FCP Native HDV editing: the good, the bad and the ugly
 

Now that the champagne celebrations heralding the arrival native 24p editing on FCP are winding down, it’s time to wake up and deal with the hangover: Native editing is not a good way to complete a project that is intended for HD delivery or filmout.

Why? If your timeline requires rendering of any kind (even real-time effects, that don’t require rendering to play) you will be recompressing your mpeg, which can add nasties to your image. I wasn’t convinced of this until I did some tests. Just the simple act of placing a superimposed title on a shot, doing a simple color correction, or performing a flop, caused blockyness to appear in the shadows - subtle, but definitely there, and not good for the big screen. If you complete in HDV, say by outputting to tape and then handing it over to a post house to up-convert, these recompression artifacts will become “baked in” to your master. (Think of all those Red Rock M2 shooters that will be rotating every single shot in their timeline!) This inconvenient truth should be a concern to anyone using HDV for high-end work.

Don’t get me wrong, I am just as excited about the arrival of native 24p editing on the Mac as the next guy. I think it is the most practical way to deal with large quantities of HDV footage (after all, transcoding is a supreme pain in the butt!). However, for projects aimed at large screen viewing, I think native editing should be treated as a means to get to a picture lock, but not to a master. The same artifacts that I mentioned above all disappeared when the shots in question were copied and pasted into uncompressed timeline. This, to me, is really good news because it means that a whole sequence that has been edited native can be easily converted to uncompressed while still in FCP, shedding it’s HDV baggage before being played out to a high end format. This technique is not new or mine – it has been suggested by others on this forum for quite some time, but it was reassuring to see it work with my own eyes. Given the surge in native editing that will surely follow Final Cut’s new capability, I thought it was a good time to bring it up again.

Of course, rendering your entire sequence to an uncompressed codec is not necessarily a practical thing to do, but you really don’t have to. Complete your edit in HDV, then submit it to your post facility as a consolidated FCP project on a firewire drive. Let them render it as uncompressed on their high-speed array and then output it via HD-SDI to your delivery format. Yes, it is a bit of a pain, but easier, I think, than converting all of your material to another format at the start. Happy cutting!

Well I dunno what things are like in FCPland, but when Avid gets around to releasing 720p25 support my workflow will be something like this:

-Capture HDV
-Change project to SD & transcode media to DV or 15:1 (probably DV, space is cheap these days)
-Edit offline with SD footage for speed's sake
-Change back to HD project, relink sequence to HDV media
-Use DNxHD to render anything while mastering
-Mixdown to a DNxHD master

I'd be curious to hear what workflows people use for HDV (especially with HDV under Avid but under any NLE would be interesting)

Exporting a "Quicktime Movie" with uncompressed sequence settings is the same net effect as dropping into an uncompressed timeline and rendering. This is a great way to deliver also.

Just to be clear for the audience tuning in: native HDV can be your path to the highest quality available; you just have to know how to do it correctly!

I can't say working with HDV media in FCP is as buttery as DV media, but it definitely doesn't slow me down.

These days I only make offline media for multicam concerts or to fit a project on a powerbook.

How is it for realtime effects like dissolves, keying, colour correction etc?

RT dissolve is full frame rate, but blocks up a tiny, tiny bit. Keying, I don't know. 3-way color corrector, I get one layer of that full res, full quality, full speed.

Everything works well enough for me to get a cut to 97%. At that point, whether a render takes 2 seconds or 8 seconds doesn't really matter anymore.

Keys and graphics usually need to be tweaked as you move from low res to high res media anyway, so I consider it to be a pretty equal trade-off.

This all being on a dual 2.0 G5, which is trumped several times over by the newest Mac Pros.

Hi Nate,
If you did this to an HDV sequence that had effects already rendered in it are you sure the uncompressed would be made from underlying HDV and not the HDV render? If so, that is an interesting option as it would not require rendering on the facility side.

Speaking as a post facility, I would probably still recommend the render approach. You could do the render at your end to save time and yet we would have access to the underlying project and all applied affects. That way, if something comes up looking funky we can look into it, even make a fix, and then relay a section off without loosing too much time.

