Hi All.

I'm interested to know how other pro users amongst us are tackling their AVCHD workflow and if anyone is successfully using any kind of hardware acceleration. At the moment, we still transcode to ProRes for several reasons - it's easier on our quad core i7 processors, it "front loads" the transcoding burden to the start of a project, rather than leaving it to the end of a project (a la Premiere Pro) and it gives us a resilient archive "mezzanine" format which will cut with anything else from our recent archive. We're cutting with Final Cut Studio 2/3. So far, our workflows have been based on the 2 following software transcoders:

1) Apple Compressor - desperately slow to load the AVCHD files. On a recent 2 camera event job, it was taking a couple of hours to load the 4GB AVCHDs and was crashing if we loaded up too many files.

2) 5DtoRGB Batch - another nightmare. "Known" bugs include the first file in the batch hanging 2 frames before completion. It didn't say that on the App Store. I got my money refunded.

Anyone got any suggestions for another proven workflow that will happily take a batch of large-ish AVCHD files and reliably transcode them overnight to ProRes 422?

Google searches suggest nothing has really progressed further than 2009 and that it's a bit of an "elephant in the room."

Thanks in advance,

Ben.

ClipWrap. It's worked perfectly for me when editing in FCP. Back up a card, point it to that folder, and transcode it to whatever flavor PR422 quick and easy. It does batches by default and I've never had a hiccup, even on clips that constantly error out on FCP L&T.

I recently bought CS6 and it's been delightful in handling even multiple streams of AVCHD on an older (~2009) MBP... then just render out what's in the cut.

ClipWrap gets a very strong second or (don't shudder) Final Cut Pro X, works raw AVCHD from the GH2 just fine. Renders fast.

Thanks for the suggestions, chaps.

I gave Clipwrap a spin today and, based very unscientifically on just transcoding one minute's worth in the demo version, it seemed much quicker and certainly didn't have the huge start-up delay we have been experiencing with Compressor.

We've got FCP X as well, so will have to be dragged kicking and screaming towards using it more often, I guess.

Thanks again,

Ben.

Compressor examines the entire file before letting you do anything with it. Of course there is nothing in the program that tells you that, just an alarming beach ball. If you have used MPEGStreamclip with DVDs made in a DVD recorder, it does the same thing with DVD files that have "broken" time code which some DVD recorders create, some don't. MPEGStreamclip, to the credit of it's programmer, tells you what it's doing. ClipWrap quickly creates a frame reference wrapper for the AVCHD files so FCP can tell how to distribute the compression without heavy calculations for off key frame cuts.

it "front loads" the transcoding burden to the start of a project, rather than leaving it to the end of a project (a la Premiere Pro)

What does this mean? Premiere Pro edits AVCHD natively - and renders very quickly. I don't know of an NLE that doesn't require rendering out a timeline, reguardless of the file format. So rendering should not be considered a disadvantage to using AVCHD with Premiere. The real question is - do you want to sit and wait for your footage to transcode to a x10 larger file before you start a project...or just start editing.

Angelo, my comment is based upon a recent 2 day course I did, courtesy of Adobe, with a pre-release of CS6.

My understanding from them was that PP essentially decompresses all formats out to base band (so they're not actually native as such) and that, unless you have the NVIDIA CUDA card with the accelerated GPU processing, you're not really taking advantage of what CS6 offers and you're effectively moving the burden of transcoding to the end of the project, when things are often most tight for time.

According to Adobe, our 2 year old quad core i7 iMacs just aren't man enough to cope with AVCHD effectively with CS6 and we'll continue to have an unsatisfactory experience playing back AVCHD/H264 files with dropped frames etc, until we do an entire tech refresh with monster spec Mac Pros etc.

Coupled with a need to deliver/archive many of our masters at higher quality than highly compressed web deliverables, that's what is driving me towards sticking with ProRes as a mezzanine format for now.

Could just be old-think and my "safe" ex-BBC approach - but it's something I'm trying to evaluate as clearly as possible before everything changes again in 5 minutes...

Ben.

"The real question is - do you want to sit and wait for your footage to transcode to a x10 larger file before you start a project...or just start editing."

BTW, Angelo, that sounds spookily like an Adobe ad... :-)

Ben.

Angelo, my comment is based upon a recent 2 day course I did, courtesy of Adobe, with a pre-release of CS6.

My understanding from them was that PP essentially decompresses all formats out to base band (so they're not actually native as such) and that, unless you have the NVIDIA CUDA card with the accelerated GPU processing, you're not really taking advantage of what CS6 offers and you're effectively moving the burden of transcoding to the end of the project, when things are often most tight for time.

According to Adobe, our 2 year old quad core i7 iMacs just aren't man enough to cope with AVCHD effectively with CS6 and we'll continue to have an unsatisfactory experience playing back AVCHD/H264 files with dropped frames etc, until we do an entire tech refresh with monster spec Mac Pros etc.

Coupled with a need to deliver/archive many of our masters at higher quality than highly compressed web deliverables, that's what is driving me towards sticking with ProRes as a mezzanine format for now.

