DPX Export from FCP?

[Matthew Celia]

I'm wondering if this workflow has any quality loss (or doesn't make use of the full RED RAW data)

1. Transcode to ProRes and edit offline
2. Online to R3D files (what's the best way? Clipfinder? L&T? Crimson?)
3. Export to DPX to deliver to a post house for Color Correction.

We're trying to bypass the expensive "conforming" process and essentially online the work ourselves, sans color correction.

Wondering if the export from FCP to DPX directly causes any loss of information from the Red files...

I understand that any rendered down files (effects, speed, etc) will have a loss of the original red data... but how much? Would it affect the coloring?

Thoughts?

Thanks!

 

[jake blackstone]

I'm wondering if this workflow has any quality loss (or doesn't make use of the full RED RAW data)

1. Transcode to ProRes and edit offline
2. Online to R3D files (what's the best way? Clipfinder? L&T? Crimson?)
3. Export to DPX to deliver to a post house for Color Correction.

We're trying to bypass the expensive "conforming" process and essentially online the work ourselves, sans color correction.

Wondering if the export from FCP to DPX directly causes any loss of information from the Red files...

I understand that any rendered down files (effects, speed, etc) will have a loss of the original red data... but how much? Would it affect the coloring?

Thoughts?

Thanks!

What's expensive about conforming? It's done instantly from EDL in the Scratch or something that works directly with RED files. You're overcomplicating things...

 

[Emery Wells]

What's expensive about conforming? It's done instantly from EDL in the Scratch or something that works directly with RED files. You're overcomplicating things...

A conform is hardly an 'instant' process no matter what the source material is. The length of the piece, number of repos, speed changes, accuracy of the EDL etc, all play a role in how long a conform can take. It *can* be instant but it more often than not... is not.

 

[Carlos Corral]

Take a read at my blog on editing with REDCode in FCP and then going to COLOR. You may just be able to higher a color correction person if you have final cut studio 2.

Editing with REDCode:
http://www.mindwarpentertainment.com/blog/2009/04/editing-with-redcode/

 

[Jeremy Newmark]

Wondering if the export from FCP to DPX directly causes any loss of information from the Red files...

I understand that any rendered down files (effects, speed, etc) will have a loss of the original red data... but how much? Would it affect the coloring?

Thoughts?

Thanks!

There are a lot of variables at play here, and as with any workflow, it can be done a handful of different ways. Going from the RAW R3D file to DPX will certainly cause loss of information. You are dropping down to 10bits with the DPX to start. However, you can absolutely get wonderful results by finishing with the DPX files. There is also the matter of how you export the DPX files and with what settings. If you already have a post house in mind, I would start by checking with them. Every post house is going to have their way of doing things and they should be able to tell you exactly what they need. Most of the major grading systems now have some sort of RED workflow, so you may not even need to go the DPX route, but don't be afraid to if you do decide to go that way. It really all comes down to the skills and talent of whose hands you wind up putting your project in. There are many, many very talented colorists out there who will give you a much nicer finished product working off of your DPX files, then a less talented colorist working from your R3D files. Sorry for the ramble, just check with your post house.

 

[jake blackstone]

A conform is hardly an 'instant' process no matter what the source material is. The length of the piece, number of repos, speed changes, accuracy of the EDL etc, all play a role in how long a conform can take. It *can* be instant but it more often than not... is not.

I challenge anyone to see any difference in speed of a conform in a Scratch be it 5 cuts or 500. Once the shots are imported into the construct, the conform is instant. There are no speed changes with R3D, although, this is not a problem, while working with DPX or something similarly uncompressed (TIFF, Cineon etc). Once the speed is specified in the EDL, this is instant as well with DPX. So, speed changes isn't even an issue. Repos can be done after the conform inside the color grading app. Accuracy of EDL? Well, in this case, any time anything done carelessly and unprofessionally will inevitably cost money...
Regardless, between the time I get the EDL, load shots into Scratch and do a full conform, there may be barely 5 to 10 minutes. It depends on the number of the shots, that needs to be loaded into construct.

