-
iMac Pro or next year’s Mac Pro?
-
[Bill Davis] “The way X works right now – those splits and M&E versions are just pointers to the original files that your desktop computer renders out on demand.”
This is another difference in our worlds. I don’t do the final mix, a sound mixer does, and then they provide the various audio stems required by the network. And then I cut them into my final sequences. It appears you do the final mix, and there’s nothing wrong with that. just saying it’s different. When I cut promos from show masters, I did deliver the final mix, as the audio provided to me was already mixed, and in the separate stems I needed.
[Bill Davis] “Seems to me that with a depository if the original frames in the cloud to link to – it would be massively faster to have the big iron behind the cloud apply the version metadata against the stored frames – rather than to replete upload the same underlying data again and again and again.. “
I’m sorry, do you mean that you load ALL of the audio into the cloud? That seems..odd. If that’s one of those new-fangled things…OK then. Still seems odd to have all the source files on a delivery platform.
[Bill Davis] “Why would you ever need to re upload multiple versions of the same exact frames? “
Because the shows I online, that are cut in FCX originally, BTW, need 8 deliverables. An HD Cut-To-Clock version with commercial blacks, texted. And then the same textless. Then an international SEAMLESS version that has 5 min of additional material, and no act breaks. Texted and textless. And then 4K versions of these, so four more. And the audio config on them is the same….10 channels of audio. And then we deliver separately the 32 different stems they require.
Shane
Little Frog Post
Read my blog, Little Frog in High Def -
[Bill Davis] “See, that’s what I just don’t get.
The way X works right now – those splits and M&E versions are just pointers to the original files that your desktop computer renders out on demand.”1. It’s generally a deliverable requirement.
2. Not everyone uses X.
3. Most often, original files are archived and not available. But if you still have to make changes, it’s often very easy to do this from split track textless masters.
4. Based on your description above (which I understand and also use), every other NLE is capable of exactly the same thing, albeit maybe with a bit more manual intervention.[Bill Davis] “Seems to me that with a depository if the original frames in the cloud to link to – it would be massively faster to have the big iron behind the cloud apply the version metadata against the stored frames – rather than to replete upload the same underlying data again and again and again.. “
That doesn’t exist today and may never exist, because files sizes are increasing and not decreasing. It’s nice in theory, though I would have a hard time trusting it and I think most people feel that way. What about something as simple as having the right third-party plug-ins? To date, no one has successfully demonstrated this theoretical cloud-based online editing concept – at full resolution (like 4K ProRes4444XQ) – with anything other than very simple demo projects.
– Oliver
Oliver Peters – oliverpeters.com
-
[Bill Davis] “And there’s nothing to stop someone from setting up a rules based upload system where small chunks of the master timeline “o-neg” are uploaded as parts of the project are approved – or they go off to full Rez required processes like VFX.”
I think we are talking past each other and I think your comments have changed from the original post. We are talking about delivering finished masters to the cloud and having a zillion variations generated on demand, BUT, from a single, self-contained master file. You seem to be talking about an online-quality editing system with all the media in the cloud. That’s quite a pipe dream.
– Oliver
Oliver Peters – oliverpeters.com
-
This thread just makes me want to buy large quantities of alcohol for all my past present and future assist, coordinators and finish teams. Ive been very fortunate to work with great support teams as I would be very very bad at doing all this kind of stuff.
-
[Andrew Kimery] “If Frame isn’t concerned about file sizes then they really need to get rid of the file size limits on user accounts. ????”
That’s my number 1 complaint against it. I like Frame.io alot and use it alot but they are incredibly tight with their storage and that makes for a very challenging work environment.
I Hate Television. I Hate It As Much As Peanuts. But I Can’t Stop Eating Peanuts.
– Orson Welles -
[Oliver Peters] “You seem to be talking about an online-quality editing system with all the media in the cloud. That’s quite a pipe dream.”
I don’t think Bill is talking about that. Going back to his rendering example, I think what Bill’s brainstorming is a function where you can selectively export portions of the timeline at ‘master quality’ resolution and have them uploaded to a cloud-based service in the background. The service would need to sync w/the NLE to make sure all the chunks ended up in the right order (and older chunks were correctly replaced by newer versions if changes were made). Once you upload your audio splits (either made by you or given to you by a rerecording mixer) you can set your audio configs, choose the texted/textless version, etc., and queue up all the different versions you need then hit the ‘go’ button.
