from your master, it doesn't get any better.

FCP doesn't store this data at 4:4:4, when you do need to render something, it takes that little bit, upconverts it to 4:4:4, does it's calculations then re-renders it back to the original codec color space. This is just keeps rounding errors from adding up and compressing the compression time and again.

But what that means is your compression errors are least when you get your footage into FCP as early in the chain as you can.

For example, taking your HDV then bumping up to HDCAM then getting into FCP gives you this:

HDV(8 bit 4:2:0)-->HDCAM(10 bit 3:1:1)--->HDSDI(10 bit 4:2:2)--->FCP (any renders or changes are made in 4:4:4)--->Final Sequence: Output to SDI(10 bit 4:2:2) ----> HDCAM (10 bit 3:1:1).

You go through several re-compressions before you get your footage out, in varying color spaces. HDCAM->HDSDI is of course, a very minor recompression, but it is one (and that's why cloning tapes in high end post houses is done in SDTI instead, which gives an actual bit for bit copy from tape to tape).

If you bring your HDV into the timeline directly and it looks like this:

HDV(8bit 4:2:0)--->FCP (any renders or changes made in 4:4:4)--->HDSDI (10 bit 4:2:2)--->HDCAM (10 bit 3:3:1).

You cut out one compression/re-compression cycle.

So you open up an uncompressed HD timeline, and drop your HDV footage in it you keep the best quality possible.

The trade off is that any time you modify the HDV footage in the timeline, you will have to re-render it. Of course, you will most likely have to render anything in HD uncompressed more than a simple fade or dissolve anyway.

I'd try a quick test, one with direct HDV input, and the other with the HDV-->HDCAM-->SDI input and see which one you like better. It may of course be academic, most likely, both will look fine to the naked eye, and even a scope might have a hard time picking out the differences.

But being digital video, all you have are a long string of 1's and 0's arranged in a cunning order. You want to get those bits into the computer with the least amount of error before you manipulate them. And HDV via fire-wire is a direct bit to bit copy, an actual clone of your footage.

So I still believe that HDV direct to FCP will give you the best image...

Cheers, and good luck with your project!

John