Could just be old-think and my "safe" ex-BBC approach - but it's something I'm trying to evaluate as clearly as possible before everything changes again in 5 minutes...

Ben.

Thanks Ben!
That makes me feel much better that I couldn't make the jump to CS6.

ProRes is an industry wide delivery format now and is a wise place to stay with for the moment. Yes, we can edit AVCHD (or MPEG2HD) natively with several NLE programs but is the experience actually helping your editing workflow? For quick, mostly cuts only, editing, dropping the AVCHD files in and getting a cut out is great. But once the edit gets complex with effects, titles and color correction, using ClipWrap to move the files over to ProRes is frequently the better way to go, especially for multi-cam switching. Pre-screening the clips before transcoding is a way to conserve drive space if that is an issue. I usually edit native for quick projects and export a ProRes edit master for distribution and transcoding to optical disc or web formats. Multi-cam edits are always converted to ProRes first, that is a big difference in editing ease.

A friend shooting in MPEG2HDV in Alaska and editing on his laptop was trying to work the files in native codec directly but his titling needs would cause the edit to grind to a halt just to preview the results. Switching to ProRes files, as I recommended, will speed up the process considerably and his laptop will run much cooler.

You will need Pro-res only in FCP. Don't know about Avid or Vegas.
I use PremierePro Cs5 and import AVCHD straight in. Mix it up with XD IMX50, Digi, etc. and it makes little difference and the timeline (using CUDA and MPE) has absolutely no problem at all.
I guess you need to think about your end format - what is the program being viewed as - Broadcast, Blueray, quicktime on the web, corprorate promotion in a boardroom. I find a lot of discussion is total overkill about miniscule output detail, when in actual fact, you need to set up your project for the end audience format.

I use Edius and AVCHD is handled like any other footage. Just place it on the timeline and edit. The files are handled natively.

Sorry to hear you have to transcode before adding to your timeline in FCP as this sounds like a FCP specific situation in dealing with AVCHD files. Not fun.

Boy this all sounds complicated. I just pull the files straight from my gh2, drop into premier and go at it. Effects, color correction, not much really gives me grief. I tend to throw everything out into quicktime animation for intermediate moves to after effects and back.

No issues thus far.

Valid answers, but I'm interested to know what kind of stuff you're cutting.

We do quite a lot of live music events - often 5 iso cameras because we can't get a switcher in there and the get in/out is tight. We shot a recent one with a UK artist called Ed Sheeran, 5 x 5DIIs and the final piece was 80 mins long. We had 24 hours to travel 350 miles, cut it, grade and caption, output to ProRes (one of our deliverables) and compress that down to a 5Mbps 720P web deliverable.

Thankfully we have 100 Meg broadband, as we tend to have to deliver this stuff right to the wire and the contract stipulates we don't get paid if it's not on the client's server by 5pm the next day.

This is one of the reasons I don't think it's as simple as "chuck it straight into Premiere" - maybe for short-form stuff it's fine, but long-form cutting with 5 layers plus captions and a basic grade? There's quite a lot at stake if we get this wrong and a ProRes workflow still feels like the lowest risk approach. It would just be good to get AVCHD material into the system more quickly.

Ben.

It would be nicer if we could get acceptable 24p out of the GH2's HDMI port and into a ProRes recorder. Alas, Panasonic wants you to purchase a $4000 camera to do that. You might be able to do that with 25p and an Atmos Ninja recorder but the portability of the camera is slightly reduced.

Interesting, I had a feeling that you were doing multi-cam. 5 cameras of AVCHD is beyond the capabilities of most systems to do a multi-cam live switch edit in full resolution without choking. And with your timeframe capturing low-res proxy files will not work. Do you have a laptop on location specifically to transfer camera files? Converting to ProRes with ClipWrap takes longer than just wrapping but it's not overnight for sure. Once again you should research FCPX as it's multi cam options are top of the field in some ways.

I use toast 10 titanium to convert to prores
It loads and processes just fine up to 32 gb
Lately audio started to get out of sync for 6 frames in end prores files, but that's fixable

William, I think you will find atomos and GH2 won't go together, something to do with the HDMI out format. There was a discussion here a while ago. Although atomos does record in pro-res, there is a compatability issue with GH2's output.
Back to the original thread, putting 80 minutes of 5 layers of AVCHD from 5 Mark 2's sounds like a huge job plus I thought they only did 12 minutes in a run, so a nightmare to synch.

We stagger when we turn over and shoot separate audience cutaways as a "get out of jail free" for the odd occasion when all 5 shots coincidentally don't work - fairly rare, and I find 4 or 5 cams the sweet spot for small venue gigs like this.

We have an assistant who takes cards and transcodes on a laptop and the remaining files are processed on the train on the way home.

It's not an ideal way of working and we much prefer to do a live switch with isos - but we've learned a lot along the way and knocked a few months off our lives in the process...

Ben.

Yikes! Perhaps you should get at least one GH2 as a cover camera as it will not shut off after 12 minutes. When I would shoot boxing events years ago with iso cameras and we couldn't get timecode on all the cameras, we would start at the beginning of each match and stop soon after the winner was declared so to make it easier to hold sync during edits and control when tapes were changed. What's a small venue where you are? Here in NYC, five cameras would occupy half the audience in some places!

