SOT:Hollywood movie post production

NormanPCN wrote on 4/10/2014, 1:35 PM
This is a long 35min video but an interesting watch.
It is about the post production of, The girl with the dragon tattoo.

Prepping for a 4K world.

I recently posted this link , buried in a long thread, but some may not read that and might be interested in some of the workflow and technology talked about in the video.


wwjd wrote on 4/10/2014, 2:17 PM
thanks! I love this kinda thing.
paul_w wrote on 4/10/2014, 4:02 PM
good find.
My own light bulb moment with 4k was thinking 'ah, now we can shoot the same res we scan film'. And apparently output it that way too now. Seems pretty obvious to me that 4k would be hit.
Sony asked my over Twitter once (no big deal, amongst many others) whether i thought 4k or 3D was the way forward for new TVs an cameras. Well, after thinking for all of about 1mS, my answer was 4k. Glad to see it out there and being received so well. Cost is coming down fast too and processors and storage are only ever going to get faster and cheaper. If not today, maybe tomorrow.

OldSmoke wrote on 4/10/2014, 4:07 PM
4K@60p that would be nice! I cant wait to get my FDR-AX100 but it only does 4k@30p which I might accept thanks to Paul.

Proud owner of Sony Vegas Pro 7, 8, 9, 10, 11, 12 & 13 and now Magix VP15&16.

System Spec.:
Motherboard: ASUS X299 Prime-A

Ram: G.Skill 4x8GB DDR4 2666 XMP

CPU: i7-9800x @ 4.6GHz (custom water cooling system)
GPU: 1x AMD Vega Pro Frontier Edition (water cooled)
Hard drives: System Samsung 970Pro NVME, AV-Projects 1TB (4x Intel P7600 512GB VROC), 4x 2.5" Hotswap bays, 1x 3.5" Hotswap Bay, 1x LG BluRay Burner

PSU: Corsair 1200W
Monitor: 2x Dell Ultrasharp U2713HM (2560x1440)

dxdy wrote on 4/10/2014, 5:02 PM
30 minutes well spent. Thanks for the pointer.
Cliff Etzel wrote on 4/10/2014, 5:43 PM
Before anyone gets too excited over 4K IMO, one should watch this video by Michael Cioni/Light Iron to see what they are having to do infrastructure wise to prepare for 6k-8k workflow. 4K is still going to require a large investment not only for post, but also for archiving and we aren't there yet for optical storage that can handle that amount of raw data.

4K sounds fine and dandy but I have to ask who is going to have the amount of investment capital needed to build an infrastructure to handle the raw data that is entailed for 4K. One has to define their delivery platform and consider their needs versus wants accordingly. And secondly, how many people are going to have the ability to watch said 4K material. I know most are looking at conforming to HD for final delivery but It seems to me that 4K is creating the ability to get sloppy in shooting technique/skills for the sake of being able to "fix in post" later.

Call me old school, but I consider fixing in camera first to alleviate longer times in post the sing of a skilled DP/shooter.
VMP wrote on 4/10/2014, 6:23 PM
Interesting video post Cliff Etzel.

Before you know it a simple laptop could do all this storing and processing in the near future. Sooner than we might expect.

But what is to say that 100K video wouldn't be the trend then.

videoITguy wrote on 4/10/2014, 7:05 PM
What 4K will probably NOT be - a chain of acquisition to a delivery device. So then you ask why do we now have 4k TV sets? well, the answer is people and pyschology of perception - 4k at home will not mature into what HD 1920x1080 acquisition did for Blu-ray. The chain to delivery will not be a constant of the resources from production to storage to delivery as it was during the HD revolution of a couple of years ago. 4k will not be perceived as that much value added as HD was over SD. Sure some value, yes, but not enough for the average consumer.

4K will be an acquisition format for home delivery in the traditional HD end-products - be it burned Blu-ray, or NetFlix stream, or the local theater screen.
wwjd wrote on 4/10/2014, 7:11 PM
for me, as a guy that wants to make films, 4K is great for so many reasons:

- way better looking HD, than HD cameras. HD upscales to 2K movie projection just fine, but why not get the extra detail if it is possible downscaling to 2k from 4k?
- reframing and zooming
- fixing in post: just because they didn't used to be ABLE to do that, doesn't mean we shouldn't be able to now. The tools got better... why not use them?
- I'm looking at 4K file space like this: MORE BETTER planning of filming shots, less "shoot tons and cherry pick"... kinda more like film days when you had your 10 minute film roll... shoot it good in 2 or 3 takes, not 30 like you can in digital