It’s an interesting idea, though still very impractical for long form projects given the current state of Internet service in the US and the limitations of cloud hosting companies (ex. Vimeo Pro only lets you upload 20gig/week). Also, there’s no way for you to QC the final output without downloading it and watching it yourself, which pretty much negates the whole point of having a cloud-based service generate and deliver your files for you.
-
[Shane Ross] “Because the shows I online, that are cut in FCX originally, BTW, need 8 deliverables. An HD Cut-To-Clock version with commercial blacks, texted. And then the same textless. Then an international SEAMLESS version that has 5 min of additional material, and no act breaks. Texted and textless. And then 4K versions of these, so four more. And the audio config on them is the same….10 channels of audio. And then we deliver separately the 32 different stems they require.”
Thanks for posting this Shane. It brings back memories for me for when I used to work on Discovery Channel shows. These tech specs sound very familiar and as I recall it was quite an exhausting process to create those masters.
I Hate Television. I Hate It As Much As Peanuts. But I Can’t Stop Eating Peanuts.
– Orson Welles -
Shane,
This to me is a logical progression of the changes I’ve experienced over the past six years.
I do VERY little the same way I did it back then.
(And yes, to Olivers point, this is not just about X – it’s about the entirety of the new era of digital delivery.)
Maybe it’s my lucky fiber line access – but I simply think about digital files differently now.
They aren’t a fixed “thing” to me anymore. They are part of a living connected stream that I can touch and change any time.
The old way I’d do that was to “load up” my project on my machine and basically master a stand-alone file to send to someone. Actually, as you indicate – lots and lots of standalone versions, often. That’s less and less how it works for me anymore. While I don’t face the same delivery needs that you do, my deliverables are basically all now connected files on-line.
I don’t “push” them to clients – they “pull” them as needed.
And increasingly, I’m starting to see that things I expected to need to do BEFORE the upload – like versioning – can be more efficiently done in the cloud responding to metadata I apply against already uploaded content.
Right now this type of thing is still somewhat unusual, I’ll grant – and I’m no big pioneer in it – I’m just noticing how what I’m being pulled toward today works much better and faster than what I used to do – and it seems to me that the benefits are so HUGE I have to believe it’s going to grow in this direction.
My delivery is an ephemeral connected file on a server. (The “connected” part is the key)
There is no actual MASTER or VERSION as a stand-alone thing – just an expression of what somebody needs in the moment – exported for the moment – revisable in 10 minutes if a flaw is discovered or a new idea is useful to the work.
It’s a change in my thinking.
And I’m just trying to figure out what’s going to be important in this new system. At least what’s important that’s distinct from what’s been important in the past – when a master was a fixed thing on a tape or disk – that was relatively unchangeable past the moment of its creation.
FWIW.
Creator of XinTwo – https://www.xintwo.com
The shortest path to FCP X mastery. -
[Andrew Kimery] “Also, there’s no way for you to QC the final output without downloading it and watching it yourself, which pretty much negates the whole point of having a cloud-based service generate and deliver your files for you.”
Well, right now the FiO Approve/Needs Work tagging system could easily grow into something to manage versions – and really, all you’d need for QC is for someone to watch the deliverable version on line and vet that the file confirms to your requirements. No need to download it, really.
Further, if it’s just matching a distribution requirement checklist – a machine learning system should easily be able to compare the deliverable to your last export and only flag it if it differs.
Fun to think about.
Creator of XinTwo – https://www.xintwo.com
The shortest path to FCP X mastery. -
[Bill Davis] “Well, right now the FiO Approve/Needs Work tagging system could easily grow into something to manage versions – and really, all you’d need for QC is for someone to watch the deliverable version on line and vet that the file confirms to your requirements. No need to download it, really.
Further, if it’s just matching a distribution requirement checklist – a machine learning system should easily be able to compare the deliverable to your last export and only flag it if it differs. “
QC’ing a relatively low res proxy stream of the high res master wouldn’t allow you to catch things like banding, gamma shifts, sync, odd motion characteristics, etc., because you wouldn’t know if those anomalies existed in the master itself or were only present in the proxy stream you were monitoring. A person, and possibly machine learning in time, could certainly compare a low res reference export to the final master export the cloud service creates to make sure there are no discrepancies in the edit, L3rds, etc.,.
Reply to this Discussion! Login or Sign Up