Who's using Re-entrant workflows ?

[PatrickFaith]

I've finally fully moved over to re-entrant workflows ... basically where sound/editing/compositing/vfx/production is all happening at the same time and brought together daily through the new packages designed for that. I think these workflows are about perfect for animation and performance capture, and the directors I look at work in detail are James Cameron, Pete Docter and Peter Jackson.

I think the new thing for me is that the packages to do this are now available to indies - I'm working with integrated package suite from thefoundry/autodesk/avid . These companies I think are working real well together now, doable on cost for indies, and are not impossible to learn.

So though the big "A" level guys are doing this, I was wondering what other indie's experiences are on these re-entrant workflows?

 

[Marc Wielage]

It can work provided you have experts at every stage, dedicated rooms with machines optimized for that particular software, and massive, high-powered servers that can deal with multiple simultaneous users. For average indies, no -- it's gonna be too expensive to be practical.

I know of situations where four or five cutting rooms were going on simultaneously, allowing editors to work on a single project at one time, but even then they were working on different scenes.

In general, I think you're making this far more complicated than it needs to be. If you hire four people and have four rooms alongside each other -- picture editorial, sound editorial, VFX, color -- they can all work at the same time on different stages of the project. They don't have to use the same software, but they do have to use a compatible file format that will work for everybody. The reality is that you won't use just one VFX artist; it could be 100, 200, even a thousand VFX artists working at the same time on a big project. Sound requires sound effects editors, dialogue editors, music editors, a pre-lay editor, and multiple re-recording mixers. Even color gets bogged down by needing dailies, temp versions, conforming, final color, revisions, then final renders.

I don't see a way to do a big project with very few people in very few rooms. If it was a very tiny project, like under a couple of million dollars, then you could have (say) half a dozen people working on it for six months. It could be done and has been done. But then, the end results hinge, not on the type of equipment and software used, but by the skill and artistry of the people involved.

 

[PatrickFaith]

.... They don't have to use the same software, but they do have to use a compatible file format that will work for everybody. The reality is that you won't use just one VFX artist; it could be 100, 200, even a thousand VFX artists working at the same time on a big project.

Even on the same program using the same file formats I am having problems between systems, just because the disk arrays are setup slightly differently, then loads into the software on the various machines are not linking correctly. With nuke studio/hiero I'm finding I really need a python coder just to integrate the streams from the different raid array's and/or system locations (nuke studio has bugs in my opinion on how the comp's are relinked between compositors/editors). Probably I waist more time then anything else, relinking .nk files in from other systems(i'm using nk files kind like edl's in the multi serve nuke studio environment). Often it's these simple things that don't work I find are the most frustrating and derail me.

 

[Marc Wielage]

It's for this reason that post houses have experienced staffs with IT pros, engineers, and sometimes even software programmers to cover the issues in interfacing one kind of software with another, or in dealing with new file formats or new delivery requirements, or to solve a problem caused by camera issues or machine crashes.

No one person is an expert in everything. As the great Brian Wilson once said, "you need a mess of people to stand alone." My advice: hire people smarter than you to fix it.

 

[PatrickFaith]

It's for this reason that post houses have experienced staffs with IT pros, engineers, and sometimes even software programmers to cover the issues in interfacing one kind of software with another, or in dealing with new file formats or new delivery requirements, or to solve a problem caused by camera issues or machine crashes.

No one person is an expert in everything. As the great Brian Wilson once said, "you need a mess of people to stand alone." My advice: hire people smarter than you to fix it.

I was talking to one of the guys that originally invented the re-entrant tool & modeling concepts, and he used the word "system integration" a lot. Listening to my problems he mentioned that the future editor could be thought of as a system "media" integrator, accepting & integrated all changes (and testing them against the story), also feeding back to the group in priority order what he needed next . With this I'm thinking the concept of the dailies would be drastically different, which would require a lead from music, editorial, sound, cinematography,vfx, & story ... to get together for a half hour each day to a dailies/meeting that was basically lead by this new type of Editor/Integrator to say what got in and what was dropped from the "timeline"/film. The film being more of a timeline of activies, that slowly layered to a finished film, the actives having a visual representation which eventually emerged as a finished film and could also be conformed to server based formats (i.e. like online games).

Another thing on system integration front, it's working with the vendors. I'm really dependent on the vendors and having someone code work arounds would put me in the $100M plus group (which isn't going to happen for me this years). A good thing about the Foundry is they do listen(I'd say vendors listening skills are now my key criteria for me picking them as one of my vendors), in January almost every one of my "whines"/[crap i have to do as work arounds] will be addressed in nuke 10. https://www.thefoundry.co.uk/products/nuke/nuke-10/

 

[Marc Wielage]

I was talking to one of the guys that originally invented the re-entrant tool & modeling concepts, and he used the word "system integration" a lot. Listening to my problems he mentioned that the future editor could be thought of as a system "media" integrator, accepting & integrated all changes (and testing them against the story), also feeding back to the group in priority order what he needed next.

