RED Workflow State of the Union 2016

[Steve Sherrick]

We are 8 years in and I'd like to get a deep discussion going about the current state of workflow as it relates specifically to RED. We have seen huge advances in R3D integration thanks to the SDK getting out there in so many more places than we had at the beginning. What I'm looking for here is what choices are you making and why? We all know each project brings it's own set of challenges and workflows can change on a per project basis but what have you found to be a very effective pipeline for projects that mostly originate on RED cameras?

I want to know how you are bringing the footage from set to post.

Are you cutting natively or are you using proxies.

How are you dealing with dailies, especially as it relates to double system sound, clip naming, metadata, etc.

How are you working with VFX? Are you using in house solutions, working with vendors, etc?

What NLE has provided the best experience for you in dealing with Red footage? Be specific about what makes you choose one over the other. Not looking for NLE battle here. I want to know specifically what is working for you.

How are you finishing?

What improvements are you hoping for? Is there anything in your current workflow that is limiting you? Are there currently tools to address those limitations? If so, what is preventing you from using those tools? Learning curve? Cost?

What are your deliverables and how are you achieving them?

I know the collective membership here is made up of many different markets from hobbyists through feature films, so please do state your main focus, the types of projects you work on and in what market geographically as that will help with the context of your workflow.

Thanks guys!

 

[Steve Sherrick]

My RED Workflows

On Set

Most of the time look is created in Camera or via Redcine-X. Sometimes the look is created via Livegrade. Fool color is on my radar but have not had a chance to delve into it yet.

Data is backed up with Silverstack, nextLAB or YoYotta. Usually to large SAS or Thunderbolt 2 RAIDs and then to secondary drives.

Dailies

For larger productions I DIT, Avid is the NLE 95% of the time. Most often I am generating offline dailies in Avid DNx36 or DNx115. There is almost always a window burn and quite often this is specific to the needs/likes of that editorial team. That's why it's very important for tools that create dailies to have highly customizable window burns.

Tools I use. For large features, nextLAB and Colorfront. For commercials and independent features I'll often use Resolve. For whatever reason, I've never been able to get into Scratch, but it's on the list of things to conquer this year. I also will utilize Redcine-X, especially on projects that I edit and that I don't have access to one of the big boy boxes.

I find that synching sound and being able to utilize the metadata from the BWF is very well implemented in tools like nextLAB, Colorfront, and Scratch. Resolve is not as good in this regard and in fact, I will actually sync in Media Composer if I'm generating dailies with Resolve.

Editorial

As mentioned, on the big projects I am seeing mostly Avid as the NLE. And almost always this is at 1080P. Although Avid has gone resolution independence route, the deliverables have not changed. That's not to say that some editors are not utilizing that new feature, but so far for projects that I hand off, not seeing it utilized during editorial.

For the projects that I edit myself, I have mostly been using Avid. It's the NLE that I feel most comfortable in. I do not cut R3D natively in Avid. Always proxies. I have also experimented with FCPX and you can follow that journey over here. http://www.reduser.net/forum/showthread.php?135643-FCPX-The-Good-The-Bad-The-Ugly&highlight=
Currently though, I am following the lead of my hero Walter Murch and testing Premiere CC 2015. As per usual, I really like to have an actual project that I can test with because it's the easiest way to find the holes. So what I have done is taken a web series that was originally cut in FCP 7 and I'm going to be recutting this as a broadcast show, as well as creating a sizzle reel for pitch purposes. While I'm at it, I'm going to be mastering it in 4K, both as a potential selling point but also to finally master something at 4K. Also, since I have always been on the side of proxies, I'm going to step out of my comfort zone and edit natively in 4K. Most of the source footage is Red One MX shot at 3840x2160, some in 4096x2304. Everything will end up being UHD. So far, I have to say it's been fun. The FCP 7 projects came over very well (minus the usual incompatibilities). I was able to relink media to the R3Ds rather than the proxies that had been used in FCP 7. And so far everything has been snappy. I edit in 1/4 res which is plenty for me for editorial purposes. The challenge will be uprezzing the non RED material,, as well as any graphics that are not in UHD. I love the integration with After Effects, although my machine is not quite handling the AE compositions in real time when they are placed on the timeline. But nonetheless, the integration with AE and Photoshop is a very attractive thing. Perhaps if I fall in love with this workflow, I'll be less concerned than I was in this thread. http://www.reduser.net/forum/showthread.php?139257-Rethinking-Adobe-Pricing-Model&highlight=

Finishing

For me, on projects with smaller budgets, I send my sequence from Avid to Resolve, grade, and then either send back to Avid or finish right in Resolve. But in an ideal world, I prefer to work with a colorist who can get the most out of the footage. Therefore, if that is the case then workflow considerations are vitally important in order to ensure a smooth transition from the NLE to the grade. I wish I had the skillsets to finish at a high level in something like Smoke, but alas there are people who have spent years mastering this and when possible I want to take advantage of that.