We're only halfway to IMAX quality in digital, but even IMAX looks better in 2K theaters: Dark Knight for example. Why NOT allow tech advances to make things better?
4K tvs and cameras available doubled over the last few months, no reason to think people won't be buying them. And I watch 4K content on my HD screen all the time - looks fantastic even if not viewed in 4K, so I imagine people will enjoy it even in HD.
Byron K wrote on 4/11/2014, 2:45 AM
Thanks for sharing that NormanPCN. Watched the whole thing and thought the talk was very interesting. I had no idea that Girl w/ Dragon Tattoo was done in 4K, which, imho, thought was a pretty good movie btw.
farss wrote on 4/11/2014, 5:53 AM
Thanks, a very good watch.
Interesting that this was a presentation to LAFCPUG and yet in the entire workflow FCP was only one cog in the machine but no one in the audience threw things at the presenter when he spoke of the virtues of PPro, AE and Clipster as tools to get the job done.

Unfortunately in this workflow FCP cannot be replaced with Vegas, I wish it could and it hinges on just one thing that Vegas cannot do, render out video and preserve source timecode. I find it quite remarkable that in the journey from film to DV to 2K and beyond the one thing that hasn't changed is timecode. On the upside V13 will soon be with us, maybe, just maybe there'll be an Easter Egg that'll end my decade of whinging about Vegas and timecode.

NormanPCN wrote on 4/11/2014, 12:05 PM
He was a little murky on the FCP to PPro switch. Something about switching out the ProRes file back to the RED file to render uncompressed. Did they do a render from FCP and bring that into PPro? Maybe in FCP you cannot swap files from under the editor.

Anyway. Vegas certainly does not mind having files swapped out from under it. This is how people have been doing proxy editing. So if Vegas were used you would cut the ProRes proxies, swap the proxies out to RED raw, and then render DPX. No need for an application switch.
farss wrote on 4/11/2014, 5:27 PM
[I]"He was a little murky on the FCP to PPro switch. Something about switching out the ProRes file back to the RED file to render uncompressed. Did they do a render from FCP and bring that into PPro? Maybe in FCP you cannot swap files from under the editor."[/I]

To go from a FCP project to a Ppro project he used a program called Automatic Duck. It's now supported by SSL.

[I]" Anyway. Vegas certainly does not mind having files swapped out from under it. This is how people have been doing proxy editing."[/I]

That's true except the way Vegas does everything is based on time into file NOT source timecode which is what everything uses and for good reasons.


VMP wrote on 4/12/2014, 11:17 AM
@ Bob. "..with Vegas, I wish it could and it hinges on just one thing that Vegas cannot do, render out video and preserve source timecode. "

Can't this be solved by setting the Vegas timeline ruler timecode according to the source file?
That would change the embedded TC in the final rendered file.

Or do you mean passing through the original take TC of each individual event-clips on the timeline? If so wouldn't the final rendered file have timecode jumps throughout playback? (According the events with different TC on the timeline)

I can only imagine what timecode will be shown during a two clip transition.

Just trying to visualize what you mean :-).

In Vegas until now I have made use of the timecode ruler setting with the render of MXF XDCAM 422 files (which embeds the time code in file), the TV broadcast system here requires it to start at 00:01:55 feature starts at 00:02:00, I guess it has something to do with the ancient tape queing system which they are still using with the MXF files.

rmack350 wrote on 4/12/2014, 4:12 PM
VMP, you've just pointed out part of the key to getting Vegas to (accidentally) preserve timecode if you're just trying to transcode clips using the timeline. Vegas isn't really the right tool for transcoding, but you can do it.

So, Vegas can write the timeline's timecode into XDCAM 422 MXF files. That's a good start. And a while back Bob brought up the question of laying out clips on the timeline to match their timecode, and it turned out to be a dead simple copy and paste action in the edit details window. So, once you've got all your clips on the timeline with the clip Timecode matching the timeline time, you could run a script to render each clip one-by-one, and they'd have the same timecode (or starting timecode anyway) as the originals.

A neat trick.

This may not be what Bob is talking about but it's one of the possibilities.

My guess is that if SCS is looking closer at proxy workflows in Vegas then they're probably also looking at better support for timecode in media files. I know I was hugely pleased to see that Vegas 12 FINALLY could read timecode in Quicktime files, so maybe they're getting serious about this sort of thing.

NormanPCN wrote on 4/12/2014, 7:03 PM
What file formats (templates) can Vegas render a timecode to a file beyond any of the MXF file variants it supports?

