HDV Workflow w/FCP4, DVHSCap, MpgStreamclip?
 

Hello everyone, I was wondering what sorta workflow would work so that I can edit using FCP 4.5 HD (not V5- so no native HDV) and send back to Sony FX1 and preserve HDV quality.
I am using DVHSCap and Mpeg Streamclip. I can capture the video (m2t) the Sony FX1 produces- and can demux it to .m2v and aiff files. Since FCP4.5HD can't edit native m2t files, what conversion do I use to make the clips editable in a FCP timeline (W/O rendering)~~ I'd then like to be able to create a TS (transport stream) to send the finished project back to the FX1- any suggestions?

{Here's a sample 7mb Quicktime video (H.264 Codec) of what I'd like to send back: this is a highly compressed FX1 clip- but nonetheless is better than anything of similar quality with DV:

Right-click and "Save As"

http://www.stevenunez.com/videos/MantisH264.mov }

Thanks you guys- you're the best.

Steve, I'm not sure what you mean by "without rendering".

Do you mean playing the HDV footage in the FCP timeline without transcoding to another codec, ie just demuxing/re-muxing? No, not in 4.5

Or...

Do you mean realtime effects in the timeline, but at HD resolutions? Depending on the power of your Mac, that would be possible if you first transcoded to DVCproHD (Streamclip can do this is you have FCP 4.5 installed). DVCproHD is about 4x the datarate of HDV.

As for creating a transport stream, currently Quicktime doesn't support transport streams, so you could try FFmpegX. It can output MPEG2 transport streams of the relevant size and datarate with suitable mp2 audio, but how precisely compliant they are I don't know.

If you're willing to use an offline/online workflow you could use Streamclip to create m2v files and DV movs (or offlineRTs if disc space is tight), edit those and then delete and reconnect to the m2vs, and output to DVCproHD or even uncompressed 10bit should you have enough space. Then use FFmpegX to convert to a HDV transport stream as before. m2v files can be loaded in FCP 4.5 but you can't get any usable playback from them, but the finished timeline could be exported (this is how the LumiereHD workflow operates).

Dylan,

What I meant was, w/o needing timeline rendering when clips are dropped in!

I'd like to be able to edit then export the project as a video (what codec) then make it compatible for sending it back to the Sony FX1.

(man do I wish I had FCP5!!!!!!!!!!!!!)

Then you need to transcode to DVCproHD 1080i which with your Dual 2.5 G5 should give you plenty of realtime effects. Streamclip can do this but be aware you WILL lose all timecode references, so you can't batch capture. Hope you have plenty of drive space. (you could delete all the mt2 files once you've made DVCproHD copies)

When you're done editing, you can output in DVCproHD codec and then use FFmpegX to create the MPEG2 HDV transport stream which DVHScap should send back to the cam. In FFmpegX choose MPEG-TS(mpeg2enc) and then adjust all the settings to reflect the HDV spec (25000Kbs, 1440*1080, 15 frame gop, 384Kbs MP2 stereo Audio etc etc).

You can't go direct from the timeline to HDV mpeg-TS with FCP4.5, you'll need 5 for that.

Dylan, thanks a million- this helps allot until I get FCP5.

Question: why do we end up needing to use codecs that are 4X or more the data rate of the original (under 4MBPS) data stream? This just ends up with tremendous file sizes and increased computer taxing? I tried Pixlet yesterday for the 1st time and I noticed it was almost identical to the same data rate the raw m2t files were: seems more like it- as for the quality- it looked great to me on the Apple Cinema display. I'm sure it's taking a hit that I'm not noticing.......I want 100% quality and of course don't want to degrade the HDV quality with these conversions......do we loose quality in changing from the raw m2t to DVCproHD? I just don't understand why there's no way to keep similar data rates as the raw m2t's?

In FCP5 I'm assuming when working with native HDV the data rates are the same as original and aren't changed in any way- just wondering how Apple seems to have come up with a "Native" solution while there's so much turmoil editing HDV with other programs.

I have a 200GB SATA drive in my Mac setup exclusively for media capture, which is fine for my uses. I'm just wondering why do we have to deal with these bloated file sizes/rates when the original is 3.5 mbps or so!

Because whereas HDV is GOP compressed, DVCpro is discrete frame compression ( or "I-frame", like DV, MJPEG etc) and is much easier for editing programs to handle.