Final Thoughts

When Red cameras first started shipping there was a lot of confusion throughout the industry about workflow. We had tools like Red Alert, and Clipfinder, etc. but there wasn't a lot of integration into the common post tools of the time. Eventually Final Cut Pro and Color were able to handle the footage through camera proxies and Color was one of the first to handle R3D natively. Now we are at a point where a lot of the mystery is gone and it's more about whether you have the hardware to support the particular resolutions you are working with. R3D support is available in just about all of the NLEs and finishing systems. The good thing is the technical side of things is starting to get out of the way and let the creative process take over.

But, it's still great to hear what people are doing, what their process is. And that's why I started this thread, to see where people have taken their workflows.

 

[Mikael Lubtchansky]

My RED Workflows

On Set

Most of the time look is created in Camera or via Redcine-X. Sometimes the look is created via Livegrade. Fool color is on my radar but have not had a chance to delve into it yet.

Steve, if you have an ethernet to lemo cable or a wifi enabled camera (REDLink, HOTlink, DSMC2, COLR...) feel free to shoot me an email. I can send you a test license for OS X version with Lift Gamma Gain / tangent panel support and access to other metadata look tools.
No iOS demo version available sorry but the OS X version has been used for on-set pre-grading workflows on House of Cards, Daredevil, Narcos etc...

 

[Eric Santiago]

Finished a fun personal project for friends wedding.

Basically my first time Directing/Writing a short film (it felt like a short).

I use all three popular NLEs and this time I went with FCPX since I had 5 days to edit an 8 day shoot (equates to about 4 days).

I spent most of it in 4K RAW but as soon as I had to start doing CG, I had no choice but to bring down the plates to HD.

Sequential 4K frames in Maya was cumbersome and roto in After Effects as well.

HD didnt speed things up but it made it manageable with file size and RAM Preview.

I like FCPX since I dont have to cut too many corners when it comes to RED files.

I worked in Premiere for the past 7 months just enough to know what to avoid and def cant deal with RAW in that one.

Avid same and even with (at work) MC 8.4.x and the DNxIO, 4K is still a battle on that side of the fence.

I can honestly say that at least all three plays well with Resolve.

I am still on 11.x.

When I find time to work 12 in too my life...if when of course

 

[David Battistella]

Steve,

Here is what I have done for my last four independent films.

Capture 6K FF anywhere from 5:1 - 8:1 compression
Back up footage via Double Data, (which I will probably now dump since testing GAMP)
LTO 6 backup of footage
One-light footage in RCX (though I might move to resolve for this)
Import into FCP X
Render proxy media
Edit
Lock
Export FCPXxml for resolve
Export FCPXxml for X2Pro
Audio mix and track layout in Logic (or prep for Protools)
FCPXxml to RESOLVE, trim media to an external drive and ship to Color Grade
Color grade (1 film in Pablo, 2 in Resolve)
Create deliverables

1X 4K/2K DCP
1X 4K 4444 Prores master
1X 2K 4444 Prores master
1x HD 4444 Prores master

Then everything gets anther LTO 6 backup and I start the next film.

Depending on a few variables and the project much of this can be altered on a project by project basis. One thing I would like to add is that I use Shotlister and Light Iron on some projects, but it all depends on the production.

 

[Manfred Lopez]

One-light footage in RCX (though I might move to resolve for this)
Import into FCP X

Your workflow is pretty much what we do. I just wanted to point out for anyone else working with r3d and Final Cut that it is very important to FIRST modify and save changes to the metadata to the red footage in RedCine BEFORE importing it into Final Cut. Otherwise Final Cut might create it's own 'phantom' metadata file which then can't be accessed by RedCine, and both programs would become disjointed and all the magic of change-anywhere-to-change-everywhere would be gone. Learned this the hard way.

Also we don't do proxies as we edit 5k with a trash can and it all seems to work. It's pretty incredible we don't have to deal with rendering or proxies until exiting resolve.