 

[Emery Wells]

I challenge anyone to see any difference in speed of a conform in a Scratch be it 5 cuts or 500. Once the shots are imported into the construct, the conform is instant. There are no speed changes with R3D, although, this is not a problem, while working with DPX or something similarly uncompressed (TIFF, Cineon etc). Once the speed is specified in the EDL, this is instant as well with DPX. So, speed changes isn't even an issue. Repos can be done after the conform inside the color grading app. Accuracy of EDL? Well, in this case, any time anything done carelessly and unprofessionally will inevitably cost money...
Regardless, between the time I get the EDL, load shots into Scratch and do a full conform, there may be barely 5 to 10 minutes. It depends on the number of the shots, that needs to be loaded into construct.

Not to hijack the purpose of this thread but while the conform action in scratch is of course instant, how accurate your result is can be another story. If every conform you do in Scratch is a one button operation than I want to be working with your clients. The real world is sloppier than you describe. EDLs often need massaging. Speed changes come into Scratch but it offers no frame blending. All of our speed changes happen outside of Scratch and then have to be brought back in. I consider repos to be part the 'conform' process.

Duplicate TC, incorrectly named reels, missing reels, wiped out metadata, and the likes are all what happens in the real world. Do we have one button conforms sometimes? Yes, but more often than not that isn't the case.

 

[Emery Wells]

Not to mention that if you are working on a movie with thousands of edits, you have to check each and every edit go through frame by frame. Export plates that may need to go to VFX. Reversals are another problem in Scratch, they don't come through. Id like to see how well you fair with a couple hundred edits that are all speed changed and reversed back and forth. You get a quicktime reference with no TC burnin. Good luck with your 5 to 10 minutes.

 

[M Most]

Duplicate TC, incorrectly named reels, missing reels, wiped out metadata, and the likes are all what happens in the real world. Do we have one button conforms sometimes? Yes, but more often than not that isn't the case.

While this is true, it's a lot less prevalent in higher end production than it is in the "indie" world. This is due in part to more experienced editors, but it's also due to the presence of very well trained assistant editors, proper front end support (often from professional facilities), and properly maintained databases in post production. It's also due largely to the fact that it's usually the indies who think they know more than they actually do, make assumptions (often incorrect ones) based on that belief, and don't ask questions when they should, either prior to, during, or after production.

 

[Michael Cioni]

Final Cut Pro & FINISHING

Final Cut Pro & FINISHING

I feel it is important that all post houses are sensitive to anyone's concern over the expense of a data conform. Back in the day, conforming 35mm cut lists used to be extremely slow and very expensive. With the advent of tape, the conforming got much faster and cheaper, but still proves to be a tedious, costly process. Finally, with data conform, the process is even faster, but still yields a noticeable cost to the customer.

Matthew-
My advice to you is to weigh the pros and cons of the data conform. There are many projects that find the cost of paying someone to do a proper conform are worth it in more ways than one. Being aware of what one is giving up by foregoing a conform can also help the decision making process.
Full debayers in a lot of the popular tools out there can help with:
-more dynamic range
-more accurate qualifiers
-less compression = less resistance
-no rendering/transcoding
-overall less time
-real time playback

But with special consideration to your concern, there is a workflow in which I recommend and successfully use here at PlasterCITY on a regular basis. If you are not going to do a conform, then one of the most efficient ways to color time your final movie is with the ProRes files themselves!
If you perform a 1/2Res High debayer (RedLog/CamRGB) in RedRushes, you can easily cut with 10bit ProRes files and simply media-manage the PicLock to UC10bit files. These can be imported as 10bit files in nearly every system on the planet and because you do a 1/2 Res debayer in Log space, you will end up with UC10bit files that will have nearly as much information as DPX files. This is to say by taking the time to render upfront in Log, you can completely avoid any reconform, plus all your transitions, speed effects, or any other work in FCP is finalized in 2K and will make the job for the colorist easier since there will be very little conforming on their end required. This will probably be a lot less work, time and nearly the same quality as your proposed workflow.