Nowdays, with and more powerful computers and redesigned NLEs it's getting easier to edit GOP material, which FCP 5 can do. So yes, editing native HDV is a great aspiration, if you have the software and hardware that can handle it - your G5 set up should be able to do pretty well - but you'll get a lot more realtime performance even with FCP5 if you're editing with the DVCproHD codec. Anyway the nature of I-frame versus GOP means, that despite DVCproHD being 4 times the datarate of the HDV originals, it is considerably less taxing on the CPU, though it does take up far more HDD space.

I'm afraid I don't know anything about Pixlet, but if it's getting that sort of result it's extraordinary. However I doubt very much that it's realtime enabled within FCP4.5 or 5.

You do loose quality with DVCproHD, and will again when you go back to HDV, but HDV is not really a suitable delivery or archiving codec. Once you move into DVCpro you'd want to find a way to stay there (for example archiving on HDD rather than tape.)

For maximum quality you CAN edit with DVCproHD (or even standard def DV proxies), then reconnect M2V and AIIF files to the timeline and your output will be directly from the HDV originals, but its a more time consuming process.

The Pixlet codec did need rendering of any effects/transitions- so no realtime other than natively dropping it in the timeline (with a Pixlet sequence setup.)

Another Question: Since you've suggested that we can do proxy editing and reconnect original m2v/aif files when the edit is complete- do we lose quality with that workflow?
What about when we send the finished edit back to the camera via FFmpeg- do we lose quality there?

Thanks for answering these questions- I've been wanting to know this stuff and you seem to know exactly what I'm looking to do, thanks again.

Interesting, I'll have to give it a go.

It depends -you could output a 10 bit uncompressed file at about 8GB a minute. If you were to output a compressed format such as DVCproHD you'd loose a bit but not much, although DVCproHD is slightly lower res than HDV.

Cineform claim their codec is "visually lossless" and the tests on their website are very impressive, but it's PC only at this time.

yes, certainly. the HDV compression scheme is fairly severe and recompressing a second time around will definitely add artifacts (depending on the footage) though in only two passes, it won't render it unwatchable. You would still have to export an intermediate in the codec of your choice as FFmpegX can't take data directly from the FCP timeline.

Basically in 4.5, HDV is not like DV where anything that's not been filtered can get passed along directly to the output with no recompression. With HDV everything is going to need to be recompressed, even when going from HDV-HDV

Now with FCP5 things might be different as the HDV is now contained in .mov wrappers, so unnaffected GOPs may be passed directly on to the export file without recompression, but I don't know.

Well, I'm also struggling to get 4.5 to deal with HDV footage (in my case on an aging 450Mhz g4!). Let me know how you're getting on.

In short there seems to be no way to get HDV edited and back to the camera without quality loss- this is sad. Let's hope FCP5 truly does edit it natively without quality loss.

To me the whole point in going HDV was to get higher quality and greater resolution for todays HD sets.....to take hits along the way doesn't seem like a good solution. But- since I'm using FCP4.5 and it's not HDV native- I'll reserve HDV's doom until I try it out with FCP5.

I thought I'd never say this- but I'm missing DV already! If FCP5 can truly work with HDV as it did with DV, Apple has a winner!
(If not- we all lose!)

The weak link is archiving to HDV tape, a 3.2MBs codec on a $6 tape in a $3000 camcorder.

In most cases the quality loss from the render from FFmpegX is minimal, sometime imperceptable given certain non critical viewing conditions. It would still be better at HD than material blown up from standard DV.

I don't know how FCP5 works in this regard, but I do know it has a lengthy "conform" stage before going back to HDV tape. This may mean it re-renders the WHOLE timeline before exporting to tape.

If you want to maximise quality for archiving you might consider alternative approaches, such as archiving on Hard Discs (they're not intended for realtime playback, just storage, so could be cheaper, slower drives.)

Dylan, thanks for the info.
I've finally gotten FCP 5 HD and all my worries are over- it works great!

It treats the HDV the same as DV (workflow wise) you just firewire the FX and capture (even batch capture) and edit just as you do with DV- can't ask for a better solution. Export back to tape in same quality should you want to showcase the project on an HDTV.

tut-tut-tut-tat tahhhhh I'm lovinnn it ! (McDee's theme)