I'm pretty sure (Note: *Pretty sure*. Hopefully some one can back me up) that any form of non-native output from a FCP timeline is processed from the original media files, not any mid-range renders.

I know it works that way when you add effects to already-rendered sequences. It'll go back to the original media file, and render with all the effects, rather than adding the effect to the already rendered file. I can't see any reason why it would behave differently on an output.

Yes, it always goes back to the original media and tosses any renders.

This is true even if you've rendered an uncompressed timeline!

Not that you're guilty of saying otherwise, but you know, those FCP guys have indeed thought a thing or two through when it comes to HDV.

My HDV workflow will go something like this:

Capture and edit native HDV.
Bump to uncompressed 8-bit SD timeline and re-render so all graphics etc suddenly look lovely.
Author SD DVD.
Back up to external HD original HDV timeline for future use of HDV/Bluray DVDs.

I have a question:

Assuming Native HDV Editing In FCP: If I color correct one clip on a timeline and DON'T render it, but then removal the color correct filter, will this affect quality? What about if I do render it, but then removal the color correction?

You can delete this post, I answered my question.

No it won't effect quality, even if you do render and then remove the effect, FCP will always revert back to the original source clip, it won't revert back to a compressed/uncompressed/compressesed again as you add/remove effects.

You can add/remove effects all day long, then when you finally remove all effects, the clip reverts to original clip hence quality is as it was when you first imported it.

Hi Nigel,
I don't think you will gain anything from the uncompressed stage if an mpeg for DVD is you final master. As long as you create the mpeg from your HDV timeline via "export to compressor" your export should be created from your original data (graphics included). Of course, it's always good to do a little test to be sure.

I will be doing tests. I have been told by a codec guru that if there are any cross-disolve type renders or other complex graphics that have been rendered out in HDV there could be blockies and artifacts, whereas a bump to uncompressed 8-bit timeline re-renders them in that format i.e. not HDV hence no blockies etc.

@ Tim D. How about renaming the thread header to "FCP Native HDV editing: the good, the bad and the ugly".

As Andy said, uncompressing the timeline for delivery is nothing new but may be new to FCP users since the recent (welcome) update.

S.Noe

So the least blocky solution from HDV to DVD 5 is.....
 

I have been searching this forum and others seeking a answer .
Here is my senario,
Shot 720 30p on HD 100 , droped it in a HDV timeline [FCP 5.2], approx 7 minutes, lots of filters and time remapping , colour correction.
Export to compressor , using best Quality DVD 90minutes 16 x 9 , import to DVD studio pro.Burn.
What I notice while viewing on a NTSC monitor and a NTSC projector, the image is lighter, also blacks where I changed the contrast , becomes blocky as talked about in this thread.
Does anybody have any suggestions as to how to reslole these issues, it would be so nice to have a finished product that is a mirror image of the piece I see being played in my timeline.
thanks

Shaun,
I don't think that you are gaining anything by transcoding to DV. Editing in HDV is equally fast - I didn't notice any delays other than the need to render some effects, such as superimposed titles. The method I used was to edit as far as you can in native HDV and when the picture is locked and you are ready for colour correction, titles and other final stuff, transcode to the highest DNxHD codec and finish in that.
My problem was getting to SD DVD directly from the HD timeline. I could do it but the picture didn't look good with washed out colours and raised blacks. I had to take the HD timeline, "Export to HDV Device" and then use the HDV tape, downconvert it via the deck to DV, recapture the master in DV and geti it to DVD the usual way (QR REF).

Thanks Shuan,
Sorry but when yoy say "downconvert via the deck" is this possible to do using the HD 100 and if so , how would I go about it .
thanks

Your guy has the right intention, but what you're doing is unnecessary.

As long as you render to another format (uncompressed, MPEG-2 for DVD, etc), FCP always goes back to the original media and makes that timeline step redundant.

I don't think the camera can do that... maybe Nate or some else can confirm this?

If you don't mind I will jump in with my FCP experience since I have been editing HDV for over a year now.

I am shooting with a Sony Z1 so my footage is 1080i60, but the theory should remain the same.