Looking at how Vegas implements the Smart proxy feature, I think all Vegas cares about for a proxy is that it is frame accurate with the source. Meaning if the source is 1000 frames then so is the proxy.

Vegas actually encodes smart proxies with a little "trick" in this regards. They are always XDCAM EX 720 24fps 35Mbps files, but no resampling occurs since they are frame accurate. So if your source was 60p and you looked at the length field in the header the proxy would be 2.5 times longer to be frame accurate.
wwjd wrote on 4/12/2014, 7:27 PM
...and my brain exploding in 3...2...1....
farss wrote on 4/12/2014, 8:19 PM
[I]"...and my brain exploding in 3...2...1.... "[/I]

It's not all that hard to get ones head around.
Simply put when a movie is shot it's enormously helpful if each frame of vision and audio can be uniquely identified using some scheme. If one makes copies, dub or proxies each frame of that should be identified by the same identifier otherwise expensive chaos can set in.

The way it was done with film was edge numbers. The editors typically cut B&W proxies (workprint) and those proxies were made optically with edge numbers printed from the camera original. When the edit was finished the camera negative was cut to match the workprint from the edge numbers.

Nothing changed really when video tape came along and nothing changes with an entirely file based workflow.

The approach SCS seems to have taken sort of seems fine. Dispense with all this timecode thing and just assume every file starts at 0. Pretty simple and easy to implement but there's a problem. That assumes that when the director called "ACTION" every thing recording sound and vision started recording at exactly the same time. Of course that never happens and to maintain order the production distributes timecode via various means. That way it doesn't matter if the sound guy hit Record 10 frames after Camera 1 and Camera 2 was 13 frames behind. Everything can be lined up in post from the timecode.

Of course you can do this in Vegas, no real problem as we've discussed but that breaks down when different people with different systems are going to work on the movie. The people doing the audio post production don't need the massive RAW files, tiny SD proxies are all they need. This is where the Vegas approach comes unstuck. If you send them proxies that start at 0:00:00;00 how are they going to sync up the audio files from the timecode in them? How is an editor going to sync up the video taken of one shot when two or more cameras we used if all he's got are the proxies that start at "0:00:00;00"? The very useful information that the camera timecode carried is lost.

musicvid10 wrote on 4/12/2014, 8:57 PM
And just when we think we've got it all figured out, along come the Variable Frame Rate camcorders:
Panasonic AG-AC160A
For.A VFC-7000

Can vfr consumer models be far behind?
farss wrote on 4/12/2014, 9:43 PM
Panasonic's Varicam has been around for ages. That was a challenge to deal with. For a while there was a camera in this country but no VCR that could play the tapes.

wwjd wrote on 4/12/2014, 10:43 PM
I'm not sure how my edits stack up as far as complexity goes, but with like 20 audio tracks and 10 video tracks of tiny little clips and overlays, and dubbed audio and sound effects etc, I just lined it up by hand. But I'm not doing a weekly TV show that has to be slapped together quickly from multiple cams and an audio mix.
Back in the day, I understood timecode to keep all the random external machines on the same page, but (and I am just a hobbist) I don't see the need now on an NLE. BUT BUT I'm not sending my stuff out for POST work either.

Not downplaying TC at all, but it seems like an extra time consuming layer in my workflow. My time code is the 23.976 fps on the time line.
rmack350 wrote on 4/13/2014, 2:33 AM
Thanks Bob,

I was looking for good examples and had completely missed the problem of syncing dual system sound. Excellent example.

ushere wrote on 4/13/2014, 4:59 AM
i have worked with time code for the last 40 years or so. bob sums it up quite succinctly...

one thing that Vegas cannot do, render out video and preserve source timecode

NormanPCN wrote on 4/13/2014, 10:43 AM
For a single file timecode it certain can. The Vegas rendered timecode comes from it's timeline. If you copy a source timecode to the timeline they will match.

If you have two or more files on the timeline and you want the timecode to change from one clip to the next, then no.
videoITguy wrote on 4/13/2014, 11:05 AM
NormanPCN - you are missing the entire point of the timecode hitherto discussed - Bob stated this "That's true except the way Vegas does everything is based on time into THE file NOT source timecode ". This is an exact description of what Vegas is engineered to do - create time out of the run of the file. Very similar to what "run time" would be in less expensive video camera running.

As pointed out - timecode for the hobbyist, run and gun shooting with a single camera , and other simple shooting scenarios is not a big deal. This is why Vegas is NOT perceived to be powerful enough for editing group collaboration. Given many cameras, many audio sources, and many post-production steps you want an NLE to preserve timecode.