Hopefully this all make sense, but the most important thing to remember is conforming is necessary on all jobs that wish to explore the best quality. But there is always more than one way to do this, so keep digging and you'll eventually find a workflow that fits your project.
best,
michael

 

[M Most]

...the most important thing to remember is conforming is necessary on all jobs that wish to explore the best quality.

Are people here (besides you and me) aware that at this moment, there are major network television dramas being finished directly from Avid DNxHD 115 files (and yes, I did say 115, not 175)? And, if shot on Red, not necessarily using anything beyond a 2K Standard debayer? Not to mention cable shows that are done in a similar fashion using ProRes (and not HQ)? I would say that if these approaches are good enough for network dramas, they're good enough for a typical indie project intended for DVD, Internet. or even festival screenings, but people will believe what they want to believe.

Besides, I really don't want to encourage the "good enough is good enough" mentality that seems to be rampant these days...

 

[jake blackstone]

Not to mention that if you are working on a movie with thousands of edits, you have to check each and every edit go through frame by frame. Export plates that may need to go to VFX. Reversals are another problem in Scratch, they don't come through. Id like to see how well you fair with a couple hundred edits that are all speed changed and reversed back and forth. You get a quicktime reference with no TC burnin. Good luck with your 5 to 10 minutes.

Again, conform is instant, confidence check is anything but. Scratch, Lustre, Baselight etc. will conform material exactly according to EDL. If editor or his assistant didn't do their job carefully, guess who's going to pay for it? Yeah, at $500 to $1000 an hour for a grading session, it will become expensive, fast. So, let's talk about confidence check. If EDL is done correctly, problem solved. Check a couple of events at the beginning and at the end. If nothing drifted by the last event, it's probably fine then. Speed changes? All of the high end systems will read EDL speed correctly, positive and negative. Unfortunately, they do not do frame interpolation, therefore this is not the recommended way of doing the slow downs. Lustre can send the timeline to the flame for example and it will do the proper speed change. Then that could be dropped into the timeline. But to do it correctly, a new "pull list" or a "cut list" needs to be created by the editor. In it all of the speed changes are removed, that EDL is conformed and graded and the graded material is exported, but as a series of separate events, preserving names and TC, so FCP can do the reassemble and then editor can apply all of the speed changes to it's hart's desire (with RE;Vision's Twixtor, for example). If everyone does their job correctly, color grading is not that expensive

 

[jake blackstone]

Are people here (besides you and me) aware that at this moment, there are major network television dramas being finished directly from Avid DNxHD 115 files (and yes, I did say 115, not 175)? And, if shot on Red, not necessarily using anything beyond a 2K Standard debayer? Not to mention cable shows that are done in a similar fashion using ProRes (and not HQ)? I would say that if these approaches are good enough for network dramas, they're good enough for a typical indie project intended for DVD, Internet. or even festival screenings, but people will believe what they want to believe.

Besides, I really don't want to encourage the "good enough is good enough" mentality that seems to be rampant these days...

Couldn't agree more! That is why I recommend my clients just to perform their edit and give me the EDL. I will do the rest- conform, transcode, grade, render out to QT, DCI etc. I work from original R3D files or transcoded DPXs for Lustre grade. The best quality and least amount of confusion. I know Prores or DNxHD route is a common these days. But it is all highly compressed and, as we all know, compression doesn't lend itself very much to grading. Primary grading is fine, but any secondaries or god forbid, any key, not good at all...

 

[Michael Cioni]

Conforming Time

Conforming Time

I also agree with Jake and Mike-

A properly formatted EDL/XML/AAF and a good list manager/AE can insure that any conform can be fast and efficient. This is very helpful to filmmakers especially when weighing the cost v. time v. quality equation.

One of our recent films hits BluRay next Tuesday, "S. Darko - A Donnie Darko Take." This film was shot on the Red, offlined in FCP and conformed/graded in Pablo (plus a 35mm print as made). The total conforming time for this 6 reel feature was about 7 hours of labor from the online editor. That includes reference movie checking, compositing (from FCP offline) and A,B & C reel layering, etc. For feature work, 7 hours of labor for 105 minutes of picture with 2600 events is very efficient.

michael

 

[Emery Wells]

I also agree with Jake and Mike-

A properly formatted EDL/XML/AAF and a good list manager/AE can insure that any conform can be fast and efficient. This is very helpful to filmmakers especially when weighing the cost v. time v. quality equation.