Most of my end use is SD DVD, but I edit everything native HDV and export a Quicktime movie with the original settings of my project. I then load this quicktime file into compressor and make a DVCProHD 720p60 version. I do this because I use a lot of H.264 for web and iPod delivery. This file seems to be a good master to then do compressions from either for DVD, web, whatever. Maybe I am adding an extra layer of compression, but my DVD's look pretty damn good.

I can then take my DVCProHD files and output to DVCProHD tape if need for an HD master. I could also go uncompressed HD if I wanted to make an HDCAM master.

Does this make sense?

Dan Weber

Down-converting and Avid:

You can't down-convert via firewire using the camera or the BRHD50 (like you can with the Sony HDV devices). The camera can downconvert via component SD (or composite) analogue but you'll lose control and TC. If you want to capture in SD for offline (for example) you can ingest via component, unbalanced analogue audio and RS422 for control and TC. This is the method I use for Avid offline.

Btw, I agree with Jiri that the HDV transcode to DV is somewhat superfluous on a fast system. Furthermore - for long-form projects - it's usually more efficient to capture compressed SD for offline and then decompose a duped sequence and conform as HDV. You can then set renders and imports (Media Creation settings) for your online HDV timeline to DNx to ensure all processing happens at optimum quality. If you happen to have Symphony Nitris, you can use uncompressed HD instead. This method keeps the quality high, minimises storage requirements and eliminates lengthy transcodes.

Hey fellas,

Very interesting to read this whole back and forth. I haven't yet tried any edits with the new FCP.. so I have nothing to add but one more question:

I've just gotten the DR-HD100 - and am awaiting the QT wrapper for 720p24 HDV recording... thinking I will finally have the streamlined HD 24p workflow I have been dreaming about since I got my HD100 a little over a year ago.

After reading all your varying assertions about the loss of quality in Native editing - I've got some basic questions which I was hoping you all might chime in on. They are:

When FCP (or perhaps in my future situation the DR-HD100) converts the MPEG2 stream into a quicktime clip - which we then use in the edit - is the resultant QT file still an MPEG2 stream? Or is it creating a reference file that refers to part of a companion m2t? Or something like that?
What happens to the GOP structure? And in this transition from mpeg2 stream to 720p24 HDV QT File is there any loss of quality?

When you convert to uncompressed HD what happens to the GOP structure?

Also, when people talk about "uppressing" to a codec with 4:2:2 color space for color correction - do they simply mean compressing HDV to dvcproHD (for example) or can this be done for via output of uncompressed HD? And what are the pros/cons of this process regarding image quality/resolution and original GOP structure?

I'm sorry - thats actually a ton of questions... any input would be greatly appreciated.. as I am in the early stages of planning a feature intended for filmout and I want to be rocksolid on the workflow from shoot to edit to output.. before I begin scheduling, budgeting, breaking down the script, equipment needs etc..

thanks for any input.

oh yeah.. sorry

so (after re-reading my last somewhat confusing post) my basic question was - would there be any downside to recording 720p24 with the QT wrapper (on my DR-HD100) - assuming that these files will be my master footage, rather than a pure .m2t - if thats not what these files represent?

thanks.

Daniel, what you say makes sense, but I would not do it. Working in DVCPROHD is only any good if you are importing DVCPROHD footage to start with. So why convert from one heavily compressed codec to a different format, but still a compressed codec? You'd be better off leaving the DVCPROHD part out of it and either bumping to an uncompressed 8-bit (not 10-bit as that is a bit twitchy and is far from perfect right now) timeline and going from there, or go from HDV straight to Compressor

When using the "native" method in FCP to bring in HDV, the muxed stream is demuxed to it's constituent m2v and mp3(?) audio streams. These streams become tracks in a Quicktime file. The process also creates "hints" I believe for the Quicktime MPEG decoder that allow Quicktime to handle the file quicker.

The Quicktime file is still MPEG2, and it hasn't been trancoded to anything else...that is, the picture data is still untouched. It's just a different vehicle for the same data, same GOP structure.


Well...uncompressed video doesn't have a GOP structure. Each frame lives alone. So each MPEG frame is decompressed to a frame buffer, then written out to the uncompressed structure where it lives alone.