 

[David Battistella]

Your workflow is pretty much what we do. I just wanted to point out for anyone else working with r3d and Final Cut that it is very important to FIRST modify and save changes to the metadata to the red footage in RedCine BEFORE importing it into Final Cut. Otherwise Final Cut might create it's own 'phantom' metadata file which then can't be accessed by RedCine, and both programs would become disjointed and all the magic of change-anywhere-to-change-everywhere would be gone. Learned this the hard way.

Also we don't do proxies as we edit 5k with a trash can and it all seems to work. It's pretty incredible we don't have to deal with rendering or proxies until exiting resolve.

Yup. Very good points. It's exactly when I put the one light step before the import step as you can edit with a pretty good version that holds up well for screenings, etc.

Cool. I just use the proxies because it's long form and generally a bit smoother overall when there is a lot of media.
I have to say, from day 1 to now the RED R3D workflow has become one of the simplest camera to finish workflows out there and prores will make that even easier.

David

 

[Martin Stevens]

Anyone using LTO 7 tapes yet?

 

[Steve Sherrick]

Anyone using LTO 7 tapes yet?

Good question. In the feature world, I would suspect it will take a little while to transition from 6 to 7. From a small shop perspective the HP single loader will run you $5k plus software. 6TB tapes are running from $125-150. If you can get over the hurdle of the upfront cost, LTO 7 with it's large capacity and very good write speeds(300MB/s) are a great way to archive. I hate to admit it but for my own productions, I'm still backing up to multiples of hard drives.

 

[Phil Holland]

I hate to admit it but for my own productions, I'm still backing up to multiples of hard drives.

Nothing to be ashamed about. The studios I shoot for typically backup to tape, but I also maintain a copy (2 or 3 copies to be exact) of all my footage on drives and I occasionally get the call to help bringing a shot back online. Weirdest one so far was something I shot in 2004 that came up last year, that shot was migrated to newer/larger drives and it was pretty simple to restore.

As long as you have generally a good practice of refreshing drives, multiple drive redundancy, and a decide cataloging system it's still pretty cost effective and fairly reliable.


One thing that's worth noting for this thread is the advancements and additional of GPU acceleration that jumped into REDCINE-X Pro and various 3rd NLEs. This more or less has changed the name of the game when ingesting and/or working with REDCODE RAW .R3Ds. I worked in hotels for 4 months processing my footage (5K-6K) daily on a fairly super powered laptop offloading to externals, pushing out dailies, doing some first light corrections, etc. Not long ago this would have been virtually impossible for a job of that scale. That was about the equivalent of just a bit over 2.2 million feet of film.

 

[PatrickFaith]

My workflow is for performance capture for multiple distribution channels:
1) VR Story development (alpha version of Quastum) VR based at 240 fps (two skeletol systems & 3d simulator )
2) Production Shot: A&B camera red dragon 6k @ 24 fps @ 3:1 redcode; a & b xsen suites @ 240 fps; lidar per shot (covers scene and camera placements)
5.1 sound (4 ccm41, center ku4 and u67 for the .1) -> 6 channel nagra
3) nuke r3d -> prores 444 & openexr clean plate ; maya & vuexstream -> openexr ; nuke (read prores 444 & exr) -> vfx final 444 at 4k
4) davinci reads 4k prores -> writes 4k dnxhr (I'm on windows for davinci and it can't write 4k prores - is a pain btw)
5) protools 5.1 (works off of davinci proxy)
6) transmedia r3d reconform (using alpha version of quaxl, can go to 12k geometrically merging 4k prores with original r3d, geometry and graded 4k) -> dci 4k, 2k, 1080p & alembic, multiple 1080p video's with projection mapping script)

Mix of macpros with windows boxes (dual cpu 10 core windows, macpros 12 core), they read from NAS, also each computer has a 24terrabyte local raid cash using thunderbolt 2. Everything is on a 10g network.

 

[Blair S. Paulsen]

Nice setup Patrick :drool:

Cheers - #19

 

[Michael Tiemann]

I thought sharks with friggin' lasers were cool. Then I saw

[...] Everything is on a 10g network.

Nice!