One of our recent films hits BluRay next Tuesday, "S. Darko - A Donnie Darko Take." This film was shot on the Red, offlined in FCP and conformed/graded in Pablo (plus a 35mm print as made). The total conforming time for this 6 reel feature was about 7 hours of labor from the online editor. That includes reference movie checking, compositing (from FCP offline) and A,B & C reel layering, etc. For feature work, 7 hours of labor for 105 minutes of picture with 2600 events is very efficient.

michael

That is an incredibly efficient time Michael and sounds much more reasonable than "5-10 minutes." I 100% agree with you that a properly managed EDL can make a pretty swift time of things, trouble is thats just not always the case. Also, Pablo can accept multilayered timelines from FCP, Scratch cannot. Even still, just QCing a perfect EDL for a feature film is going to take hours, not minutes and that was my only point.

As Mike Most suggested, the process is often not managed properly from the start. Sloppy workflows are even more rampant in the commercial world where everyone thinks they can open their own edit facility. Education of proper file based workflows is steady but slow and we are still dealing with shops everyday who are doing their 'first' red workflow. I suspect things will get better and yes its true that a data conform is significantly faster than tape.

Im still astonished that the most prevalent exchange of edit lists is still a CMX3600 EDL. EDL's are absolutely primitive and the industry is really in need of a new modern standard.

 

[M Most]

Im still astonished that the most prevalent exchange of edit lists is still a CMX3600 EDL. EDL's are absolutely primitive and the industry is really in need of a new modern standard.

Well, English is now a very old language, but it still seems to work just fine. My rather obtuse point is that things get replaced when they need to be replaced, not simply because they're simple. Simple is often good. One of the biggest issues I've seen in the last couple of years has been the emergence of self taught editors, some of whom have real editing talent, but few of whom have the desire or the discipline to learn the craft properly. They can create cuts that work, but when they try to pass along their projects for a professional finish, the finishing artist finds a rats nest of multiple video tracks (some of which serve no purpose because they're overwritten by other tracks), hidden effects that aren't documented, sizing changes that are not documented because they're already rendered, clips that don't exist because they were created as intermediate elements, nested sequences that aren't identified because the timeline that was nested is not supplied, time codes that are incorrect because they used the Final Cut default of "drop frame," 24 frame shows that were cut at 30 frames because the editor (who didn't have an assistant, of course) either couldn't understand or couldn't be bothered with converting an NTSC transfer back to 24 frames, 2 hour shows that haven't been balanced on <20 minute reels even though they were told to do so, audio that is incomprehensible - and all of this accompanied by an attitude of "what do you mean you can't figure it out?"

The fact is that the basic EDL has been modified over the years to convey just about everything it needs to. The problem is that until you have multiple editing system vendors agreeing to document effects and speed changes in an identical way, numbers that represent such things are going to be meaningless. I think vendors like Autodesk and Quantel (not to mention Automatic Duck) have done a very good job in writing interpreters for the "native" formats and descriptors used by both Avid and Apple. The need for multiple video tracks really only exists when there are multilayer effects going on - a three way superimposition, or a multipanel effect, things of that nature. For 90% of almost any show, it's cuts and dissolves. But self taught editors rarely seem to understand - or want to understand - the value and need for simplifying and clarifying their cuts so that others can understand them. That's the real problem.

 

[Brandon Kraemer]

I'm wondering if this workflow has any quality loss (or doesn't make use of the full RED RAW data)

1. Transcode to ProRes and edit offline
2. Online to R3D files (what's the best way? Clipfinder? L&T? Crimson?)
3. Export to DPX to deliver to a post house for Color Correction.

We're trying to bypass the expensive "conforming" process and essentially online the work ourselves, sans color correction.

Wondering if the export from FCP to DPX directly causes any loss of information from the Red files...