Boy, you're sure worried about those GOPs.

The above methods of outputting your final to uncompressed do away with the need to go to an intermediate step. Pros and cons? Going to uncompressed needs lots of disk space, and a fast drive if you expect to be able to play the resulting file in real time.

I have to say, this is an awful lot of talk about a simple aspect of how FCP works. What seems like voodoo or hearsay in this thread is kind of common knowledge amongst the FCP post houses here in Los Angeles...with all the questions and discussion, it seems much more complicated than it is!

we never go back to HDV and seldom ever back to Tape for that matter in HD editing.

But if I did we would:

shoot in HDV with the HD100u-A

capture and edit in HDV.

export and uncompressed 10-bit quicktime for output later to HDCAM.


pure and simple and no need for complications.
I've looked at this kind of thing closely and have never seen anything ugly.

It really is not all that complicated as Nate has said.

Nate and Mark others,

thanks for your comments/clarifications... much appreciated.

the nice thing about his forum (and this camera too, it seems) is that it works for both experts and intermediates - with regard the technical aspect. Thank you for clearing up my questions/confusions re: HDV format and the editing process

I myself am not incredibly tech-oriented.. and actually the process of figuring out a feasible hd24p workflow during the last year has forced me to learn a bit more of the technical side of file management - which has in turn made me a more efficient shooter and editor.

Big props to the forum members who take the time to explain these issues, to the less technically proficient. I hope in some way it has been beneficial to you work through your explanations of our various questions. It has certainly been a big help to me... when in many cases, JVC and Apple didn't really have much to offer in the way of support/answers.

Perhaps the problem stems from the fact that Avid does reuse renders and the fact that from the days of Hi8 and DV there was an obsession with bumping to something better.

I'll kill if I read one more time about the "fragile" nature of HDV. It's data in a file and no more fragile than your income tax data files. It gets decoded once in FCP. The uncompressed result of ANY FX is recompressed once during export to HDV, DVD, DV, HDCAM, DVCPRO HD, etc. There is one decompress and one compress.

If you waste the money to go to uncompressed there will be -- one decompress and one compress.

And, if I remember right -- FCP decodes HDV to 4:4:4 not 4:2:2. And, you can ask for all computations to be done in 16-bit (or maybe that's Liquid) or FPP. Then the bit-depth is ONLY determined by the export codec. So choose a 10-bit codec.

You're working with alot of NLE's I see. Liquid works 422 8bit and renders subpixel 16 bit depth in the YUV space. There is no re-render unless the timeline is fused to a codec or fused to export. So you can load effect upon effect on the HDV and you are still working with the original source code to get only to second generation. This is a superior way to handle the data. I had the same conversation with David Newman (cineform) on the Cow last year and the subject was compression and recompression and recompression and how Cineforms codec handles recompresses. It's true their codec handles recompresses very well but with the Native way of cutting you're not recompressing over and over, instead your FX are calculated off the source material no matter how much FX and CC is done to the material. Add to that the uncompression and subpixel processes and you'll soon see that native editing using an uncompressed timeline is an excellent workflow.

Nate has written (a few months ago) about working in FCP the same way we work in Liquid. Cutting the timeline on MP@HL for the smooth workflow and then changing the codec to uncompressed for final output. This is the smoothest workflow and gives excellent, predictable results as Andy eludes to in the original post. It seems FCP users can now benefit from the same workflow as Liquid 7.1 with 24p. I'm sure there will be some hiccups like we had back on Liquid 6.1 with 24fps HDV. I'd be surprised to see Apple & FCP get the workflow right, right out of the gate.

Right now is the time to get geared up for 720p60 and 1080p60. We (AL 7.1 users) are already working in 60p native workflows for Sony and JVC. I can tell you that the 60p workflow for 720p and 1080p is already set in Liquid including the DVD workflow. I hope it's working in the FCP release for Mac users as well. Fun Times!

I have had real bad luck going right to compressor from HDV. I think that the issue is that I am working with interlaced footage. I find that if I convert to a progressive format, it works better.