I don't think this is true. I think this saddles the editor with far too much responsibility. To me, this is what a Chief Engineer, an IT Dept. head, a Chief Technical Officer, or some other technical person does. The editor should be more concerned with just artistic issues: telling the story, making the footage work, having enough coverage, getting the emotional tone right, coming up with workable solutions for the director, and so on.

Don't assume that just because a single person can wear multiple hats, they should wear multiple hats. I still say you're better off having an effects person, a sound mixer, an editor, and a colorist. If you have one person doing all of these things, it'd be like running a restaurant with the same person being the Maitre D', the waiter, the cook, the dishwasher, and the janitor: sure, you could do that, but it'd be far more efficient to have five people doing five jobs. Don't assuming that doing it the cheapest possible way is the best possible way. There are ways to cut corners, but forcing one person to do the job of multiple people tends to burn people out and create the potential for more stress and problems. Even hiring part-time people will work better than saddling one person with too much work and responsibility.

In the case of post, you also need a post supervisor to coordinate dealing with the editor, the assistant editor, the producers, the sound mixers, the VFX department, and the colorist. To me, 90% of the job of the post supervisor is to save time and money, and also to help avoid technical problems and make sure the post crew has the equipment and software needed to do the job.

In respect to dailies: more and more these days, the execs are watching dailies on mobile platforms, so they need access to secure connections via Pix Systems, Cinesync, Framelight, Axlevideo, and so on. The advantage of having all dailies in the cloud is that no physical media or drives need to be shipped anywhere; the disadvantage is that if you're in the middle of nowhere, connectivity becomes an issue.

 

[Hans von Sonntag]

Patrick, check out this: http://www.autodesk.com/products/flame-family/overview and this http://www.reduser.net/forum/showthread.php?137934-Autodesk-Flame-Unleashed

I'm a longtime Flame user, in earlier days mainly on the couch as a TVC director. But since 2009 I do my own projects on Smoke and upgraded recently to Flame. I'm running it on a Z-Box on Linux but at the end of this month Flame will be available on OSX.

To make a very long story short: its the long awaited Super-NLE with deep, very, very deep VFX capabilities. But Flame can do more, much more: Editing like in PP or other NLEs, sound-desig with sufficient tools, the limiter and the EQ are very good, for more a proven AAF workflow is available. Grading can be performed deeply with anything you may think of. The conforming engine is second to none, so is archiving. With the new Reactor3 Engine Flame is super fast, when using a high-end GPU and a quick raid. It swallows many codecs, including R3Ds, allows rendering to EXR to ProRes (any flavor), Avid codecs, you name it. It is able to export sequences source shot based like Resolve or RedCine do, that means you can use Flame as an massively over-capable dailies generator.

Traditionally Flame as a huge edge on compositing/finishing. Its offers 3 VXF workflows: there is a traditional TL based VFX workflow as know from other NLEs, it offers a TL based and node based VFX workflow and last but not least a clip based and node based VFX workflow which is similar to that what Nuke offers.

Flame runs circles around anything available. Downsides: it has a pretty steep learning curve and requires quite a bit of knowledge in postproduction to get started. Setting up a Flame system with a second workstation running Flame Assist on Mac needs some deeper understanding in IT. But this will be done by your local ADSK-dealer.

Flame is the ideal companion to a Red because it offers true, uncompressed and RT working 4K workflows.

Flame has always been the base of any hero suite in commercial land, was very, very expensive and is now available on subscription for almost everyone who wants to take the challenge.

Great times!


Hans

 

[Marc Wielage]

Flame runs circles around anything available. Downsides: it has a pretty steep learning curve and requires quite a bit of knowledge in postproduction to get started. Setting up a Flame system with a second workstation running Flame Assist on Mac needs some deeper understanding in IT.

I'd say it has three other problems:

1) it's very expensive

2) it's very slow

3) Autodesk is not updating Lustre in an expedient way.

I think Flame is still a very useful program, but I personally believe it's a bad idea to consider any post software to be a Swiss Army knife capable of doing everything well. I think you're much better off getting a sound editing/mixing program like Pro Tools for sound, editing software like Avid to edit, VFX software like After Effects for VFX, and color correction software like Resolve for color. And four computers and four trained artists to do the work.

Flame is still a good program used every day on $100M+ features, and can work well if you have a lot of assistants and a decent support staff, along with very, very fast hardware. I don't think it's practical for modest projects, but I concede it's a subjective opinion.

 

[PatrickFaith]

Patrick, check out this: http://www.autodesk.com/products/flame-family/overview and this http://www.reduser.net/forum/showthread.php?137934-Autodesk-Flame-Unleashed

I'm a longtime Flame user, in earlier days mainly on the couch as a TVC director. But since 2009 I do my own projects on Smoke and upgraded recently to Flame. I'm running it on a Z-Box on Linux but at the end of this month Flame will be available on OSX.