 

[Steve Sherrick]

My workflow is for performance capture for multiple distribution channels:
1) VR Story development (alpha version of Quastum) VR based at 240 fps (two skeletol systems & 3d simulator )
2) Production Shot: A&B camera red dragon 6k @ 24 fps @ 3:1 redcode; a & b xsen suites @ 240 fps; lidar per shot (covers scene and camera placements)
5.1 sound (4 ccm41, center ku4 and u67 for the .1) -> 6 channel nagra
3) nuke r3d -> prores 444 & openexr clean plate ; maya & vuexstream -> openexr ; nuke (read prores 444 & exr) -> vfx final 444 at 4k
4) davinci reads 4k prores -> writes 4k dnxhr (I'm on windows for davinci and it can't write 4k prores - is a pain btw)
5) protools 5.1 (works off of davinci proxy)
6) transmedia r3d reconform (using alpha version of quaxl, can go to 12k geometrically merging 4k prores with original r3d, geometry and graded 4k) -> dci 4k, 2k, 1080p & alembic, multiple 1080p video's with projection mapping script)

Mix of macpros with windows boxes (dual cpu 10 core windows, macpros 12 core), they read from NAS, also each computer has a 24terrabyte local raid cash using thunderbolt 2. Everything is on a 10g network.

I read this and I think wow, I'm not very ambitious.

Seriously, I'd love to see your results. You're always doing something cool.

 

[PatrickFaith]

Nice setup Patrick :drool:

Cheers - #19

Hey, thanks for your sugestions in reduser, I've tried to implement a bunch of them.

I thought sharks with friggin' lasers were cool. Then I saw

Nice!

They 10g btw is the noisest thing I've done, in my blog I show me pulling up an isobox ... which btw I love and has made a .5 petabyte network sound less then a fan. I'm btw in love with my production 5.1 recording sound system, doesn't sound too techie but has been the hardest thing to do and I think is the coolest thing.

I read this and I think wow, I'm not very ambitious.
Seriously, I'd love to see your results. You're always doing something cool.

There's a couple of shots on pudls.com, I'll try and post a bit more there - I keep my film notes at that site which is more about Robert Rodriguez sugestion to keep a diary(I really like Robert's general approach to film making). Late this month I need to start gathering things together for marketing, so I'll have something in February. There's a ton of alpha stuff in my production workflow, so it's not something I recommend for people that don't control thier delivery time schedules. The transmedia conform though, I think could help your workflow Steve, in that if your fanatical about maintaining r3d naming conventions - it allows a merge of the r3d with the "final" prores release(which could have been at 1080p 7 bit) to give back the raw resolution and bit depth (this is total alpha stuff now, but I could see this being solid in a year). Long term the transmedia conform will have a node in nuke that will have two inptuts (i.e. final edited and graded 7 bit 1080p with another r3d raw input), then output 4k prores at 12 bit.

 

[David Battistella]

I am going to dive into your blog patrick. Far out! Love it.

 

[Steve Sherrick]

4K Upscaling

Anyone have a best practice, best tools for the job they have used for great results for cutting in non-RED, non 4K material? I tend to get disappointed in the Red Giant products, so not sure their Instant 4K will be the exception despite them claiming they get better results than After Effects. Any plugins for Photoshop image sequence that give amazing results? I'm just trying to avoid heavy pixelation not trying to match the fidelity/sharpness of 4K which of course will be impossible.

 

[jake blackstone]

4K Upscaling

Anyone have a best practice, best tools for the job they have used for great results for cutting in non-RED, non 4K material? I tend to get disappointed in the Red Giant products, so not sure their Instant 4K will be the exception despite them claiming they get better results than After Effects. Any plugins for Photoshop image sequence that give amazing results? I'm just trying to avoid heavy pixelation not trying to match the fidelity/sharpness of 4K which of course will be impossible.

One word- Nucoda.

 

[Nick Morrison]

Steve, our workflow is pretty basic. Our deliverables are much like yours: my clients are all broadcast, and want/demand DNX 115 Quicktimes (my clients all work off the same "spec" sheet, more or less). They also want the camera RAW footage as well, as the GFX/DESIGN departments often work from the 6K/5K masters.

WORKFLOW:

1) On set, we just shoot, and DIT to three drives for redundancy.

2) That same night (or next morning), Jake Blackstone typically logs into my computer, and we grade dailies together (remotely). We do "Best Lights"; meaning we go through every take and make sure it's optimal, matches, and has our "look". We use RESOLVE, and also sync the audio as well (we shoot with jammed/matching timecode on all cameras...with Sound guy being the master).

3) I often deliver a drive within 24hrs to client, with a RAW folder, a DNX Graded folder, and a sound folder. Recently there's been some demand for 4K Prores from GFX/DESIGN, so I may start adding those as well.

This workflow is for the PROMOS we shoot, which are our bread and butter.

Hope this helps!

best

 

[Marcus Friedlander]

Hey Nick,

Quick question. For step 2, how deep into grading do you go? Do you just do primary corrections, or do you also do secondaries as well?