What are y'all using for post-Flame versioning?

We are drowning in versions, so to speak. After I export a spot from Flame the folks in my machine room have to make a gazillion files for digital delivery. They’re using a cocktail of Media Encoder, Compressor, Sorenson Squeeze (I know, but, it works) and more. This is a process that is begging for some automation. Is anyone out there using an all-in-one or configurable/scriptable system to handle this? Would love to hear about it.

2 Likes

Back at my agency job, I looked into a program called CONTENT AGENT. It seemed pretty awesome. You can build a node tree where you give it one master file - let’s say a Slated 8Ch ProRes w 5.1 and Stereo, and build the deliverables that can do things like:

  • ProRes deliverable that truncates the name to the ISCI only (for Extreme Reach)
  • h264 deliverable that encodes only the stereo audio (for traffic to approve)
  • h264 that chops off the slate, stereo only and adds 1 sec of black to head and tail (for client review)
  • ProRes downconvert to NTSC/PAL-Stereo (this was a few years ago I looked into it)

Lots of other cool stuff, but I was really attracted to the idea of exporting ONE master, and having automation do all of those audio and slate/no slate tedious stuff… You could even have it use different watch folders for different clients/deliverable trees.

It was a Windows only product, and I ran a demo on virtual machine, but I never got it up and running on a production machine.

Curious if anyone else is using that and if it really is the bee’s knees…

1 Like

Thanks Dave. Yeah…I’d love to see if there’s a Mac solution out there, only because were all Mac based in that dept. I’ll check it out!

@Gunpowder friends have a ton of experience in this world.

This looks cool. Any idea what it costs?

I think we were dealing with a reseller that set up the demo. He was recommending a pretty powerhouse PC with dual GRFX cards for encoding power, etc… I want to say the ballpark was $10k for a functioning system, and my boss was like, “Well - we don’t pay you overtime, so why do I care about saving you time?” Paraphrasing, of course…

1 Like

Dang. That’s about what I expected given the “contact us” on their site. We have a sweet IO department that handles everything with a smattering of Mac-based tools, but $10K/seat will be a hard sell. Thanks, though!

Yeah - it would be spendy if each seat were 10k, but I’d hope it was more like you spend $10k once to set it up, and then once you set up the tree for each client, everything is automated and you could get by with only one system - at least for a shop of our size. We had 3 Flame suites at the time. I can’t remember if it was expandable to create a “farm” or not…

Ive always liked vidispine/vidinet - Media Services in Vidinet | Vidispine

It certainly used to be very heavily code driven and would require a tonne of custom work, but these days vidinet seems pretty straight forward, and i think you get $300 of credits when you sign up.

Never used it in production, but given the time and the need, vidinet would probably be the first place id look for an automated transcoding solution

I suspect that since we are such a niche within a niche industry when it comes to these kinds of encoding problems and this scale that whilst we are in fact drowning in versioning-debt so to speak, we are chump change. Back at the old place, some of the solutions we sought at our scale were probably 6 figures and still weren’t great. To have automation and high speed rendering likely means the inability to do really simple things, like, add your own encoder presets. So, they aren’t quite built for the willy nilly tech specs and schedules of commercials.

2 Likes

Years ago I worked at a shop that used telestream(episodes?) and iogates

2 Likes

This is where we need a “not like” button. I know that attitude too well.

I make every version that is it’s own unique pairing of audio and video and slate. That usually amounts to 3 to 6 versions of every spot. I export those at native size, 422HQ 24b audio. The graphics assistant makes codecs and sizes from those using media encoder (I think.) I’m the only one generating final picture, however, so it’s not overwhelming for the assist. Just me. If we were a bigger shot, I might consider only making the unique picture, then having an assist do slates and audio on a flame assist.

When you say gazillions of versions do you mean wmv, real, mp4 etc?

I thought this had all gone away with the advent of YouTube and Instagram?

In my time I have used content agent and telestream and Terran media cleaner is you remember that! My point is, it’s a faff. Can’t you sub it out?

To Quinn’s point, we are using ffmpeg with a python hook to make after edit dubs for client approval. It’s very possible. I didn’t make it but can say that when I’ve used ffmpeg on command line on Mac it’s been a piece of piss.

Versions look daunting on the spreadsheet. This is my wheelhouse and what I am comfortable with so I will briefly share a few of my sorting tips.

  1. To my knowledge there isn’t a quick solution to the labeling of files. This requires a human to make the decisions. The label from the export is key. ISCI first please.
  2. After sorting through the spec sheets, sometimes there will be several specs that are the same as other requests but requires specific labeling for your non-technical clients to identify and feel special. The YouTube, Hulu and Twitter specs go a long way…with correct labeling.
  3. Adobe Media Encoder (CPU instesive - do not multitask on this step) and ffmpeg both offer drop folders. In theory you can “set it and forget it” however it still requires a human to sort, label, QC and direct the files (product) to the final destination.
    Delivery is real.
    I hope this helps.
3 Likes

Thanks Tracy, and welcome! :slight_smile:

1 Like

@johnt & @tmhollins I’m talking about the pile of files that are made after flame export. I’ll export a Prores HQ or 4444 and my MCR team has to make 1-40 files from that. Diff codecs & specs. Couple of examples that go outside of Media Encoder (which we are using):

  1. I export with a slate, but they need a version without a slate. Would be great to have a preset that trimmed that.
  2. I export a 23.98 master. 2997i and SD need to be made. Occasionally there will be a flicker frame across the 2pop. I end up doing this version in flame so I can manually check for that offending 2pop flicker frame and remove it.
  3. Audio relays. Stereo, 5.1, whatever.

To Tracey’s point: A lot of this needs to be checked by a human being. That’s why, currently we are doing a lot of this in Flame. It seems like Media Encoder does a bunch of stuff, but then lacks. I guess they all do. That’s why I was curious to see what people have had experience with.

Thanks everyone!

2 Likes

Hey Andy,

We’ve made two Python scripts in house we call ‘WIP’ & ‘Master Out’. When I say we, I mean I thought of it but my super talented engineer colleague actually made it.

Master out for example will take any number of sequences and perform multiple exports. You just need to set up your master sequence with clapper and set your in and out points for the picture only . Additionally we name the top track ‘supers’ and lay 2 tracks of audio (one for broadcast one for web). The script will do all the work, make clappered, superless, broadcast and web versions. We’ve also set up 1:1, 4:5 & 9:16. You just check the box and it does it’s thing. It also individually labels, date and time stamps all versions

Long explanation sorry and I know it doesn’t solve all your problems but I can’t tell you what a time saver this is.

D

1 Like

Ooh! Amazing.

For us Yankees, mind defining the word “clappered?”

Hey Drew!

I have a similar script, although I couldn’t figure out how to get the play head to move to a named track. Or did you mean that you keep the play head on the generic track and then have 1 script set to “use top track”?