To make a very long story short: its the long awaited Super-NLE with deep, very, very deep VFX capabilities. But Flame can do more, much more: Editing like in PP or other NLEs, sound-desig with sufficient tools, the limiter and the EQ are very good, for more a proven AAF workflow is available. Grading can be performed deeply with anything you may think of. The conforming engine is second to none, so is archiving. With the new Reactor3 Engine Flame is super fast, when using a high-end GPU and a quick raid. It swallows many codecs, including R3Ds, allows rendering to EXR to ProRes (any flavor), Avid codecs, you name it. It is able to export sequences source shot based like Resolve or RedCine do, that means you can use Flame as an massively over-capable dailies generator.

Traditionally Flame as a huge edge on compositing/finishing. Its offers 3 VXF workflows: there is a traditional TL based VFX workflow as know from other NLEs, it offers a TL based and node based VFX workflow and last but not least a clip based and node based VFX workflow which is similar to that what Nuke offers.

Flame runs circles around anything available. Downsides: it has a pretty steep learning curve and requires quite a bit of knowledge in postproduction to get started. Setting up a Flame system with a second workstation running Flame Assist on Mac needs some deeper understanding in IT. But this will be done by your local ADSK-dealer.

Flame is the ideal companion to a Red because it offers true, uncompressed and RT working 4K workflows.

Flame has always been the base of any hero suite in commercial land, was very, very expensive and is now available on subscription for almost everyone who wants to take the challenge.

Great times!


Hans

On my next film I'm seriously thinking of flipping over to linux. Right now I'm a mix of windows 7&8 and Mac's, with my compositing primarily on windows 7. I really like maya, but it's lack of node based processing can be frustrating for me (i.e. I really like houdini's node based approach, but it's often too heavy duty for me). I do think autodesk is getting there, and I look forward to my next film to really test this out.

I'd say it has three other problems:

1) it's very expensive

2) it's very slow

3) Autodesk is not updating Lustre in an expedient way.

I think Flame is still a very useful program, but I personally believe it's a bad idea to consider any post software to be a Swiss Army knife capable of doing everything well. I think you're much better off getting a sound editing/mixing program like Pro Tools for sound, editing software like Avid to edit, VFX software like After Effects for VFX, and color correction software like Resolve for color. And four computers and four trained artists to do the work.

Flame is still a good program used every day on $100M+ features, and can work well if you have a lot of assistants and a decent support staff, along with very, very fast hardware. I don't think it's practical for modest projects, but I concede it's a subjective opinion.

On my first film I'm pretty much scraping the bucket and doing a bunch of things that straight out aren't good. I'm hoping for my next film to get into a more "normal" budget range, LMAO if my first film is ever done. I just finished reading Jeff Bartsch book "editing better", which I really liked, he was talking about mastering tools. On mastering tools, at this point I can only really attempt to master three: protools&nukex&maya. On the liftgamagain I was reading your discussion on Walter Murch's recent interview ( http://liftgammagain.com/forum/index.php?threads/walter-murch-on-apple-vs-adobe-vs-avid.5582/ ), i btw was searching for my old copy of The Blink of an Eye, and just received his 2nd version yesterday ... rereading that ... any way I really like Walter's approach to using tools and that is a good discussion you started over there.

 

[Michael Tiemann]

On my next film I'm seriously thinking of flipping over to linux. Right now I'm a mix of windows 7&8 and Mac's, with my compositing primarily on windows 7. I really like maya, but it's lack of node based processing can be frustrating for me (i.e. I really like houdini's node based approach, but it's often too heavy duty for me). I do think autodesk is getting there, and I look forward to my next film to really test this out.

I'm curious what you think of Blender, if you have looked at it. Having Python built-in as a scripting language is pretty nice.

 

[PatrickFaith]

I'm curious what you think of Blender, if you have looked at it. Having Python built-in as a scripting language is pretty nice.

In 2012 my initial treatments were with blender ... https://www.youtube.com/watch?v=p3pWVPf2RAM . Thinking back on it, what I was really doing was more of a self assessment then a technology assessment: did I have what it would take to make a A level feature with theater distribution or would I continue with making art house type shorts. At that time the component workflow wasn't in blender, also I was needing to drop into the C code a lot with blender and didn't like how the parallel code worked(this is before I started exploring OpenCL). For other reasons I had a lot of walk throughs of A level houses and noticed they where almost always using nuke(or mistika), maya and protools. So focusing more on story, I decided to basically just to copy what the big guys where doing and stopped oscillating on the packages I was learning(this is when I also fully committed to shooting on RED). So using nuke/maya/protools was more of a executive "lets stop screwing around" decision after I realized I wasn't even competent enough to evaluate what the best packages were.

Both maya and nuke are doing pretty good on open standard formats though, so I'll be thinking hard on OpenCL integration to packages in 2016 which will define what my top level package will be. My 2nd film will start in 2017, lot has changed in blender, so I will move to more open standard processing for that film.

[edit, I should also add that all my focus is on performance capture ... I.e golem in hobbit, avatar and planet of the apes]