I understand that any rendered down files (effects, speed, etc) will have a loss of the original red data... but how much? Would it affect the coloring?

Thoughts?

Thanks!

Monkey extract probably is the best toolset for this job, and you can skip the L&T step and cut with proxy files (drag and drop). Export an XML or CMX 3600 EDL, open with Monkey, monkey uses Red Alert in the background and will generate DPX sequences, and will even update the DPX headers and EDL file to create proper reel ID's for the color correction facility.

 

[jake blackstone]

That is an incredibly efficient time Michael and sounds much more reasonable than "5-10 minutes." I 100% agree with you that a properly managed EDL can make a pretty swift time of things, trouble is thats just not always the case. Also, Pablo can accept multilayered timelines from FCP, Scratch cannot. Even still, just QCing a perfect EDL for a feature film is going to take hours, not minutes and that was my only point.

As Mike Most suggested, the process is often not managed properly from the start. Sloppy workflows are even more rampant in the commercial world where everyone thinks they can open their own edit facility. Education of proper file based workflows is steady but slow and we are still dealing with shops everyday who are doing their 'first' red workflow. I suspect things will get better and yes its true that a data conform is significantly faster than tape.

Im still astonished that the most prevalent exchange of edit lists is still a CMX3600 EDL. EDL's are absolutely primitive and the industry is really in need of a new modern standard.

Efficient conform is directly proportionate to the quality of the supplied EDL. Just a couple of examples. A music video client. The usual 3 minutes video, around 250 cuts. After an hour of trying to conform on Baselight, I finally realize, that about half of the events are actually dissolves 1 frame long. That is correct 1 frame long dissolves, which by definition make them cuts. And it is a nasty bug, that renders Baselight unresponsive. Do I blame Filmlight for nor foreseeing this possibility and the cost of it? Of coarse not. BTW, they fixed it now. How about the EDL with 99 video layers? Yep, another music video. How about not being able to reassemble the project in FCP after the grade? Do you start seeing the commonality? Yes, the common denominator of all of them is the director, who thinks that he can edit. I have done the conform in minutes or in some cases hours. So, 5-10 minutes of conform for a simple commercial is very possible. Just get a good editor

 

[Emery Wells]

Efficient conform is directly proportionate to the quality of the supplied EDL. Just a couple of examples. A music video client. The usual 3 minutes video, around 250 cuts. After an hour of trying to conform on Baselight, I finally realize, that about half of the events are actually dissolves 1 frame long. That is correct 1 frame long dissolves, which by definition make them cuts. And it is a nasty bug, that renders Baselight unresponsive. Do I blame Filmlight for nor foreseeing this possibility and the cost of it? Of coarse not. BTW, they fixed it now. How about the EDL with 99 video layers? Yep, another music video. How about not being able to reassemble the project in FCP after the grade? Do you start seeing the commonality? Yes, the common denominator of all of them is the director, who thinks that he can edit. I have done the conform in minutes or in some cases hours. So, 5-10 minutes of conform for a simple commercial is very possible. Just get a good editor

That is a statement we agree on upon then. I think we are saying the same things.

 

[Paul E. McCarthy]

Efficient conform is directly proportionate to the quality of the supplied EDL. Just a couple of examples. A music video client. The usual 3 minutes video, around 250 cuts. After an hour of trying to conform on Baselight, I finally realize, that about half of the events are actually dissolves 1 frame long. That is correct 1 frame long dissolves, which by definition make them cuts. And it is a nasty bug, that renders Baselight unresponsive. Do I blame Filmlight for nor foreseeing this possibility and the cost of it? Of coarse not. BTW, they fixed it now. How about the EDL with 99 video layers? Yep, another music video. How about not being able to reassemble the project in FCP after the grade? Do you start seeing the commonality? Yes, the common denominator of all of them is the director, who thinks that he can edit. I have done the conform in minutes or in some cases hours. So, 5-10 minutes of conform for a simple commercial is very possible. Just get a good editor

What rules and/or parameters do you give to editors to cut under to insure the cleanest conform in scratch.
Do they vary between Avid and FCP?