The Atmos works with 30p on the GH2 and perhaps 25p, it's the 24p output that is wonky and not usable.

We've bought 2 GH2s in the last few months and have used the hack just to extend recording times - just got caught out on the recent conference job we did which then took an entire weekend to transcode (simply because Compressor and 5DtoRGB were both crashing/hanging.)

Clipwrap has been bought...

Ben.

Valid answers, but I'm interested to know what kind of stuff you're cutting.

We do quite a lot of live music events - often 5 iso cameras because we can't get a switcher in there and the get in/out is tight. We shot a recent one with a UK artist called Ed Sheeran, 5 x 5DIIs and the final piece was 80 mins long. We had 24 hours to travel 350 miles, cut it, grade and caption, output to ProRes (one of our deliverables) and compress that down to a 5Mbps 720P web deliverable.

Thankfully we have 100 Meg broadband, as we tend to have to deliver this stuff right to the wire and the contract stipulates we don't get paid if it's not on the client's server by 5pm the next day.

This is one of the reasons I don't think it's as simple as "chuck it straight into Premiere" - maybe for short-form stuff it's fine, but long-form cutting with 5 layers plus captions and a basic grade? There's quite a lot at stake if we get this wrong and a ProRes workflow still feels like the lowest risk approach. It would just be good to get AVCHD material into the system more quickly.

Ben.


Ah, you have some pretty difficult circumstances there. Though I'm curious now as to the capabilities of premiere with such a workload. It might be surprising depending on the system.
It would seem if you could avoid the initial prores ritual it would really save a lot of time.

Ben, I must say I am confused about which cameras/codecs you are referring to in the thread. 5DMKIIs or GH2s?

The 5DMKII does not shoot AVCHD while the GH2 does. Are you looking to switch from using 5DMKIIs or did I misread and you are using 5 - GH2s?

In any event, I am guessing you are tied to the Apple platform? Because in the quick turnaround situation you are working in, I would suggest a look at Edius. This is where Edius is strongest, high quality & quick turnaround. A lot of same day edit folks use Edius.

Sorry to offer a PC option but AVCHD is a native format for Edius. Meaning put the original clip on the timeline, do your primary & secondary color correction and it still plays back in realtime. When your cut is finished you export. Simple. The multicam is great as well.

Tim.

Sorry for the confusion - I have a couple of GH2s that I'd like to use more, but I used the shoot with 5DIIs as way of an example of the tight deadlines we often have to work with. With our recent transcoding workflow, we'd never be able to deliver an AVCHD multicam shoot with that timeline.

We have an entirely Mac setup.

Thanks,

Ben.

Both cameras shoot h.264/MPEG4 and wrap them in in their own ways, Panasonic in the MTS file format and Canon in the MOV format. Neither play well with Final Cut 7 without rewrapping or transcoding.

Importing GH2 footage into Final Cut X is very quick (and I would assume the same with the 5D) and I suggest you experiment with it with a project that has already been completed. Try importing the raw files vs. having "optimize media" checked in terms of time saved. The multi cam function in X is very, very good although I think a laptop would choke on 5 cameras in AVCHD but I could be wrong. You have it, try it, the program is way better than the bad press would lead you to believe.

I've been following this thread and as a recent GH2 puchaser, I am more than a little bit intimidated by all the chatter about AVCHD. Plus, it sounds like you're all on Macs, except maybe for Tim Polster (Edius is PC only, if I'm not mistaken) and I'm on a PC still running PP CS4. Unlike you heavyweights, shooting is my second job and can't quite afford the jump up to CS 6. The only saving grace is a fairly strong computer with an i7 CPU and 16 gigs of memory, and a decent video card. But you mateys are scaring the bejeezus out of me, so I'll stay tuned for any suggestions you may have! ;-)

Cheers!

Nothing to be scared about. Mr. Giles is working in an exceptional situation where the recording codec becomes an issue to consider. Your computer is fine for all sorts of editing with AVCHD however I would suggest that you look beyond PP CS4 as my experience is that Premier really became a great program with CS5. There are a number of decent low cost PC editing programs that have been updated recently and will leverage your computer's capabilities. Perhaps someone with more experience with these programs can give their opinion.

This is one of the reasons I don't think it's as simple as "chuck it straight into Premiere" - maybe for short-form stuff it's fine, but long-form cutting with 5 layers plus captions and a basic grade?.

I have a I7 950 and can edit multicam with 4 layers of canon dslr 1080p footage native in edius 6 including grading, works with no issue, only when I add a 5th layer it stutters. Also exporting 4 layers multicam of graded dslr to a 1080p file takes 20 minutes for 1 hour of footage into canopus own hq avi codec and about 25 minutes if i would convert to a 720p/25mbs file. Edius was build for heavy multicam lifting and it can do all that without without any advanced gpu support or even having to be 64bit. I"m pretty sure that a recent cpu will handle 5 dslr streams with ease considering my cpu is allready ancient.